Reliability and Availability

Introduction:

The changing size of projects and designs has caused an increased adoption of reliability and
availability determinations. These activities have been in place for many years but generally fell
into a select few design areas. The increased interest in these methods has been driven by
several factors. If a design proves to be unreliable, a significant cost can be incurred to repair
the devices. As scientific detectors become larger and more complicated, repairing tens of
thousands of circuit boards would have a severe cost impact and has the potential to cause the
experiment to be canceled after a great deal of money and effort has been invested.

The following article is reprinted with permission from Prenscia HBK (Reliasoft).
HBMPrenscia.com

It does not represent an endorsement of the company or its products but rather it is an
example of a methodology for Reliability and Availability calculation methods.

MIL-217, Bellcore/Telcordia and Other Reliability Prediction Methods for Electronic Products -
ReliaSoft

The article discusses both Reliability and Availability.

MIL-217, Bellcore/Telcordia and Other

Reliability Prediction Methods for
Electronic Products
[Please note that the following article — while it has been updated from our newsletter archives
— may not reflect the latest software interface and plot graphics, but the original methodology
and analysis steps remain applicable.]

Software Used: Lambda Predict, Weibull++, ALTA

In today's competitive electronic products market, having higher reliability than competitors is
one of the key factors for success. To obtain high product reliability, consideration of reliability
issues should be integrated from the very beginning of the design phase. This leads to the
concept of reliability prediction. Historically, this term has been used to denote the process of
applying mathematical models and component data for the purpose of estimating the field
reliability of a system before failure data are available for the system. However, the objective of
reliability prediction is not limited to predicting whether reliability goals, such as MTBF, can be
reached. It can also be used for:
 • Identifying potential design weaknesses
• Evaluating the feasibility of a design
• Comparing different designs and life-cycle costs
• Providing models for system reliability/availability analysis
• Establishing goals for reliability tests
• Aiding in business decisions such as budget allocation and scheduling

Once the prototype of a product is available, lab tests can be utilized to obtain more accurate
reliability predictions. Accurate prediction of the reliability of electronic products requires
knowledge of the components, the design, the manufacturing process and the expected operating
conditions. Several different approaches have been developed to achieve the reliability prediction
of electronic systems and components. Each approach has its unique advantages and
disadvantages. Among these approaches, three main categories are often used within government
and industry: empirical (standards based), physics of failure and life testing. In this article, we
will provide an overview of all three approaches.

First, we will discuss empirical prediction methods, which are based on the experiences of
engineers and on historical data. Standards, such as MIL-HDBK-217 and Bellcore/Telcordia, are
widely used for reliability prediction of electronic products. Next, we will discuss physics of
failure methods, which are based on root-cause analysis of failure mechanisms, failure modes
and stresses. This approach is based upon an understanding of the physical properties of the
materials, operation processes and technologies used in the design. Finally, we will discuss life
testing methods, which are used to determine reliability by testing a relatively large number of
samples at their specified operation stresses or higher stresses and using statistical models to
analyze the data.

Empirical (or Standards Based) Prediction Methods

Empirical prediction methods are based on models developed from statistical curve fitting of
historical failure data, which may have been collected in the field, in-house or from
manufacturers. These methods tend to present good estimates of reliability for similar or slightly
modified parts. Some parameters in the curve function can be modified by integrating
engineering knowledge. The assumption is made that system or equipment failure causes are
inherently linked to components whose failures are independent of each other. There are many
different empirical methods that have been created for specific applications. Some have gained
popularity within industry in the past three decades. The table below lists some of the available
prediction standards and the following sections describe two of the most commonly used
methods in a bit more detail.

Prediction Method Applied Industry Last Update

MIL-HDBK-217F and Notice 1 and 2 Military 1995
Bellcore/Telcordia Telecom 2011
NSWC Mechanical 2011
FIDES Commercial/French Military 2009
MIL-HDBK-217 Predictive Method

MIL-HDBK-217 is very well known in military and commercial industries. It is probably the
most internationally recognized empirical prediction method, by far. The latest version is MIL-
HDBK-217F, which was released in 1991 and had two revisions: Notice 1 in 1992 and Notice 2
in 1995.

The MIL-HDBK-217 predictive method consists of two parts; one is known as the parts
count method and the other is called the part stress method [1]. The parts count method assumes
typical operating conditions of part complexity, ambient temperature, various electrical stresses,
operation mode and environment (called reference conditions). The failure rate for a part under
the reference conditions is calculated as:

where:

• λref is the failure rate under the reference conditions

• i is the number of parts

Since the parts may not operate under the reference conditions, the real operating conditions will
result in failure rates that are different from those given by the "parts count" method. Therefore,
the part stress method requires the specific part’s complexity, application stresses, environmental
factors, etc. (called Pi factors). For example, MIL-HDBK-217 provides many environmental
conditions (expressed as πE) ranging from "ground benign" to "cannon launch." The standard also
provides multi-level quality specifications (expressed as πQ). The failure rate for parts under
specific operating conditions can be calculated as:

where:

• πS is the stress factor

• πT is the temperature factor
• πE is the environment factor
• πQ is the quality factor
• πA is the adjustment factor

Figure 1 shows an example using the MIL-HDBK-217 method (in ReliaSoft Lambda
Predict software) to predict the failure rate of a ceramic capacitor. According to the handbook,
the failure rate of a commercial ceramic capacitor of 0.00068 μF capacitance with 80% operation
voltage, working under 30 degrees ambient temperature and "ground benign" environment is
0.0217 / 106 hours. The corresponding MTBF (mean time before failure) or MTTF (mean time to
failure) is estimated to be 4.6140 / 107 hours.

Figure 1: MIL-HDBK-217 capacitor failure rate example

Bellcore/Telcordia Predictive Method

Bellcore was a telecommunications research and development company that provided joint R&D
and standards setting for AT&T and its co-owners. Because of dissatisfaction with military
handbook methods for their commercial products, Bellcore designed its own reliability
prediction standard for commercial telecommunication products. In 1997, the company was
acquired by Science Applications International Corporation (SAIC) and the company's name was
changed to Telcordia. Telcordia continues to revise and update the standard. The latest two
updates are SR-332 Issue 2 (September 2006) and SR-332 Issue 3 (January 2011), both called
"Reliability Prediction Procedure for Electronic Equipment."

The Bellcore/Telcordia standard assumes a serial model for electronic parts and it addresses
failure rates at the infant mortality stage and at the steady-state stage with Methods I, II and III
[2-3]. Method I is similar to the MIL-HDBK-217F parts count and part stress methods. The
standard provides the generic failure rates and three part stress factors: device quality factor ( πQ),
electrical stress factor (πS) and temperature stress factor (T). Method II is based on combining
Method I predictions with data from laboratory tests performed in accordance with specific SR-
332 criteria. Method III is a statistical prediction of failure rate based on field tracking data
collected in accordance with specific SR-332 criteria. In Method III, the predicted failure rate is
a weighted average of the generic steady-state failure rate and the field failure rate.

Figure 2 shows an example in Lambda Predict using SR-332 Issue 3 to predict the failure rate of
the same capacitor in the previous MIL-HDBK-217 example (shown in Figure 1). The failure
rate is 9.655 Fits, which is 9.655 / 109 hours. In order to compare the predicted results from MIL-
HBK-217 and Bellcore SR-332, we must convert the failure rate to the same units. 9.655 Fits is
0.0009655 / 106 hours. So the result of 0.0217 / 106 hours in MIL-HDBK-217 is much higher
than the result in Bellcore/Telcordia SR-332. There are reasons for this variation. First, MIL-
HDBK-217 is a standard used in the military so it is more conservative than the commercial
standard. Second, the underlying methods are different and more factors that may affect the
failure rate are considered in MIL-HDBK-217.

Figure 2: Bellcore capacitor failure rate example

Discussion of Empirical Methods

Although empirical prediction standards have been used for many years, it is always wise to use
them with caution. The advantages and disadvantages of empirical methods have been discussed
a lot in the past three decades. A brief summary from the publications in industry, military and
academia is presented next [5-9].

Advantages of empirical methods:

1. Easy to use, and a lot of component models exist

2. Relatively good performance as indicators of inherent reliability
3. Provide an approximation of field failure rates

Disadvantages of empirical methods:

1. A large part of the data used by the traditional models is out-of-date

2. Failure of the components is not always due to component-intrinsic mechanisms but can
be caused by the system design
3. The reliability prediction models are based on industry-average values of failure rate,
which are neither vendor-specific nor device-specific
4. It is hard to collect good quality field and manufacturing data, which are needed to define
the adjustment factors, such as the Pi factors in MIL-HDBK-217

Physics of Failure Methods

In contrast to empirical reliability prediction methods, which are based on the statistical analysis
of historical failure data, a physics of failure approach is based on the understanding of the
failure mechanism and applying the physics of failure model to the data. Several popularly used
models are discussed next.

Arrhenius's Law

One of the earliest and most successful acceleration models predicts how the time-to-failure of a
system varies with temperature. This empirically based model is known as the Arrhenius
equation. Generally, chemical reactions can be accelerated by increasing the system temperature.
Since it is a chemical process, the aging of a capacitor (such as an electrolytic capacitor) is
accelerated by increasing the operating temperature. The model takes the following form.

where:
 • L(T ) is the life characteristic related to temperature
• A is the scaling factor
• Ea is the activation energy
• k is the Boltzmann constant
• T is the temperature.

Eyring and Other Models

While the Arrhenius model emphasizes the dependency of reactions on temperature, the Eyring
model is commonly used for demonstrating the dependency of reactions on stress factors other
than temperature, such as mechanical stress, humidity or voltage.

The standard equation for the Eyring model [10] is as follows:

where:

• L(T ,S) is the life characteristic related to temperature and another stress
• A, α, B and C are constants
• S is a stress factor other than temperature
• T is absolute temperature

According to different physics of failure mechanisms, one more term (i.e., stress) can be either
removed or added to the above standard Eyring model. Several models are similar to the standard
Eyring model. They are:

Two Temperature/Voltage Model:

Three Stress Model (Temperature-Voltage-Humidity):

Corrosion Model:

Electronic devices with aluminum or aluminum alloy with small percentages of copper and
silicon metallization are subject to corrosion failures and therefore can be described with the
following model [11]:

where:

• B0 is an arbitrary scale factor

• α is equal to 0.1 to 0.15 per % RH
• f(V) is an unknown function of applied voltage, with empirical value of 0.12 to 0.15

Hot Carrier Injection Model:

Hot carrier injection describes the phenomena observed in MOSFETs by which the carrier gains
sufficient energy to be injected into the gate oxide, generate interface or bulk oxide defects and
degrade MOSFETs characteristics such as threshold voltage, transconductance, etc. [11]:

For n-channel devices, the model is given by:

where:

• B is an arbitrary scale factor

• Isub is the peak substrate current during stressing
• N is equal to a value from 2 to 4, typically 3
• Ea is equal to -0.1eV to -0.2eV

For p-channel devices, the model is given by:

where:

• B is an arbitrary scale factor

• Igate is the peak gate current during stressing
• M is equal to a value from 2 to 4
• Ea is equal to -0.1eV to -0.2eV
Since electronic products usually have a long time period of useful life (i.e., the constant line of
the bathtub curve) and can often be modeled using an exponential distribution, the life
characteristics in the above physics of failure models can be replaced by MTBF (i.e., the life
characteristic in the exponential distribution). However, if you think your products do not exhibit
a constant failure rate and therefore cannot be described by an exponential distribution, the life
characteristic usually will not be the MTBF. For example, for the Weibull distribution, the life
characteristic is the scale parameter eta and for the lognormal distribution, it is the log mean.

Black Model for Electromigration

Electromigration is a failure mechanism that results from the transfer of momentum from the
electrons, which move in the applied electric field, to the ions, which make up the lattice of the
interconnect material. The most common failure mode is "conductor open." With the decreased
structure of Integrated Circuits (ICs), the increased current density makes this failure mechanism
very important in IC reliability.

At the end of the 1960s, J. R. Black developed an empirical model to estimate the MTTF of a
wire, taking electromigration into consideration, which is now generally known as the Black
model. The Black model employs external heating and increased current density and is given by:

where:

• A0 is a constant based on the cross-sectional area of the interconnect

• J is the current density
• Jthreshold is the threshold current density
• E a is the activation energy
• k is the Boltzmann constant
• T is the temperature
• N is a scaling factor

The current density (J) and temperature (T) are factors in the design process that affect
electromigration. Numerous experiments with different stress conditions have been reported in
the literature, where the values have been reported in the range between 2 and 3.3 for N, and 0.5
to 1.1eV for Ea. Usually, the lower the values, the more conservative the estimation.

Coffin-Manson Model for Fatigue

Fatigue failures can occur in electronic devices due to temperature cycling and thermal shock.
Permanent damage accumulates each time the device experiences a normal power-up and power-
down cycle. These switch cycles can induce cyclical stress that tends to weaken materials and
may cause several different types of failures, such as dielectric/thin-film cracking, lifted bonds,
solder fatigue, etc. A model known as the (modified) Coffin-Manson model has been used
successfully to model crack growth in solder due to repeated temperature cycling as the device is
switched on and off. This model takes the form [9]:

where:

• Nf is the number of cycles to failure

• Α is a coefficient
• f is the cycling frequency
• ΔT is the temperature range during a cycle
• Α is the cycling frequency exponent
• Α is the temperature exponent
• G(Tmax) is equal to:

which is an Arrhenius term evaluated at the maximum temperature in each cycle.

Three factors are usually considered for testing: maximum temperature (Tmax), temperature range
(ΔT) and cycling frequency (f). The activation energy is usually related to certain failure
mechanisms and failure modes, and can be determined by correlating thermal cycling test data
and the Coffin-Manson model.

Discussion of Physics of Failure Methods

A given electronic component will have multiple failure modes and the component's failure rate
is equal to the sum of the failure rates of all modes (i.e., humidity, voltage, temperature, thermal
cycling and so on). The system's failure rate is equal to the sum of the failure rates of the
components involved. In using the above models, the model parameters can be determined from
the design specifications or operating conditions. If the parameters cannot be determined without
conducting a test, the failure data obtained from the test can be used to get the model parameters.
Software products such as ReliaSoft ALTA can help you analyze the failure data.

We will give an example of using ALTA to analyze the Arrhenius model. For this example, the
life of an electronic component is considered to be affected by temperature. The component is
tested under temperatures of 406, 416 and 426 Kelvin. The usage temperature level is 400
Kelvin. The Arrhenius model and the Weibull distribution are used to analyze the failure data in
ALTA. Figure 4 shows the data and calculated parameters. Figure 5 shows the reliability plot
and the estimated B10 life at the usage temperature level.
Figure 4: Data and analysis results in ALTA with the Arrhenius-Weibull model
 Figure 5: Reliability vs. Time plot and calculated B10 life

From Figure 4, we can see that the estimated activation energy in the Arrhenius model is 0.92.
Note that, in ALTA, the Arrhenius model is simplified to a form of:

Using this equation, the parameters B and C calculated by ALTA can easily be transformed to
the parameters described above for the Arrhenius relationship.

Advantages of physics of failure methods:

1. Accurate prediction of wearout using known failure mechanisms

2. Modeling of potential failure mechanisms based on the physics of failure
3. During the design process, the variability of each design parameter can be determined

Disadvantages of physics of failure methods:

 1. Need detailed component manufacturing information (such as material, process and
design data)
2. Analysis is complex and could be costly to apply
3. It is difficult to assess the entire system

Life Testing Method

As mentioned above, time-to-failure data from life testing may be incorporated into some of the
empirical prediction standards (i.e., Bellcore/Telcordia Method II) and may also be necessary to
estimate the parameters for some of the physics of failure models. However, in this section of the
article, we are using the term life testing method to refer specifically to a third type of approach
for predicting the reliability of electronic products. With this method, a test is conducted on a
sufficiently large sample of units operating under normal usage conditions. Times-to-failure are
recorded and then analyzed with an appropriate statistical distribution in order to estimate
reliability metrics such as the B10 life. This type of analysis is often referred to as Life Data
Analysis or Weibull Analysis.

ReliaSoft Weibull++ software is a tool for conducting life data analysis. As an example, suppose
that an IC board is tested in the lab and the failure data are recorded. Figure 6 shows the data
entered into Weibull++ and analyzed with the 2-parameter Weibull lifetime distribution, while
Figure 7 shows the Reliability vs. Time plot and the calculated B10 life for the analysis.
Figure 6: Data and analysis results in Weibull++ with the Weibull distribution
 Figure 7: Reliability vs. Time plot and calculated B10 life for the analysis

Discussion of the Life Testing Method

The life testing method can provide more information about the product than the empirical
prediction standards. Therefore, the prediction is usually more accurate, given that enough
samples are used in the testing.

The life testing method may also be preferred over both the empirical and physics of failure
methods when it is necessary to obtain realistic predictions at the system (rather than component)
level. This is because the empirical and physics of failure methods calculate the system failure
rate based on the predictions for the components (e.g., using the sum of the component failure
rates if the system is considered to be a serial configuration). This assumes that there are no
interaction failures between the components but, in reality, due to the design or manufacturing,
components are not independent. (For example, if the fan is broken in your laptop, the CPU will
fail faster because of the high temperature.) Therefore, in order to consider the complexity of the
entire system, life tests can be conducted at the system level, treating the system as a "black
box," and the system reliability can be predicted based on the obtained failure data.
Conclusions

In this article, we discussed three approaches for electronic reliability prediction. The empirical
(or standards based) methods can be used in the design stage to quickly obtain a rough estimation
of product reliability. The physics of failure and life testing methods can be used in both design
and production stages. In physics of failure approaches, the model parameters can be determined
from design specs or from test data. On the other hand, with the life testing method, since the
failure data from your own particular products are obtained, the prediction results usually are
more accurate than those from a general standard or model.

References

[1] MIL-HDBK-217F, Reliability Prediction of Electronic Equipment, 1991. Notice 1 (1992) and
Notice 2 (1995).

[2] SR-332, Issue 1, Reliability Prediction Procedure for Electronic Equipment, Telcordia, May
2001.

[3] SR-332, Issue 2, Reliability Prediction Procedure for Electronic Equipment, Telcordia,
September 2006.

[4] ITEM Software and ReliaSoft, D490 Course Notes: Introduction to Standards Based
Reliability Prediction and Lambda Predict, 2015.

[5] B. Foucher, J. Boullie, B. Meslet and D. Das, "A Review of Reliability Prediction Methods
for Electronic Devices," Microelectron. Wearout., vol. 42, no. 8, August 2002, pp. 1155-1162.

[6] M. Pecht, D. Das and A. Ramarkrishnan, "The IEEE Standards on Reliability Program and
Reliability Prediction Methods for Electronic Equipment," Microelectron. Wearout., vol. 42,
2002, pp. 1259-1266.

[7] M. Talmor and S. Arueti, "Reliability Prediction: The Turnover Point," 1997 Proc. Ann.
Reliability and Maintainability Symp., 1997, pp. 254-262.

[8] W. Denson, "The History of Reliability Prediction," IEEE Trans. On Reliability, vol. 47, no.
3-SP, September 1998.

[9] D. Hirschmann, D. Tissen, S. Schroder and R.W. de Doncker, "Reliability Prediction for
Inverters in Hybrid Electrical Vehicles," IEEE Trans. on Power Electronics, vol. 22, no. 6,
November 2007, pp. 2511-2517.

[10] NIST Information Technology Library. [Online document] Available HTTP:

www.itl.nist.gov

[11] Semiconductor Device Reliability Failure Models. [Online document] Available HTTP:
www.sematech.org/docubase/document/3955axfr.pdf
Reliability links:
A Guide to Reliability Prediction Standards & Failure Rate | Relyence
How to Perform Reliability Predictions Easily and Efficiently (relyence.com)

▪ FMEA (Failure Mode and Effects Analysis) identifies potential failures, provides a
way to assess the criticality of those failures, and then tracks ways to eliminate
or mitigate them.
▪ FRACAS (Failure, Reporting, Analysis and Corrective Action System) and its
related CAPA (Corrective and Preventive Action) enable you to effectively track
and manage your corrective action process.
▪ FTA (Fault Tree Analysis) assesses the risk of catastrophic events.
▪ Reliability Prediction computes MTBF metrics and provides a platform for
“designing-in” reliability.
▪ RBD (Reliability Block Diagram) offers full scale system modeling and analysis of
complex designs including those that use redundancy.
▪ Maintainability Prediction provides the ability to ensure repair and maintenance
procedures are effective and efficient.
▪ Weibull analysis is a versatile tool for predictive analytics using life data.
▪ ALT (Accelerated Life Testing) allows you to take accelerated life data and
extrapolate real world system performance.

FMEA, or Failure Mode and Effects Analysis, is an organized, systematic approach for
assessing potential system failures and the resulting consequences of those failures. The
objective of a FMEA is to evaluate the risk associated with the identified failure effects and
come up with a plan to detect, prevent, or mitigate those deemed most critical.
Fault Tree Analysis (FTA) uses a top-down deductive approach to assess the likelihood of
occurrence of an undesired, often catastrophic, event. FTA provides an important measured-
based approach for risk analysis.
RAMS analysis is a well-established approach for evaluating four critical factors related to
system performance: reliability, availability, maintainability, and safety. Widely used in
engineering disciplines, RAMS analysis ensures that systems meet operational requirements
throughout the lifecycle. The objective of RAMS analysis is to assess reliability, availability,
maintainability, and safety in an organized way, identify areas of concern, and facilitate
improvements to ensure that program goals are met.
Reliability is defined as the probability, or likelihood, that an item will perform a desired function
without failure under stated conditions for a stated period of time. In general, reliability is an
indicator of the likelihood a product will operate without failure.
Availability is defined as the probability that a repairable system is in a working state when it is
required to be operational
Maintainability is defined in MIL-STD-721 as “the measure of the ability of an item to be
retained in or restored to a specified condition when maintenance is performed by personnel
having specified skill levels, using prescribed procedures and resources, at each prescribed
level of maintenance and repair.”
Safety is a term with a much clearer definition! When used in reference to RAMS analysis,
safety analysis is performed in order evaluate ways to prevent harm to people and the
environment.
 A Reliability Analysis
of the
Mu2e Calorimeter Front End Amplifier Board

By Gary Drake1, Giovanni Corradi2

Mar. 19, 2020

Version 1.0

1
Fermi National Accelerator Laboratory, Batavia, IL, USA
2
INFN, Inst. Nazionale Di Fisica Nucleare, Frascati, Italy

Abstract
This note describes the estimation process and the calculation results in performing a
reliability analysis for the Mu2e Calorimeter Front End Amplifier Board. The analysis
is based upon the procedures set forth in the military handbook, “Reliability Prediction of
Electronic Equipment,” also known as MIL-HDBK-217F. The analysis shows that the value
of the Mean Time to Failure for this board is estimated to be 1.92 x 106 hours.
Estimates of the probability of failure and a prediction of the number of board failures as
a function of time are presented.
 2

Table of Contents
1. Introduction ..................................................................................................................................... 3
1.1. Scope ........................................................................................................................................ 3
1.2. Description of the Board ...................................................................................................... 4
1.3. Limits of Scope ....................................................................................................................... 7
2. Methodology ..................................................................................................................................... 9
2.1. Overview of the Analysis Process ....................................................................................... 9
2.2. Resistors ................................................................................................................................ 11
2.3. Capacitors ............................................................................................................................. 15
2.4. Low Frequency Diodes ....................................................................................................... 19
2.4.1. Temperature Factor for General Purpose Diodes .......................................... 19
2.4.2. Temperature Factor for Voltage Regulator Diodes........................................ 20
2.4.3. Stress, Contact Construction, Quality, and Environmental Factors ............ 21
2.5. Low-Frequency, Silicon MOSFETS .................................................................................... 24
2.6. Low-Frequency Bipolar Transistors ................................................................................ 27
2.7. Linear Integrated Circuits.................................................................................................. 32
2.7.1. Temperature Factor for Linear ICs.................................................................... 32
2.7.2. Environmental Factor for Linear ICs................................................................. 34
2.7.3. Quality and Learning Factors for Linear ICs .................................................... 35
2.8. Connectors ............................................................................................................................ 37
3. Analysis Results ............................................................................................................................. 40
3.1. Analysis Results 1 – No Harsh Environment Factors .................................................... 40
3.2. Analysis Results 2 – With Harsh Environment Factors ................................................ 45
3.3. Discussion ............................................................................................................................. 49
3.4. Interpretation of this Analysis .......................................................................................... 51
4. Appendix I – Overview of Reliability Analysis Methodology ................................................ 53
5. References ...................................................................................................................................... 59
 3

1. Introduction
1.1. Scope

This note describes an analysis of the reliability of the Mu2e Calorimeter Front
End Electronics Board. The analysis is based upon the methodology described in the
military handbook, “Reliability Prediction of Electronic Equipment,” also known as MIL-
HDBK-217F [1] (hereafter referred to as “the handbook,”) which was developed by the
Dept. of Defense for analyzing the reliability of military and aerospace systems. The results
from this reliability analysis provide a prediction of the average failure rate for this board
in the Mu2e Calorimeter instrumentation system. The analysis is limited to the evaluation
of the components on the board, using the guidance set forth in the handbook, which uses
specific weighting or acceleration factors in the calculation of the failure rate for each
individual part on the board. These factors are functions of certain aspects of the
application and parts choices, including temperature, voltage, power, packaging,
complexity, fabrication technology, environment, and quality of manufacturing. Once the
failure rate for each part is calculated, they are then combined to obtain the overall failure
rate for the board. From this calculation, the “Mean Time to Failure,” or MTTF, can be
calculated. This is the standard quantity used in reliability analysis. From the MTTF,
estimates of the probability of failure and expected number of failures as functions of time
for the system can be obtained.
 4

1.2. Description of the Board

The Mu2e Calorimeter [2-3] is comprised of two disks, each constructed as an array
of cesium iodide (CsI) crystals. A rendering of the detector is shown in Fig. 1.2.1. There
are a total of 1348 crystals in the detector, split evenly between the two disks. Each crystal
is configured with four sets of three silicon photo-multipliers (SiPMs) across the face of
the crystal as shown in Fig. 1.2.2. Each set of three SiPMs is connected in series. The
signals from two such groups are summed together and instrumented with a Front End
Board connected to the back side of the SiPM holder as shown. Thus, each crystal is read
out using two Front End Boards. This yields a total of 2,696 Front End Boards in the
system.

Fig. 1.2.1. (A) Configuration of two Calorimeter Disks in the Mu2e detector
(B) Configuration of CsI Crystals Looking into the Face of a Disk

Fig. 1.2.2. Configuration of SiPMs and Associated Front End Boards on a CsI Crystal
 5

A diagram of the readout electronics is shown in Figure 1.2.3. The Front End
Boards process the charge signals from the SiPMs, and send analog voltages off-board
differentially to be digitized. The analog signals are passed through a Mezzanine Board,
and on to the waveform digitizer board, which is call the DIRAC. These boards reside a
short distance away from the Front End Boards in crates located on the outer ring of the
Calorimeter Disks, as sown in Fig. 1.2.1. Each DIRAC digitizes signals from 20 Front End
Boards, and sends the digitized data off-detector to the back-end data acquisition system
over optical data cables.

Fig. 1.2.3. Diagram of the On-Detector Readout for the Calorimeter

A block diagram of a Front End Board is shown in Fig. 1.2.4. The board has two
main functions. One is to process the charge signals from the SiPMs as described
previously. The second is to control and monitor the bias voltage that is needed by the
SiPMs. To achieve this, the board contains an analog-to-digitize converter (ADC) for
digitizing the bias voltage, and a digital-to-analog converter (DAC) for producing an
analog voltage that controls the bias voltage value. The overall control of the bias circuit
is implemented using an ARM microprocessor that resides on the Mezzanine Board. The
microprocessor distributes the bias voltage reference values to the DACs on the Front End
Boards, and then reads back the digitized values from the ADCs, adjusting the DAC values
as needed to achieve the desired bias voltage. The bias control and monitor data are also
read out through the DIRAC and then passed to the back-end Detector Control System
(DCS) over the optical data cable. Groups of 20 Front End Boards are controlled by each
Mezzanine Board. The Front End Board also contains a regulator section for regulating
the voltages needed by the board, as shown in the figure. Design notes and performance
reports for the Front End Board can be found in [4-12].
 6

Fig. 1.2.4. Block Diagram of a Front End Board

Pictures of the Front End Board are shown in Fig. 1.2.5. The top side contains
circuitry for the amplifier section, while the back side contains circuitry for the bias control
and voltage regulators. There is a total of 86 different parts, some having multiple instances
that yield 176 parts total. All of the parts have surface mount packaging (SMT), including
resistors and capacitors with several different package sizes, diodes, discrete transistors,
integrated circuits (ICs), and connectors.

Fig. 1.2.5. Picture of a Front End Board

Left is the Top (AMP) Side; Right is the Bottom (HV) Side

In the reliability analysis described in this note, the parts are categorized according
to the types defined in the handbook. The handbook then defines acceleration factors for
each part type and prescribes how to calculate them based upon operating conditions,
packaging, etc. The details of these calculations are presented in Section 2 for each part
type. Note that this analysis only pertains to the Front End Boards. The reliability of the
DIRAC and Mezzanine Boards will be covered separately.
 7

1.3. Limits of Scope

This analysis does not include consideration of the quality or reliability of the
printed circuit board itself, neither the fabrication nor the assembly. These aspects are
more difficult to assess and require intimate knowledge of the practices and materials used
by individual vendors, which may be a function of time. Fortunately, experience in
designing and supporting large High Energy Physics (HEP) detector instrumentation
systems [13-21] has shown that the dominate failure mode of electronics tends to be due
component failures, provided that care is taken to select printed circuit board fabrication
and assembly vendors who have been qualified, such as ISO900x certification.

This analysis does not include consideration of radiation damage. Indeed, for HEP
applications, this can be a significant aspect in reliability analysis, and this is true for the
Mu2e experiment. Specifications for radiation tolerance have been developed for the
different subsystems in the experiment [22], and radiation tolerance measurement
campaigns are either in progress or have been completed. However, at the time of this
report, results for the Calorimeter Front End Board were not available in a form that lend
themselves to the framework of this analysis, i.e. the multiplicative acceleration factors
that modify base failure rates for individual parts. Radiation damage aspects that affect the
reliability of these electronics will therefore not be addressed here, and instead will be
reported separately.

In lieu of formal consideration of radiation damage, the handbook does provide the
means for considering aspects that affect reliability related to the nature of the environment
that the equipment operates in. Not surprisingly, since the handbook was developed by the
Dept. of Defense, the environments defined in the handbook tend to be related to military
applications, such as ground-based military, naval, aerospace, missile launch, etc. One of
these environments, AUC, Airborne, Uninhabited Cargo, has similarities to a HEP
experiment, where human access is limited, and with somewhat harsh environmental
conditions. The analysis performed for the Calorimeter Front End Board includes
consideration of the impact on the MTTF if this environmental factor is applied. Even for
the military applications, these environmental factors are estimates or approximations,
providing a means to evaluate the trend in decreased reliability that various harsh
conditions can cause, albeit with factors that likely have large uncertainties. To the extent
that the AUC environment has similarities to an HEP experiment, this analysis serves to
illustrate how the reliability of this board can degrade by extreme environmental
conditions, albeit with the caveat concerning uncertainties. This effect will be described in
the discussion of the results in Section 3.

This analysis also does not include any part-specific reliability information from
manufacturers or vendors. In general, it has proved to be very difficult to obtain reliability
information for non-mil-spec, commercial off-the-shelf (COTS) parts. If manufacturers
have this information at all, it is often not published in data sheets. Sometimes it can be
obtained through private inquiry, but this is rare, and it is often difficult to find the right
contact person. From a manufacturer perspective, performing reliability measurements and
adhering to advertised limits on a production line certainly adds cost, in a competitive
 8

environment where cost is often weighted more highly than reliability. Indeed, COTS
parts are generally not manufactured with the same quality regimen as high-reliability
parts, so publishing reliability information can have a negative marketing effect. Lastly,
potential liability concerns are also factors that disfavor publishing reliability data. The
MIL-HDBK-217F handbook provides a means for electronics design teams to evaluate
system reliability as a function of different parts choices, quality level choices, testing
levels, etc., when information from the component vendors is not available. This has
inherent limitations, which are discussed further in Section 3.
 9

2. Methodology
2.1. Overview of the Analysis Process

The following is a short summary of the mathematical concepts and relationships

that form the basis for this analysis, primarily derived from [32], although there have been
many good textbooks published on the subject. A more detailed treatment is provided in
the Appendix.

In the general case, a system may be thought of as being composed of a number of

units. Each unit can have a number of components. The goal in a reliability analysis is to
determine the failure rate of the units in the system. This in turn involves the analysis of
the failure rate of the individual components within a unit. In the simplest type of reliability
analysis, the failure of any component in a unit represents a failure of that unit.

In general, the probability of failure of electronic components and systems during

their useful lifetime tends to have an exponential distribution [33-34]. The quantity F(t),
called the Cumulative Distribution Function (CDF), is defined as the accumulated number
of failures as a function of time, normalized as a fraction of the total number of a given units
in a system. For the case where the probability of failure is exponential in nature, F(t) has
the form:

=1− , 0≤t≤∞ (2.1.1)

where λ is a constant, called the hazard rate or the average failure rate. The units of λ are
in “number of failures per unit time.” In the electronics industry, this is often expressed
as the number of failures in 1E9 hours of operation, and is called “Failures in Time,” or
FITs.

The quantity R(t), called the Reliability function or the Survival function, is defined
as the number of units that survive at time t, again normalized as a fraction of the total
number of a given component type. R(t) is related to F(t) as:

= 1− , 0≤t≤∞ (2.1.2)

For the case where the probability of failure is exponential in nature, R(t) has the form:

= 1− = , 0≤t≤∞ (2.1.3)

For a printed circuit board containing M components, each with hazard rates λ1, λ2, …λM
respectively, the hazard rates are added together to give an overall hazard rate for the
board:
= ∑ (2.1.4)
 10

Once the hazard rate for a board is known, the probability of having a failure in
the system at a time ti, can be calculated as:

= !1 − "# ∗ % & ,0 ≤ ≤ ∞ (2.1.5)

For a system consisting of N identical units, an estimate of the accumulated number of

failures that will have occurred at time ti, is given by:

# = , ∗ !1 − "# ∗ % & ,0 ≤ ≤ ∞ (2.1.6)

For any given component on a board, there may be several factors that contribute
to the hazard rate. Examples include temperature, mechanical or electrical stress, overall
quality of the part, the environment, etc. One could model this as individual hazards or
failure mechanisms. The approach used in MIL-HDBK-217F is to define a base hazard
rate for each type of part, λb, and then define multiplicative factors that are functions of a
particular failure mechanism. These factors are called weighting factors or acceleration
factors, or sometimes “pi factors.” The resulting hazard rate for a part then has the form,

λp = λb * π1 * π2 * … * πK , for K failure mechanisms (2.1.7)

The base rate can be thought of as the failure rate under baseline conditions when all
acceleration factors equal 1.

In the subsections that follow, a description of the failure mechanisms and hazard
rate factors is presented addressing specifically the components on this board, which
again is based upon the approach used in MIL-HDBK-217F.
 11

2.2. Resistors

From MIL-HDBK-217F, Section 9.1, the hazard rate for resistors is specified as
the following:

λp = λb * πT * πP * πS * πQ * πE (2.2.1)

where: λp = the overall hazard rate

λb = the base hazard rate
πT = the temperature factor
πP = the power factor
πS = the stress factor
πQ = the quality factor
πE = the environment factor (2.2.2)

The handbook lists several different types of resistors, including: carbon

composition, metal film, thick film, wire wound, chip, etc. Each type tends to have the
acceleration factors shown above, although the values may differ from resistor type to
resistor type. For this board, the resistors used are all surface mount, type RM, Resistor,
Fixed, Film, Chip, Established Reliability. The type RM resistor is specified to have a
bas hazard rate of 3.7 failures per 1E9 hours. There is also one thermistor on the board,
which is denoted as type RTH, Thermistor, and has a base hazard rate of 1.9 failures per
1E9 hours of operation.

The aging of resistors is accelerated with increasing temperature, compared to

operation at a reference temperature. The temperature factor, πT, modifies the base rate as
shown in equation (2.2.1). For the type RM resistor, it is modeled by the following
relationship:

012 6 6
/ ∗: AB
-. = 3.567∗ 6809 ;2 <=7> ;?1@ <=7>
(2.2.3)

where Ea is the activation energy, Ta is the ambient temperature, and TREF is a reference
temperature. For the type RM resistor, the temperature parameters correspond to column
2 in the handbook. The activation energy is specified as 0.08 Joules. The reference
temperature is usually taken to be room temperature (25 C.) A plot of the temperature
factor as a function of operating temperature for a chip resistor is shown in Fig. 2.2.1. For
the type RTH thermistor, temperature is not a factor in the aging, and the temperature factor
is specified to be 1.
 12

Fig. 2.2.1. Temperature Factor vs. Temperature for Chip Resistors

Power dissipation in a resistor creates heat, and therefore contributes to lifetime

acceleration in a similar way as temperature. The acceleration factor for power, πP,
modifies the base rate as shown in equation (2.2.1), and is a function of the power
dissipation in the resistor. For the type RM resistor, the relationship is modeled by:

-C = D E F G H.IJ
(2.2.4)

A plot of the temperature factor as a function of operating temperature is shown in Fig.

2.2.2.

Fig. 2.2.2. Power Factor vs. Power Dissipation for Chip Resistors

Power dissipation in a resistor also creates stress, which contributes to accelerated

lifetime. The acceleration factor for stress, πS, also modifies that base rate as shown in
 13

equation (2.2.1). For the type RM resistor, column 1 of the stress table for resistors is used.
The stress is a function of the actual power compared to the rating. The relationship is
modeled by:
MN OPQ CRSTU
-K 0.71 ∗ . ∗K
CRSTU VP WX
, where S = (2.2.5)

A plot of the stress factor is shown in Fig. 2.2.3. For the type RTH thermistor, stress is
not a factor in the aging, and the stress factor is specified to be 1.

Fig. 2.2.3. Stress Factor vs. Stress Ratio for Chip Resistors

The quality factor, πQ, is an attribute of the quality level in manufacturing from
the vendor. The table from MIL-HDBK-217F for resistors is shown in Table 2.2.1.
Unless specifically called out in the Bill of Materials, it will be assumed that the quality
level is “Non-Established Reliability.”

Designation πQ

S 0.03
R 0.1
P 0.3
Q 1.0
Non-Established Reliability 3.0
Commercial or Unknown 10.0

Table 2.2.1. Quality Factors for Resistors as a Function of Quality Levels

 14

The environmental factor, πE, is an attribute of the environment. The table from
MIL-HDBK-217F for resistors is shown in Table 2.2.2. As discussed earlier, this analysis
will use the “AUC, Airborne, Uninhabited Cargo” for consideration of harsh conditions in
HEP experiments that may accelerate failure. This will be discussed in the Section 3.

Designation Meaning πE

GB Ground, benign 1.0

GF Ground, Fixed 4.0
GM Ground, Mobile 16.0
NS Naval, Sheltered 12.0
NU Naval, Unsheltered 42.0
AIC Airborne, Inhabited, Cargo 18.0
AIF Airborne, Inhabited, Fighter 23.0
AUC Airborne, Uninhabited, Cargo 31.0
AUF Airborne, Uninhabited, Fighter 43.0
ARW Airborne, Rotary, Winged 63.0
SF Space Flight 0.5
MF Missile, Flight 37.0
ML Missile, launch 87.1
CL Cannon, Launch 1728

Table 2.2.2. Environmental Factors for Resistors as a Function of Environment

 15

2.3. Capacitors

From MIL-HDBK-217F, Section 10.1, the hazard rate for capacitors is given as the
following:

λp = λb * πT * πC * πV * πSR * πQ * πE (2.3.1)

where: λp = the overall hazard rate.

λb = the base hazard rate
πT = the temperature factor
πC = the capacitance factor
πV = the voltage stress factor
πSR = the series resistance factor (tantalum capacitors)
πQ = the quality factor
πE = the environment factor (2.3.2)

There are several different types of capacitors listed in the handbook, most notably
differing by the type of dielectric, including: paper, metalized plastic, mica, ceramic, glass,
electrolytic, tantalum, etc. There are many different packaging options listed as well. Each
type tends to have the acceleration factors shown above, although the values may differ
from type to type. For this board, the capacitors are all surface mount with ceramic
dielectric, type CDR, Capacitor, Chip, Multiple Layer, Fixed, Ceramic Dielectric,
Established Reliability. The base rate for the CDR capacitor is specified as 2.0 failures
per 1E9 hours of operation.

The aging of capacitors is accelerated with increasing temperature, compared to

operation at a reference temperature. The temperature factor, πT, modifies the base rate as
shown in equation (2.3.1). The value of πT is modeled by the following relationship:

012 6 6
/ ∗: AB
-. 3.567∗ 6809 ;2 <=7> ;?1@ <=7>
(2.3.3)

where Ea is the activation energy, Ta is the ambient temperature, and TREF is a reference
temperature. For the type CDR capacitor, the temperature parameters correspond to
column 2 in the handbook. The activation energy is specified as 0.35 Joules. The reference
temperature is usually taken to be room temperature (25 C.) A plot of the temperature
factor as a function of operating temperature for a chip capacitor is shown in Fig. 2.3.1.
 16

Fig. 2.3.1. Temperature Factor vs. Temperature for Chip Capacitors

(Referenced to 25Deg. C Ambient)

The acceleration factor for capacitance, πC, modifies the base rate as shown in
equation 2.3.1. Generally, the larger the capacitance, the higher the probability of failure.
For the type CDR capacitor, column 1 of the capacitance factor table is used. The data is
modeled by the equation:

-Y Z F [ G[ G H.HJ
(2.3.4)

A plot of the capacitance factor as a function of capacitance value for a chip capacitor is
shown in Fig. 2.3.2.

Fig. 2.3.2. Capacitance Factor vs. Capacitance for Chip Capacitors

 17

The operating voltage applied to a capacitor also creates stress, which can lead to
accelerated failures. The acceleration factor for voltage stress, πV, also modifies that base
rate as shown in equation (2.3.1). For the type CDR capacitor, column 3 of the stress table
for capacitors is used. The stress is a function of the applied voltage compared to the
voltage rating. The relationship is modeled by:

MaaQ Tb \RQ PXT

]⁄0.6
I
-\ ` 1 , where S = \RQ PXT VP WX
(2.3.5)

A plot of the voltage factor as a function of applied voltage for a chip capacitor is shown
in Fig. 2.3.3.

Fig. 2.3.3. Voltage Stress Factor vs. Stress Ratio for Chip Capacitors

The quality factor, πQ, is an attribute of the quality level in manufacturing from
the vendor. The table from MIL-HDBK-217F for capacitors is shown in Table 2.3.1.
Unless specifically called out in the Bill of Materials, it will be assumed that the quality
level is “Non-Established Reliability.”
 18

Designation πQ

D 0.001
C 0.01
S,B 0.03
R 0.1
P 0.3
M 1.0
L 1.5
Non-Established Reliability 3.0
Commercial or Unknown 10.0

Table 2.3.1. Quality Factors for Capacitors as a Function of Quality Levels

The environmental factor, πE, is an attribute of the environment. The table from
MIL-HDBK-217F for capacitors is shown in Table 2.3.2. Consideration of the reliability
of this board under harsh conditions as characterized by the military designation AUC will
be discussed in Section 3.

Designation Meaning πQ

GB Ground, benign 1.0

GF Ground, Fixed 10.0
GM Ground, Mobile 20.0
NS Naval, Sheltered 7.0
NU Naval, Unsheltered 15.0
AIC Airborne, Inhabited, Cargo 12.0
AIF Airborne, Inhabited, Fighter 15.0
AUC Airborne, Uninhabited, Cargo 25.0
AUF Airborne, Uninhabited, Fighter 30.0
ARW Airborne, Rotary, Winged 40.0
SF Space Flight 0.5
MF Missile, Flight 20.0
ML Missile, launch 50.0
CL Cannon, Launch 570

Table 2.3.2. Environmental Factors for Capacitors as a Function of Environment

 19

2.4. Low Frequency Diodes

From MIL-HDBK-217F, Section 6.1, the hazard rate for low frequency diodes is
specified as the following:

λp = λb * πT * πS * πC * πQ * πE (2.4.1)

where: λp = the overall hazard rate.

λb = the base hazard rate
πT = the temperature factor
πS = the stress factor
πC = the contact construction factor
πQ = the quality factor
πE = the environment factor (2.4.2)

There are several different types of low frequency diodes defined in the handbook.
The types that are included in this category are: general purpose analog, switching, fast
recovery, power rectifier, transient suppressor, current regulator, voltage regulator, and
voltage reference. Each type tends to have the failure factors shown above, although the
values may differ from type to type. In this design, there are two types of diodes used:
general purpose (GP), and voltage regulator (VR). For the general purpose diode, the
base failure rate is specified as 3.8 failures per 1E9 hours. For the voltage regulator diode,
the base failure rate is 2.0 failures per 1E9 hours.

2.4.1. Temperature Factor for General Purpose Diodes

The aging of diodes is accelerated with increasing temperature, compared

to operation at a reference temperature. The temperature factor, πT, modifies the
base rate as shown in equation (2.4.1). It is a function of the junction temperature,
Tj, which is given by:

cd cP + e * P) (2.4.3)

Where: Tj = Junction temperature

Ta = Ambient temperature
Rth = Thermal resistance between junction & ambient
P = Power dissipation (2.4.4)

Once the junction temperature is known, the temperature factor, πT, can be
found. For the general purpose diode, the first set of the temperature tables is used.
The value of πT is modeled by the equation:

6 6
f IHJ ∗/ Bh
-. =
;g<=7> ;2 <=7>
(2.4.5)
 20

where the temperatures are in Celsius. A plot of the temperature factor as a

function of the junction temperature for a general purpose diode is shown in Fig.
2.4.1.

Fig. 2.4.1. Temperature Factor vs. Temperature for General Purpose Diodes
(Referenced to 25 Deg. C Ambient)

2.4.2. Temperature Factor for Voltage Regulator Diodes

For the voltage regulator diode, the second set of the temperature tables is
used. The value of πT is modeled by the equation:

6 6
f Jij ∗/ Bh
-.
;g<=7> ;2 <=7>
(2.4.6)

where the temperatures are in Celsius. A plot of the temperature factor as a

function of the junction temperature for a voltage regulator diode is given in Fig.
2.4.2.
 21

Fig. 2.4.2. Temperature Factor vs. Temperature for Voltage Regulator Diodes
(Referenced to 25 Deg. C Ambient)

2.4.3. Stress, Contact Construction, Quality, and Environmental Factors

Voltage stress can occur in diodes under reverse bias conditions. The
acceleration factor for stress, πS, also modifies that base rate as shown in equation
(2.2.1). For the voltage regulator diode, the stress factor is specified to be 1.0,
voltage stress is not a factor for these devices. For all low frequency diodes, the
stress is modeled by:

-K 0.054 , for 0 ≤ VS ≤ 0.3 (2.4.7)

-K 0.54 ∗ i.mI ∗ K
, for 0.3 < VS ≤ 1, (2.4..8)
MN OPQ VTnTUoT \RQ PXT
Pp VTnTUoT \RQ PXT VP WX
where VS = .

A plot of the stress factor is shown in Fig. 2.4.3.

 22

Fig. 2.4.3. Stress Factor vs. Stress Ratio for Low Frequency Diodes under Reverse Bias

The quality factor, πQ, is an attribute of the quality level in manufacturing

from the vendor. The table from MIL-HDBK-217F for low frequency diodes is
shown in Table 2.4.1. Unless specifically called out in the Bill of Materials, it
will be assumed that the quality level is “JAN.”

Designation πQ

JANTXV 0.7
JANTX 1.0
JAN 2.4
Lower 5.5
Plastic 8.0

Table 2.4.1. Quality Factors for Low Frequency Diodes as a Function of Quality Levels

The environmental factor, πE, is an attribute of the environment. The table

from MIL-HDBK-217F for diodes is shown in Table 2.4.2. Consideration of the
reliability of this board under harsh conditions as characterized by the military
designation AUC will be discussed in Section 3.
 23

Designation Meaning πQ

GB Ground, benign 1.0

GF Ground, Fixed 6.0
GM Ground, Mobile 9.0
NS Naval, Sheltered 9.0
NU Naval, Unsheltered 19.0
AIC Airborne, Inhabited, Cargo 13.0
AIF Airborne, Inhabited, Fighter 29.0
AUC Airborne, Uninhabited, Cargo 20.0
AUF Airborne, Uninhabited, Fighter 43.0
ARW Airborne, Rotary, Winged 24.0
SF Space Flight 0.5
MF Missile, Flight 14.0
ML Missile, launch 32.0
CL Cannon, Launch 320

Table 2.4.2. Environmental Factors for Low Frequency Diodes as a Function of Environment
 24

2.5. Low-Frequency, Silicon MOSFETS

There are several different types of Field Effect Transistors (FETs), including N-
channel, P-channel, enhancement mode, depletion mode, power, JFETs, GaAsSFETs, etc.
They differ in construction depending on the application, such as small-signal, switching,
or power. FETS can come as discrete, or as part of a larger integrated circuit. Also, FETs
are fabricated in many different technologies and feature sizes. The MIL-HDBK-217F has
chosen to divide FETs into three main categories: low-frequency silicon MOSFETS (less
than or equal to 400 MHz); high frequency silicon MOSFETs, and GaAsFETs. JFETs are
included in the low-frequency silicon MOSFETs. Integrated circuits are considered
separately. In this design, only low-frequency silicon MOSFETs are used, designated as
MOS, LF.

From MIL-HDBK-217F, Section 6.4, the hazard rate for low-frequency, silicon
MOSFETs is given as the following:

λp = λb * πT * πA * πQ * πE (2.5.1)

where: λp = the overall hazard rate.

λb = the base hazard rate
πT = the temperature factor
πA = the application factor
πQ = the quality factor
πE = the environment factor (2.5.2)

The base hazard rate, λb, is the hazard rate of a part under normal operation. For
the low-frequency MOSFETs, the base rate is specified as 12.0 failures per 1E9 hours of
operation.

The aging of MOSFETs is accelerated with increasing temperature, compared to

operation at a reference temperature. The temperature factor, πT, modifies the base rate as
shown in equation (2.5.1). It is a function of the junction temperature, Tj, which is given
by:

cd cP + e * P) (2.5.3)

Where: Tj = Junction temperature

Ta = Ambient temperature
Rth = Thermal resistance between junction & ambient
P = Power dissipation (2.5.4)

Once the junction temperature is known, the temperature factor, πT, can be found.
For the MOSFET, the value of πT is modeled by the equation:
 25

6 6
f Jij ∗/ Bh
-.
;g<=7> ;2 <=7>
(2.5.5)

where the temperatures are in Celsius. A plot of the temperature factor as a function of
the junction temperature for the low-frequency MOSFET is shown in Fig. 2.5.1.

Temperature Factor vs. Junction Temperature - MOSFETS

10.0
8.0
6.0
4.0
MOSFET Data
2.0
0.0
0 50 100 150 200
Junction Temperature - Degrees C

Fig. 2.5.1. Temperature Factor vs. Temperature for the Low-frequency MOSFET
(Referenced to 25 Deg. C Ambient)

The application factor, πA, accounts for stress on the device that is
application-dependent. The handbook divides into three categories: linear,
switching, and power. The application factor table from MIL-HDBK-217F for
low-frequency silicon MOSFETs is shown in Table 2.5.1. For this design, the
application is linear, so the value for πA will be taken to be 1.5.

Application πA

Linear 1.5
Small Signal Switching 0.7
Power, 2W≤ P < 5W 2.0
Power, 5W≤ P < 50W 4.0
Power, 50W≤ P < 250W 8.0
Power, P ≥ 250W 10

Table 2.5.1. Quality Factors for MOSFETS as a Function of Application

 26

The quality factor, πQ, is an attribute of the quality level in manufacturing

from the vendor. The table from MIL-HDBK-217F for MOSFETs is shown in
Table 2.5.2. Unless specifically called out in the Bill of Materials, it will be
assumed that the quality level is JAN.

Designation πQ

JANTXV 0.5
JANTX 1.0
JAN 2.0
Lower 5.0

Table 2.5.2. Quality Factors for MOSFETS as a Function of Quality Levels

The environmental factor, πE, is an attribute of the environment. The table

from MIL-HDBK-217F for MOSFETs is shown in Table 2.5.3. Consideration of
the reliability of this board under harsh conditions as characterized by the military
designation AUC will be discussed in the analysis section.

Designation Meaning πQ

GB Ground, benign 1.0

GF Ground, Fixed 6.0
GM Ground, Mobile 9.0
NS Naval, Sheltered 9.0
NU Naval, Unsheltered 19.0
AIC Airborne, Inhabited, Cargo 13.0
AIF Airborne, Inhabited, Fighter 29.0
AUC Airborne, Uninhabited, Cargo 20.0
AUF Airborne, Uninhabited, Fighter 43.0
ARW Airborne, Rotary, Winged 24.0
SF Space Flight 0.5
MF Missile, Flight 14.0
ML Missile, launch 32.0
CL Cannon, Launch 320

Table 2.5.3. Environmental Factors for MOSFETS as a Function of Environment

 27

2.6. Low-Frequency Bipolar Transistors

There are three main types of junction transistors: bipolar junction transistors
(BJT), unijunction, and heterojunction. For bipolar junction transistors, there are two main
topologies: NPN and PNP. The types differ in fabrication and construction details
depending the application, including small signal, high frequency, or power. Different
fabrication technologies can be used, including silicon, germanium, and gallium arsenide,
and can have different feature sizes. The MIL-HDBK-217F has chosen to divide junction
transistors into three main categories: low-frequency silicon bipolar (less than or equal to
200 MHz); low-noise, high-frequency silicon bipolar; high-frequency power bipolar, and
unijunction. In this design, only low-frequency silicon bipolar transistors are used,
designated as BJT, LF.

From MIL-HDBK-217F, Section 6.3, the hazard rate for low-frequency (< 200
MHz) silicon bipolar transistors is specified as the following:

λp = λb * πT * πA * πR * πS * πQ * πE (2.6.1)

where: λp = the overall hazard rate.

λb = the base hazard rate
πT = the temperature factor
πA = the application factor
πR = the power rating factor
πS = the voltage stress factor
πQ = the quality factor
πE = the environment factor (2.6.2)

The base hazard rate, λb, is the hazard rate of a part under normal operation. For
the low-frequency bipolar transistors, the base rate is specified as 0.74 failures per 1E9
hours of operation. It is the same for NPN and PNP devices.

The aging of BJTs is accelerated with increasing temperature, compared to

operation at a reference temperature. The temperature factor, πT, modifies the base rate as
shown in equation (2.6.1). It is a function of the junction temperature, Tj, which is given
by:

cd cP + e * P) (2.6.3)

Where: Tj = Junction temperature

Ta = Ambient temperature
Rth = Thermal resistance between junction & ambient
P = Power dissipation (2.6.4)
 28

Once the junction temperature is known, the temperature factor, πT, can be found.
For the BJT, the value of πT is modeled by the equation:

6 6
f i m ∗/ Bh
-.
;g<=7> ;2 <=7>
(2.6.5)

where the temperatures are in Celsius. A plot of the temperature factor as a function of
the junction temperature for the low-frequency bipolar transistor is shown in Fig. 2.6.1.

Fig. 2.6.1. Temperature Factor vs. Junction Temperature for the Low-frequency MOSFET
(Referenced to 25 Deg. C Ambient)

The application factor, πA, accounts for stress on the device that is
application-dependent. The handbook divides into two categories: linear and
switching. The application factor table from MIL-HDBK-217F for low-
frequency bipolar transistors is shown in Table 2.6.1. For this design, the
application is linear, so the value for πA will be taken to be 1.5.

Application πA

Linear 1.5
Switching 0.7

Table 2.6.1. Quality Factors for LF BJTs as a Function of Application

 29

Power dissipation in a transistor creates heat, and therefore contributes to age

acceleration in a similar way as temperature. The acceleration factor for power rating, πR,
modifies the base rate as shown in equation (2.6.1). The handbook defines this factor as a
function of the power rating only, with increasing values as the power rating increases. For
the LF BJT, the relationship is modeled by:

-V 0.43 for PR ≤ 0.1Watt (2.6.6)

-V = D Gr H.Is
for PR > 0.1Watt (2.6.7)

A plot of the power rating factor as a function of operating temperature is shown in Fig.
2.6.2.

Power Rating Factor vs. Power Rating - LF BJTs

14.0000

12.0000

10.0000

8.0000

6.0000

4.0000

2.0000

0.0000
-3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00
Log(Power Rating)

Fig. 2.6.2. Power Factor vs. Power Rating for LF BJTs

The applied voltage to a transistor also creates stress, which can lead to early
failure. The acceleration factor for voltage stress, πS, also modifies that base rate as shown
in equation (2.6.1). The stress is a function of the applied voltage compared to the voltage
rating. For the low-frequency BJT, the relationship is modeled by:

\
-K 0.045 ∗ I. ∗ \t
, where VS = \ u1 (2.6.8)
u1v

A plot of the stress factor as a function of the ratio VS for a LF BJT is shown in Fig. 2.6.3.
 30

Fig. 2.6.3. Voltage Stress Factor vs. Stress Ratio for LF BJTs

The quality factor πQ is an attribute of the quality level in manufacturing

from the vendor. The table from MIL-HDBK-217F for BJTs is shown in Table
2.6.2. Unless specifically called out in the Bill of Materials, it will be assumed
that the quality level is “JAN.”

Designation πQ

JANTXV 0.7
JANTX 1.0
JAN 2.4
Lower 5.5
Plastic 8.0

Table 2.6.2. Quality Factors for LF BJTs as a Function of Quality Levels

 31

The environmental factor, πE, is an attribute of the environment. The table

from MIL-HDBK-217F for BJTs is shown in Table 2.6.3. Consideration of the
reliability of this board under harsh conditions as characterized by the military
designation AUC will be discussed in the analysis section.

Designation Meaning πQ

GB Ground, benign 1.0

GF Ground, Fixed 6.0
GM Ground, Mobile 9.0
NS Naval, Sheltered 9.0
NU Naval, Unsheltered 19.0
AIC Airborne, Inhabited, Cargo 13.0
AIF Airborne, Inhabited, Fighter 29.0
AUC Airborne, Uninhabited, Cargo 20.0
AUF Airborne, Uninhabited, Fighter 43.0
ARW Airborne, Rotary, Winged 24.0
SF Space Flight 0.5
MF Missile, Flight 14.0
ML Missile, launch 32.0
CL Cannon, Launch 320

Table 2.6.3. Environmental Factors for LF BJTs as a Function of Environment

 32

2.7. Linear Integrated Circuits

Section 5 of MIL-HDBK-217F covers the reliability of all types of integrated

circuits. The hazard rate of linear integrated circuits, denoted as IC, LIN is defined in
Section 5.1 as:

λp = (C1 * πT + C2 * πE) * πQ * πL (2.7.1)

where: λp = the overall hazard rate.

C1 = the die complexity failure rate
πT = the temperature factor
C2 = the package failure rate
πE = the environmental factor
πQ = the quality factor
πL = the learning factor (2.7.2)

Factors C2, πT, πE, πQ, and πL are defined in sections 5.8 – 5.11. Note that there is no
base hazard rate defined for this category.

2.7.1. Temperature Factor for Linear ICs

The temperature factor includes the die complexity factor, C1, which
accounts for increasing failure rate with increasing complexity. Generally,
complexity is defined as being a function of the number of transistors in a device.
This is shown in Table 2.7.1. The handbook specifies the same table for both
bipolar and CMOS ICs, and makes no distinction on feature size. In general, unless
the IC design is custom, it is very difficult to ascertain how many devices are in an
IC design. For the purposes of this analysis, the guidance shown in the righthand
column of Table 2.7.1, developed by the author based upon will be used.

No. Transistors C1 Example Devices (Annotated)

1 to 100 0.010 Voltage References

101 to 300 0.020 Op Amps
301 to 1000 0.040 DACs
1001 to 10,000 0.060 ADCs

Table 2.7.1. Die Complexity Factors as a Function of the Number of Transistors

 33

The temperature factor, πT, modifies the base rate as shown in equation
(2.7.1). It is a function of the junction temperature, Tj, which is given by:

cd cP ` e * P) (2.7.3)

Where: Tj = Junction temperature

Ta = Ambient temperature
Rth = Thermal resistance between junction & ambient
P = Power dissipation (2.7.4)

Once the junction temperature is known, the temperature factor can be found.
It is modeled by the equation:

012 6 6
f ∗/ Bh
-. 0.1 ∗
3.567∗ 6809 ;g<=7> ;2 <=7>
(2.7.5)

where the temperatures are in Celsius. Ea is the activation energy, which for
linear ICs is taken to be 0.65. A plot of the temperature factor as a function of the
junction temperature for a linear IC is shown in Fig. 2.7.1.

Fig. 2.7.1. Temperature Factor vs. Junction Temperature for Linear ICs
(Referenced to 25 Deg. C Ambient)
 34

2.7.2. Environmental Factor for Linear ICs

The environmental factor includes the package failure rate factor, C2, which
is a function of the number of pins in a package, and also the package construction.
This is described in section 5.9 of the handbook. Package types considered include
hermetic DIPs and SMTs, cans, and DIPs with glass seals, and non-hermetic DIPs
and SMTs. Generally, the factors grow larger with package type, respectively. For
this analysis, all of the linear ICs are non-hermetic SMT. The relationship of the
factor C2 as a function of the number of pins in the package for the non-hermetic
SMT package is modeled by the equation:
.Hw
Zi 3.6 ∗ 10 m
∗ !,a & 2.7.6

where Np is the number of pins. A plot of this relationship is shown in Fig. 2.7.2.

Fig. 2.7.2. Package Complexity Factor C2 vs. Package Pin Count

 35

The environmental factor, πE, for linear ICs is defined in Section 5.10 of the
handbook. The table is shown in Table 2.7.2. Consideration of the reliability of
this board under harsh conditions as characterized by the military designation AUC
will be discussed in the analysis section.

Designation Meaning πQ

GB Ground, benign 0.5

GF Ground, Fixed 2.0
GM Ground, Mobile 4.0
NS Naval, Sheltered 4.0
NU Naval, Unsheltered 6.0
AIC Airborne, Inhabited, Cargo 4.0
AIF Airborne, Inhabited, Fighter 5.0
AUC Airborne, Uninhabited, Cargo 5.0
AUF Airborne, Uninhabited, Fighter 8.0
ARW Airborne, Rotary, Winged 8.0
SF Space Flight 0.5
MF Missile, Flight 5.0
ML Missile, launch 12.0
CL Cannon, Launch 220.0

Table 2.7.2. Environmental Factors for Linear ICs as a Function of Environment

2.7.3. Quality and Learning Factors for Linear ICs

The handbook specifies a detailed description for determining the quality

factor, πQ, as a function of different levels of testing in Section 5.10. For this
project, none of the ICs have undergone pre-testing or burn-in. Thus, the default
value for commercial or unknown screening levels will be used:

-y 10 2.7.7

The handbook defines a learning factor, πL, which takes into account the
number of years that a part has been in production. The general idea is that for
complex integrated circuits, there may be bugs in the design that are evident early
in the production, but become identified and are addressed as time goes on so that
later production cycles have a lower probability of having problems. After a couple
of years, the handbook projects that all potential bugs have been addressed. The
model as specified in Section 5.10 is given as:
 36

-z 0.01 ∗ j.Ij H.Ij∗{

, for 0 ≤ Y < 2 (2.7.8)
-z = 1 for Y ≥ 2 (2.7.9)

A plot of the learning factor as a function of production years is shown in Fig.

2.7.3.

Learning Factor vs. Years in Production - Lin

ICs
2.5000

2.0000

1.5000

1.0000

0.5000

0.0000
0 0.5 1 1.5 2 2.5 3 3.5
Years in production

Fig. 2.7.3. Learning Factor vs. Number of Years in production for Linear ICs
 37

2.8. Connectors

From MIL-HDBK-217F, Section 15.1, the hazard rate for a mated pair of
connectors is specified as the following:

λp = λb * πT * πK * πQ * πE (2.8.1)

where: λp = the overall hazard rate

λb = the base hazard rate
πT = the temperature factor
πK = the mating/un-mating factor
πQ = the quality factor
πE = the environment factor (2.8.2)

There are many different types of connectors identified in the handbook, including:
circular/cylindrical, card edge (PCB), hexagonal, rack and panel, rectangular, RF coaxial,
telephone, power, and triaxial. Each type tends to have the acceleration factors shown
above, although the values may differ from type to type. For this board, there two types of
connectors used: a Rectangular Connector, RC, and mating pins for the the SiPM
connection. There is no type defined in the handbook for the mating pins, so it will be
assumed that they are similar in reliability to Power Connectors, PC, since they are robust
and large gauge.

The base hazard rate, λb, is the hazard rate of a part under nominal operation. For
the type RC connector, the base rate is specified as 46 failures per 1E9 hours of operation.
For the type PC connector, the base rate is specified as 7 failures per 1E9 hours of
operation.

The aging of connectors is accelerated with increasing temperature, compared to

operation at a reference temperature. In this context, aging is synonymous with failure
rate. The temperature factor, πT, modifies the base rate as shown in equation (2.8.1). The
value of πT is modeled by the following relationship:

08.6| 6 6
/ ∗: AB
-. 3.567∗ 6809 ;8<=7> ;?1@ <=7>
(2.8.3)

where T0 is the contact temperature, and TREF is a reference temperature. The contact
temperature has a provision to include self-heating due to current flow, having a general
form:

cH = cP + ∆ . = cP + [ • ∗ € .wj
] (2.2.4)

where ∆T is the insert temperature, Ta is the ambient temperature, I is the current in

Amperes flowing through the connector, and K is a factor that is dependent upon the gauge
of the contacts. The reference temperature is usually taken to be room temperature (25 C.)
The value of K varies from 0.1 for 12 gauge contacts, to 3.256 for 32 gauge contacts. This
 38

relationship is the same for all connector types. Since the current flowing through the
connectors is significantly less than 1 amp, the difference between the contact temperature
and the ambient temperature is of order ~a few degrees, and will be ignored since it is
small. A plot of the temperature factor as a function of operating temperature for
connectors is shown in Fig. 2.8.1.

Fig. 2.8.1. Temperature Factor vs. Temperature for Connectors

Plot is for low currents, and neglects any self-heating.

Mating and un-mating a connector pair creates stress on the connector contacts, as
well as in the connections of the pins and sockets of the connector to the wires or cables.
The acceleration factor for mating/un-mating, πK, modifies the base rate as shown in
equation (2.8.1), and is a function of the frequency of the interconnects, as shown in Table
2.8.1. A cycle includes both connect and interconnect. The values are the same for all
connector types. In normal operation, the plugging and unplugging of this board will be
done rarely, so the low-frequency value will be assumed.

Mating/Un-mating Cycles πK
(per 1000 hours)

0 to 0.05 1.0
0.05 to 0.5 1.5
0.5 to 5 2.0
5 to 50 3.0
> 50 4.0

Table 2.8.1. Mating/Un-Mating Factors for Connectors as a Function of Interconnect Frequency.

 39

The quality factor, πQ, is an attribute of the quality level in manufacturing from
the vendor. The table from MIL-HDBK-217F for connectors is shown in Table 2.8.2.
Unless specifically called out in the Bill of Materials, it will be assumed that the quality
level is “Lower.”

Designation πQ

Mil-Spec 1
Lower 2

Table 2.8.2. Quality Factors for Connectors as a Function of Quality Levels

The environmental factor, πE, is an attribute of the environment. The table from
MIL-HDBK-217F for connectors is shown in Table 2.8.3. As discussed earlier, this
analysis will use the “AUC, Airborne, Uninhabited Cargo” for consideration of harsh
conditions in HEP experiments that may accelerate failure. This will be discussed in the
Section 3.

Designation Meaning πE

GB Ground, benign 1.0

GF Ground, Fixed 1.0
GM Ground, Mobile 8.0
NS Naval, Sheltered 5.0
NU Naval, Unsheltered 13.0
AIC Airborne, Inhabited, Cargo 3.0
AIF Airborne, Inhabited, Fighter 5.0
AUC Airborne, Uninhabited, Cargo 8.0
AUF Airborne, Uninhabited, Fighter 12.0
ARW Airborne, Rotary, Winged 19.0
SF Space Flight 0.5
MF Missile, Flight 10.0
ML Missile, launch 27.1
CL Cannon, Launch 490

Table 2.8.3. Environmental Factors for Connectors as a Function of Environment

 40

3. Analysis Results

3.1. Analysis Results 1 – No Harsh Environment Factors

The parts used on the board were identified and categorized according to the
methodology in MIL-HDBK-217F, as described in Section 2. The base hazard rates for
each part were defined. The various acceleration factors for each part were calculated,
based upon the expected operating conditions. For this stage of analysis, the environmental
factors πE for all components were set to 1.

The result of this analysis is shown in Table 3.1.1. The hazard rates for all
components were summed to give the overall hazard rate for the board, as described in
equation (2.1.4). This is shown at the top of Table 3.1.1.

The overall hazard rate for the board, λBD, is found to be:

519 F 1‚9 ƒ (3.1.1)

This gives the Mean Time to Failure (MTTF) as:

…J
„cc = = 1.92‚6 Hours (3.1.2)
"#
 41

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: OFF λNET: 519.5382
Version: V2 Ambient T: 12 MTTF: 1.9248E+06
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

1 1 C1 100n 2.0000 1.3160 2.6320 2.6320

2 1 C2 100n 2.0000 1.3262 2.6523 2.6523
3 1 C3 100n 2.0000 1.3262 2.6523 2.6523
4 1 C6 100n 2.0000 1.3160 2.6320 2.6320
5 1 C7 100n 2.0000 1.3096 2.6192 2.6192
6 1 C8 100n 2.0000 1.3108 2.6217 2.6217
7 1 C14 100n 2.0000 1.3122 2.6244 2.6244
8 1 C17 100n 2.0000 1.3262 2.6523 2.6523
9 1 C19 100n 2.0000 1.3262 2.6523 2.6523
10 1 C52 100n 2.0000 1.3262 2.6523 2.6523
11 1 C53 100n 2.0000 1.3095 2.6190 2.6190
12 1 C56 100n 2.0000 1.3119 2.6238 2.6238
13 1 C63 100n 2.0000 1.3096 2.6192 2.6192
14 1 C4 10n 2.0000 1.0779 2.1559 2.1559
15 1 C12 10n 2.0000 1.0664 2.1327 2.1327
16 1 C62 10n 2.0000 1.0661 2.1322 2.1322
17 1 C5 1u 2.0000 1.6111 3.2221 3.2221
18 1 C9 1u 2.0000 1.6111 3.2221 3.2221
19 1 C61 1u 2.0000 1.6112 3.2224 3.2224
20 1 C10 47p 2.0000 0.6571 1.3141 1.3141
21 1 C13 4p7 2.0000 0.5342 1.0685 1.0685
22 1 C24 100n 2.0000 4.4136 8.8271 8.8271
23 1 C25 100n 2.0000 1.3344 2.6687 2.6687
24 1 C29 100n 2.0000 1.3344 2.6687 2.6687
25 1 C36 100n 2.0000 1.3103 2.6206 2.6206
26 1 C39 100n 2.0000 1.3096 2.6193 2.6193
27 1 C26 22p 2.0000 1.9619 3.9238 3.9238
28 1 C27 100p 2.0000 2.9160 5.8320 5.8320
29 1 C28 10n 2.0000 4.9017 9.8034 9.8034
30 1 C30 1n2 2.0000 0.8657 1.7314 1.7314
31 1 C31 1n 2.0000 0.8657 1.7314 1.7314
32 1 C32 10n 2.0000 1.0693 2.1387 2.1387
33 1 C38 10n 2.0000 1.0645 2.1289 2.1289
34 1 C44 10n 2.0000 1.0846 2.1692 2.1692
35 1 C68 10n 2.0000 1.0693 2.1387 2.1387
36 1 C72 10n 2.0000 1.0650 2.1301 2.1301
37 1 C74 10n 2.0000 1.0680 2.1360 2.1360
38 1 C78 10n 2.0000 1.0647 2.1295 2.1295
39 1 C80 10n 2.0000 1.0693 2.1387 2.1387
40 1 C81 10n 2.0000 1.0644 2.1288 2.1288
41 1 C46 12p 2.0000 0.5813 1.1627 1.1627
42 1 C47 12p 2.0000 0.5813 1.1627 1.1627
43 1 C70 12p 2.0000 0.5813 1.1627 1.1627
44 1 C76 12p 2.0000 0.5813 1.1627 1.1627
45 1 C77 12p 2.0000 0.5813 1.1627 1.1627
46 1 C51 22p 2.0000 0.6137 1.2273 1.2273
47 1 C54 47p 2.0000 0.6573 1.3146 1.3146
48 1 C55 10n 2.0000 1.2564 2.5128 2.5128
49 1 C64 10u 2.0000 2.0866 4.1731 4.1731
50 1 C65 10u 2.0000 2.0866 4.1731 4.1731
51 1 C66 10u 2.0000 2.0866 4.1731 4.1731
52 1 C67 10u 2.0000 2.0866 4.1731 4.1731
53 1 C69 1u 2.0000 1.6185 3.2370 3.2370
54 1 C71 1u 2.0000 1.6120 3.2240 3.2240
55 1 C73 1u 2.0000 1.6165 3.2330 3.2330
56 1 C75 1u 2.0000 1.6165 3.2330 3.2330
57 1 C79 1u 2.0000 1.6165 3.2330 3.2330
58 1 D1 CMAD6001 3.8000 0.1296 0.4925 0.4925
59 1 D3 CMAD6001 3.8000 0.1296 0.4925 0.4925
60 1 D4 CMAD6001 3.8000 0.1296 0.4925 0.4925
61 1 D5 CMAD6001 3.8000 0.1296 0.4925 0.4925
62 1 D6 CMAD6001 3.8000 0.1296 0.4925 0.4925

Table 3.1.1a. Calorimeter Front End Board parts List with Calculated FITS Values, Nominal Environment
(Partial, 1 of 3)
 42

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: OFF λNET: 519.5382
Version: V2 Ambient T: 12 MTTF: 1.9248E+06
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

64 4 J1 SIPM_MU2e 7.0000 1.5596 10.9175 43.6701

65 1 J2 HSE 46.0000 1.5596 71.7438 71.7438
66 1 J3 J_GAIN 46.0000 1.5596 71.7438 71.7438
67 1 M1 AO3162 12.0000 3.8624 46.3490 46.3490
68 1 M2 BSS126 12.0000 3.7735 45.2820 45.2820
69 1 M3 AO3162 12.0000 3.8624 46.3490 46.3490
70 1 Q1 MMBT5771 0.7400 0.1090 0.0807 0.0807
71 1 Q2 MMBT5771 0.7400 0.1218 0.0902 0.0902
72 1 Q6 MMBT5771 0.7400 0.1218 0.0902 0.0902
73 1 Q3 MMBT3904 0.7400 0.5183 0.3835 0.3835
74 1 Q4 MMBT3904 0.7400 0.1226 0.0907 0.0907
75 1 R1 100R 3.7000 0.0510 0.1886 0.1886
76 1 R59 100R 3.7000 0.0510 0.1886 0.1886
77 1 R2 1k 3.7000 0.0207 0.0767 0.0767
78 1 R10 1k 3.7000 0.0207 0.0767 0.0767
79 1 R11 1k 3.7000 0.0207 0.0767 0.0767
80 1 R13 1k 3.7000 0.0207 0.0767 0.0767
81 1 R16 1k 3.7000 0.0207 0.0767 0.0767
82 1 R19 1k 3.7000 0.0207 0.0767 0.0767
83 1 R21 1k 3.7000 0.0207 0.0767 0.0767
84 1 R61 1k 3.7000 0.0034 0.0127 0.0127
85 1 R3 460R 3.7000 0.3182 1.1774 1.1774
86 1 R4 4k7 3.7000 0.1768 0.6540 0.6540
87 1 R5 22k 3.7000 0.1216 0.4499 0.4499
88 1 R24 22k 3.7000 0.1004 0.3715 0.3715
89 1 R81 22k 3.7000 0.1004 0.3715 0.3715
90 1 R6 50R 3.7000 0.0018 0.0068 0.0068
91 1 R14 50R 3.7000 0.0669 0.2476 0.2476
92 1 R83 50R 3.7000 0.0669 0.2476 0.2476
93 1 R84 50R 3.7000 0.0669 0.2476 0.2476
94 1 R7 2k 3.7000 0.2544 0.9414 0.9414
95 1 R82 2k 3.7000 0.2544 0.9414 0.9414
96 1 R8 220R 3.7000 0.0315 0.1164 0.1164
97 1 R9 220R 3.7000 0.0315 0.1164 0.1164
98 1 R15 220R 3.7000 0.1797 0.6648 0.6648
99 1 R12 68R 3.7000 0.0593 0.2194 0.2194
100 1 R17 10R 3.7000 0.1271 0.4704 0.4704
101 1 R18 1k6 3.7000 0.0101 0.0372 0.0372
102 1 R20 56k 3.7000 0.0720 0.2663 0.2663
103 1 R23 330R 3.7000 0.0018 0.0068 0.0068
104 1 R60 330R 3.7000 0.0320 0.1183 0.1183
105 1 R26 56R 3.7000 0.3444 1.2744 1.2744
106 1 R34 330R 3.7000 0.1514 0.5600 0.5600
107 1 R35 4K7 3.7000 0.0113 0.0420 0.0420
108 1 R36 1K 3.7000 0.0018 0.0068 0.0068
109 1 R37 3Meg 3.7000 0.3557 1.3159 1.3159
110 1 R38 15K 3.7000 0.1034 0.3827 0.3827
111 1 R39 38K3 3.7000 0.1073 0.3971 0.3971
112 1 R40 1K 3.7000 0.0207 0.0767 0.0767
113 1 R106 1K 3.7000 0.0207 0.0767 0.0767
114 1 R41 100R 3.7000 0.0510 0.1886 0.1886
115 1 R42 220R 3.7000 0.7567 2.7999 2.7999
116 1 R43 22K 3.7000 0.0062 0.0230 0.0230
117 1 R44 10K 3.7000 0.0084 0.0313 0.0313
118 1 R45 270K 3.7000 0.0023 0.0086 0.0086
119 1 R46 1K2 3.7000 0.0193 0.0715 0.0715
120 1 R47 10K 3.7000 0.0084 0.0313 0.0313
121 1 R51 2R2 3.7000 0.2441 0.9032 0.9032
122 1 R55 2R2 3.7000 0.2370 0.8771 0.8771

Table 3.1.1b. Calorimeter Front End Board parts List with Calculated FITS Values, Nominal Environment
(Partial, 2 of 3)
 43

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: OFF λNET: 519.5382
Version: V2 Ambient T: 12 MTTF: 1.9248E+06
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

123 1 R99 2R2 3.7000 0.2370 0.8771 0.8771

124 1 R54 5K23 3.7000 0.1367 0.5059 0.5059
125 1 R56 100R 3.7000 0.0509 0.1885 0.1885
126 1 R105 100R 3.7000 0.0509 0.1885 0.1885
127 1 R57 10R 3.7000 0.1263 0.4673 0.4673
128 1 R107 10R 3.7000 0.1263 0.4673 0.4673
129 1 R58 120R 3.7000 0.1460 0.5402 0.5402
130 1 R62 2k2 3.7000 0.0128 0.0474 0.0474
131 1 R63 2k2 3.7000 0.0128 0.0474 0.0474
132 1 R85 22R 3.7000 0.0075 0.0276 0.0276
133 1 R86 22R 3.7000 0.0075 0.0276 0.0276
134 1 R87 PT1000 3.7000 0.1181 0.4370 0.4370
135 1 R88 806R 3.7000 0.0226 0.0835 0.0835
136 1 R90 806R 3.7000 0.0226 0.0835 0.0835
137 1 R89 2K55 3.7000 0.0144 0.0533 0.0533
138 1 R91 3k9 3.7000 0.0122 0.0451 0.0451
139 1 R92 33R 3.7000 0.0788 0.2917 0.2917
140 1 R94 33R 3.7000 0.0788 0.2917 0.2917
141 1 R96 33R 3.7000 0.0788 0.2917 0.2917
142 1 R100 33R 3.7000 0.0788 0.2917 0.2917
143 1 R101 33R 3.7000 0.0788 0.2917 0.2917
144 1 R93 11K 3.7000 0.0018 0.0068 0.0068
145 1 R95 10K 3.7000 0.1051 0.3890 0.3890
146 1 R97 10K 3.7000 0.1051 0.3890 0.3890
147 1 R104 10K 3.7000 0.0880 0.3256 0.3256
148 1 R98 1K 3.7000 0.0731 0.2704 0.2704
149 1 R102 4K22 3.7000 0.2118 0.7835 0.7835
150 1 R103 4K22 3.7000 0.1246 0.4611 0.4611
151 1 U1 AD8014ARTZ 1.0000 0.0005 0.0005 0.0005
152 1 U3 AD8014ARTZ 1.0000 0.0006 0.0006 0.0006
153 1 U2 AD8099ARD 1.0000 0.0004 0.0004 0.0004
154 1 U7 OPA180 1.0000 0.0005 0.0005 0.0005
155 1 U8 LMP7707 1.0000 0.0005 0.0005 0.0005
156 1 U9 LMP7707 1.0000 0.0005 0.0005 0.0005
157 1 U10 LMP7707 1.0000 0.0004 0.0004 0.0004
158 1 U15 AD8038AKSZ 1.0000 0.0006 0.0006 0.0006
159 1 U16 TL1963_ADJ 1.0000 2.4337 2.4337 2.4337
160 1 U17 TL1963_ADJ 1.0000 0.0134 0.0134 0.0134
161 1 U18 DAC121S101 1.0000 0.0010 0.0010 0.0010
162 1 U19 ADC128S102 1.0000 0.0048 0.0048 0.0048
163 1 JUMP1 0.0000 0.0000 0.0000 0.0000

Total failure Rate (per 1E9 hours): 519.5382

Calculated MTTF: 1.9248E+06

# Failures in 1st Year: 9.80

Probability of Failure in 1st Year: 0.36%

# Failures in 5 Years: 48.64

Probability of Failure in 5 Years: 1.80%

Table 3.1.1c. Calorimeter Front End Board parts List with Calculated FITS Values, Nominal Environment
(Partial, 3 of 3)
 44

The probability of failure at time ti is given by the CDF:

= = 1− "# ∗ … J∗ %
(3.1.3)

For a system consisting of N boards, with no repairs being conducted, the expected
cumulative number of failures at time ti is given by:

# =,∗ = , ∗ [1 − "# ∗ … J∗ %
] (3.1.4)

Assuming that the Mu2e Calorimeter readout system has 2,696 Front End Boards in the
full readout system, and assuming 80% up-time each year (with the power to the electronics
turned off during the down-time), with no repairs performed, the associated probabilities
and expected number of failures in the system are summarized in Table 3.1.2.

Year Probability of Failure Predicted Number of Failures

1 0.36% 9.80
2 0.73% 19.56
3 1.09% 29.29
4 1.45% 38.98
5 1.80% 48.64

Table 3.1.2. Predicted Probability of Failure and Numbers of Failures as a Function of Time, Nominal Environment
 45

3.2. Analysis Results 2 – With Harsh Environment Factors

As discussed in in Section 1.3, the methodology of MIL-HDBK-217F provides for

each electronic component to include an environmental factor. For the analysis described
in Section 3.1, the environmental factors were set to 1. In the following, the same analysis
was performed, but with the environmental factors for the different components set to the
values corresponding with environment AUC, Airborne, Uninhabited Cargo, as described
in Section 2. The result of this analysis is shown in Table 3.2.1. As before, the resulting
hazard rates were summed to give the overall hazard rate for the board, as given by equation
(2.1.4). This is shown at the top of Table 3.2.1.

The overall hazard rate λBD_E with environmental factors included is given by:

_… 9,182 F 1‚9 ƒ (3.2.1)

This gives the Mean Time to Failure for the AUC environment, denoted as MTTFE to be:
…J ˆROUo
„cc … = = 1.09‚5 Hours (3.2.2)
"#_1

The MTTF of the board is reduced by a factor of 17.6 in this environment.

As in Section 3.1, the probability of failure and the expected number of failures at time ti
can be calculated.

Again assuming that the Mu2e Calorimeter readout system has 2,696 Front End
Boards in the full readout system, and assuming 80% up-time per year (with the power to
the electronics turned off during the down-time), with no repairs performed, the associated
probabilities and expected number of failures are summarized in Table 3.2.2.

Year Probability of Failure Predicted Number of Failures

1 7.50% 202.25
2 14.44% 389.34
3 20.86% 562.38
4 26.80% 722.45
5 32.29% 870.50

Table 3.2.2. Predicted Probability of Failure and Numbers of Failures as a Function of Time, Harsh Environment
 46

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: ON λNET: 9182.7787
Version: V2 Ambient T: 12 MTTF: 1.0890E+05
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

1 1 C1 100n 2.0000 32.8998 65.7996 65.7996

2 1 C2 100n 2.0000 33.1539 66.3078 66.3078
3 1 C3 100n 2.0000 33.1539 66.3078 66.3078
4 1 C6 100n 2.0000 32.8998 65.7996 65.7996
5 1 C7 100n 2.0000 32.7401 65.4802 65.4802
6 1 C8 100n 2.0000 32.7707 65.5415 65.5415
7 1 C14 100n 2.0000 32.8045 65.6091 65.6091
8 1 C17 100n 2.0000 33.1539 66.3078 66.3078
9 1 C19 100n 2.0000 33.1539 66.3078 66.3078
10 1 C52 100n 2.0000 33.1539 66.3078 66.3078
11 1 C53 100n 2.0000 32.7380 65.4761 65.4761
12 1 C56 100n 2.0000 32.7979 65.5959 65.5959
13 1 C63 100n 2.0000 32.7401 65.4802 65.4802
14 1 C4 10n 2.0000 26.9485 53.8970 53.8970
15 1 C12 10n 2.0000 26.6592 53.3183 53.3183
16 1 C62 10n 2.0000 26.6527 53.3054 53.3054
17 1 C5 1u 2.0000 40.2765 80.5531 80.5531
18 1 C9 1u 2.0000 40.2765 80.5531 80.5531
19 1 C61 1u 2.0000 40.2794 80.5589 80.5589
20 1 C10 47p 2.0000 16.4264 32.8528 32.8528
21 1 C13 4p7 2.0000 13.3558 26.7116 26.7116
22 1 C24 100n 2.0000 110.3392 220.6784 220.6784
23 1 C25 100n 2.0000 33.3588 66.7176 66.7176
24 1 C29 100n 2.0000 33.3588 66.7176 66.7176
25 1 C36 100n 2.0000 32.7569 65.5139 65.5139
26 1 C39 100n 2.0000 32.7407 65.4813 65.4813
27 1 C26 22p 2.0000 49.0470 98.0941 98.0941
28 1 C27 100p 2.0000 72.9004 145.8007 145.8007
29 1 C28 10n 2.0000 122.5421 245.0842 245.0842
30 1 C30 1n2 2.0000 21.6423 43.2846 43.2846
31 1 C31 1n 2.0000 21.6423 43.2846 43.2846
32 1 C32 10n 2.0000 26.7336 53.4673 53.4673
33 1 C38 10n 2.0000 26.6114 53.2229 53.2229
34 1 C44 10n 2.0000 27.1151 54.2301 54.2301
35 1 C68 10n 2.0000 26.7336 53.4673 53.4673
36 1 C72 10n 2.0000 26.6258 53.2517 53.2517
37 1 C74 10n 2.0000 26.7003 53.4005 53.4005
38 1 C78 10n 2.0000 26.6183 53.2367 53.2367
39 1 C80 10n 2.0000 26.7336 53.4673 53.4673
40 1 C81 10n 2.0000 26.6106 53.2211 53.2211
41 1 C46 12p 2.0000 14.5332 29.0664 29.0664
42 1 C47 12p 2.0000 14.5332 29.0664 29.0664
43 1 C70 12p 2.0000 14.5332 29.0664 29.0664
44 1 C76 12p 2.0000 14.5332 29.0664 29.0664
45 1 C77 12p 2.0000 14.5332 29.0664 29.0664
46 1 C51 22p 2.0000 15.3417 30.6834 30.6834
47 1 C54 47p 2.0000 16.4328 32.8656 32.8656
48 1 C55 10n 2.0000 31.4100 62.8199 62.8199
49 1 C64 10u 2.0000 52.1640 104.3280 104.3280
50 1 C65 10u 2.0000 52.1640 104.3280 104.3280
51 1 C66 10u 2.0000 52.1640 104.3280 104.3280
52 1 C67 10u 2.0000 52.1640 104.3280 104.3280
53 1 C69 1u 2.0000 40.4630 80.9260 80.9260
54 1 C71 1u 2.0000 40.2998 80.5997 80.5997
55 1 C73 1u 2.0000 40.4125 80.8249 80.8249
56 1 C75 1u 2.0000 40.4125 80.8249 80.8249
57 1 C79 1u 2.0000 40.4125 80.8249 80.8249
58 1 D1 CMAD6001 3.8000 2.5920 9.8496 9.8496
59 1 D3 CMAD6001 3.8000 2.5920 9.8496 9.8496
60 1 D4 CMAD6001 3.8000 2.5920 9.8496 9.8496
61 1 D5 CMAD6001 3.8000 2.5920 9.8496 9.8496
62 1 D6 CMAD6001 3.8000 2.5920 9.8496 9.8496
63 1 D2 BZX84C12 2.0000 48.0000 96.0000 96.0000

Table 3.2.1a. Calorimeter Front End Board parts List with Calculated FITS Values, Harsh Environment
(Partial, 1 of 3)
 47

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: ON λNET: 9182.7787
Version: V2 Ambient T: 12 MTTF: 1.0890E+05
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

64 4 J1 SIPM_MU2e 7.0000 12.4772 87.3402 349.3609

65 1 J2 HSE 46.0000 12.4772 573.9501 573.9501
66 1 J3 J_GAIN 46.0000 12.4772 573.9501 573.9501
67 1 M1 AO3162 12.0000 77.2483 926.9790 926.9790
68 1 M2 BSS126 12.0000 75.4700 905.6398 905.6398
69 1 M3 AO3162 12.0000 77.2483 926.9790 926.9790
70 1 Q1 MMBT5771 0.7400 2.1803 1.6134 1.6134
71 1 Q2 MMBT5771 0.7400 2.4365 1.8030 1.8030
72 1 Q6 MMBT5771 0.7400 2.4365 1.8030 1.8030
73 1 Q3 MMBT3904 0.7400 10.3651 7.6701 7.6701
74 1 Q4 MMBT3904 0.7400 2.4521 1.8145 1.8145
75 1 R1 100R 3.7000 1.5805 5.8477 5.8477
76 1 R59 100R 3.7000 1.5805 5.8477 5.8477
77 1 R2 1k 3.7000 0.6428 2.3785 2.3785
78 1 R10 1k 3.7000 0.6428 2.3785 2.3785
79 1 R11 1k 3.7000 0.6428 2.3785 2.3785
80 1 R13 1k 3.7000 0.6428 2.3785 2.3785
81 1 R16 1k 3.7000 0.6428 2.3785 2.3785
82 1 R19 1k 3.7000 0.6428 2.3785 2.3785
83 1 R21 1k 3.7000 0.6428 2.3785 2.3785
84 1 R61 1k 3.7000 0.1067 0.3947 0.3947
85 1 R3 460R 3.7000 9.8645 36.4986 36.4986
86 1 R4 4k7 3.7000 5.4798 20.2752 20.2752
87 1 R5 22k 3.7000 3.7691 13.9455 13.9455
88 1 R24 22k 3.7000 3.1123 11.5154 11.5154
89 1 R81 22k 3.7000 3.1123 11.5154 11.5154
90 1 R6 50R 3.7000 0.0573 0.2119 0.2119
91 1 R14 50R 3.7000 2.0746 7.6762 7.6762
92 1 R83 50R 3.7000 2.0746 7.6762 7.6762
93 1 R84 50R 3.7000 2.0746 7.6762 7.6762
94 1 R7 2k 3.7000 7.8873 29.1829 29.1829
95 1 R82 2k 3.7000 7.8873 29.1829 29.1829
96 1 R8 220R 3.7000 0.9752 3.6084 3.6084
97 1 R9 220R 3.7000 0.9752 3.6084 3.6084
98 1 R15 220R 3.7000 5.5698 20.6081 20.6081
99 1 R12 68R 3.7000 1.8385 6.8024 6.8024
100 1 R17 10R 3.7000 3.9410 14.5817 14.5817
101 1 R18 1k6 3.7000 0.3116 1.1530 1.1530
102 1 R20 56k 3.7000 2.2310 8.2546 8.2546
103 1 R23 330R 3.7000 0.0573 0.2119 0.2119
104 1 R60 330R 3.7000 0.9909 3.6664 3.6664
105 1 R26 56R 3.7000 10.6776 39.5071 39.5071
106 1 R34 330R 3.7000 4.6921 17.3609 17.3609
107 1 R35 4K7 3.7000 0.3515 1.3005 1.3005
108 1 R36 1K 3.7000 0.0573 0.2119 0.2119
109 1 R37 3Meg 3.7000 11.0254 40.7941 40.7941
110 1 R38 15K 3.7000 3.2067 11.8647 11.8647
111 1 R39 38K3 3.7000 3.3273 12.3111 12.3111
112 1 R40 1K 3.7000 0.6428 2.3785 2.3785
113 1 R106 1K 3.7000 0.6428 2.3785 2.3785
114 1 R41 100R 3.7000 1.5805 5.8477 5.8477
115 1 R42 220R 3.7000 23.4583 86.7958 86.7958
116 1 R43 22K 3.7000 0.1925 0.7123 0.7123
117 1 R44 10K 3.7000 0.2618 0.9688 0.9688
118 1 R45 270K 3.7000 0.0724 0.2679 0.2679
119 1 R46 1K2 3.7000 0.5987 2.2152 2.2152
120 1 R47 10K 3.7000 0.2618 0.9688 0.9688
121 1 R51 2R2 3.7000 7.5674 27.9995 27.9995
122 1 R55 2R2 3.7000 7.3484 27.1892 27.1892

Table 3.2.1b. Calorimeter Front End Board parts List with Calculated FITS Values, Harsh Environment
(Partial, 2 of 3)
 48

MIL-HDBK-217F Board Reliability

Electronic Board Reliability Analysis Calculation Results
G. Drake, Mar. 12, 2020, V2.00
Project: Mu2e Calorimeter Front End Board Harsh: ON λNET: 9182.7787
Version: V2 Ambient T: 12 MTTF: 1.0890E+05
Designer: G. Corradi
Part Information from
Date: March 12, 2020 Bill of Material
# Boards: 2696

Nominal Reliability Weighted Total per

Item Qnty/Bd Reference Part
FITS Weighting FITS (Calc) Board

123 1 R99 2R2 3.7000 7.3484 27.1892 27.1892

124 1 R54 5K23 3.7000 4.2390 15.6843 15.6843
125 1 R56 100R 3.7000 1.5794 5.8439 5.8439
126 1 R105 100R 3.7000 1.5794 5.8439 5.8439
127 1 R57 10R 3.7000 3.9156 14.4878 14.4878
128 1 R107 10R 3.7000 3.9156 14.4878 14.4878
129 1 R58 120R 3.7000 4.5262 16.7471 16.7471
130 1 R62 2k2 3.7000 0.3971 1.4693 1.4693
131 1 R63 2k2 3.7000 0.3971 1.4693 1.4693
132 1 R85 22R 3.7000 0.2313 0.8556 0.8556
133 1 R86 22R 3.7000 0.2313 0.8556 0.8556
134 1 R87 PT1000 3.7000 3.6617 13.5482 13.5482
135 1 R88 806R 3.7000 0.6993 2.5873 2.5873
136 1 R90 806R 3.7000 0.6993 2.5873 2.5873
137 1 R89 2K55 3.7000 0.4462 1.6508 1.6508
138 1 R91 3k9 3.7000 0.3780 1.3987 1.3987
139 1 R92 33R 3.7000 2.4440 9.0428 9.0428
140 1 R94 33R 3.7000 2.4440 9.0428 9.0428
141 1 R96 33R 3.7000 2.4440 9.0428 9.0428
142 1 R100 33R 3.7000 2.4440 9.0428 9.0428
143 1 R101 33R 3.7000 2.4440 9.0428 9.0428
144 1 R93 11K 3.7000 0.0573 0.2119 0.2119
145 1 R95 10K 3.7000 3.2595 12.0603 12.0603
146 1 R97 10K 3.7000 3.2595 12.0603 12.0603
147 1 R104 10K 3.7000 2.7281 10.0940 10.0940
148 1 R98 1K 3.7000 2.2653 8.3814 8.3814
149 1 R102 4K22 3.7000 6.5648 24.2899 24.2899
150 1 R103 4K22 3.7000 3.8635 14.2950 14.2950
151 1 U1 AD8014ARTZ 1.0000 0.0024 0.0024 0.0024
152 1 U3 AD8014ARTZ 1.0000 0.0032 0.0032 0.0032
153 1 U2 AD8099ARD 1.0000 0.0022 0.0022 0.0022
154 1 U7 OPA180 1.0000 0.0024 0.0024 0.0024
155 1 U8 LMP7707 1.0000 0.0024 0.0024 0.0024
156 1 U9 LMP7707 1.0000 0.0024 0.0024 0.0024
157 1 U10 LMP7707 1.0000 0.0021 0.0021 0.0021
158 1 U15 AD8038AKSZ 1.0000 0.0028 0.0028 0.0028
159 1 U16 TL1963_ADJ 1.0000 12.1685 12.1685 12.1685
160 1 U17 TL1963_ADJ 1.0000 0.0670 0.0670 0.0670
161 1 U18 DAC121S101 1.0000 0.0050 0.0050 0.0050
162 1 U19 ADC128S102 1.0000 0.0239 0.0239 0.0239
163 1 JUMP1 0.0000 0.0000 0.0000 0.0000

Total failure Rate (per 1E9 hours): 9182.7787

Calculated MTTF: 1.0890E+05

# Failures in 1st Year: 168.03

Probability of Failure in 1st Year: 6.23%

# Failures in 5 Years: 741.75

Probability of Failure in 5 Years: 27.51%

Table 3.2.1c. Calorimeter Front End Board parts List with Calculated FITS Values, Harsh Environment
(Partial, 3 of 3)
 49

3.3. Discussion

With the environment factors set to 1, the MTTF was found to be 1.92E6 hours.
The resulting probability of failure was found to be 0.36% in the first year. The reliability
requirements for the Calorimeter, as specified in [35], state that the overall failure rate
should be “at the percent level” per year [24]. This includes all electronics in the readout
chain, plus the silicon photo-multipliers that comprise the active detector. Based upon this
analysis, the predicted failure rate is within the stated requirements, approximately 1/3 of
the allocation per year, which falls within the specification, with margin.

One advantage that the Calorimeter Front End Board electronics has is the active
cooling system that will be in place, providing an operating environment estimated to have
a temperature of 12 C. The normal reference temperature for this analysis is 25 C. As
described throughout the document, higher ambient temperatures accelerate the lifetime of
electronics. It is also true that lowering the ambient temperature will decelerate aging.
This can be seen in Table 3.3.1. The cooler environment for this system provides an
improvement of about 27% in the hazard rate, MTTF, and probability of failure. This also
translates to an improvement in failure rate, with approximately 25% fewer boards
predicted to fail in the first year.

Reliability vs. Temp. 25C 12C

Hazard Rate (per 1E9 Hrs) 710 519

MTTF 1.41E6 1.92E6
Prob. Failure 1st Yr 0.50% 0.36%

Table 3.3.1. Comparison of Reliability Performance

at Cooler Operating Temperature vs. Room Temperature

The analysis can be used to identify the parts that have the poorest reliability. The
parts having the highest predicted failure rates for the board are shown in Table 3.3.2. The
table is expressed in FIT values (failures per 1E9 hours.) Topping the list are the
connectors. These have a high base hazard rate, as prescribed from the handbook, although
the acceleration factors are modest. The next highest category of failures comes from the
FETs, which also have a relatively large base hazard rate, as prescribed by the handbook..
It is worth noting that in the early days of FET fabrication, the reliability was not as good
as it is today, so these values may not be truly representative of modern-day FET
performance. High voltage capacitors C24 and C28 round out the list. C28 is a bypass
capacitor on the 230V coming into the board for biasing the SiPMs. Likewise, C24 is a
bypass capacitor on the output side of the bias regulator, which nominally operates at 200V.
Both capacitors are rated for 250V. The fact that these capacitors operate so close to the
rating is the primary contribution to the high value of the weighted FIT.
 50

Reference Part Base Acc. Weighted

FITS Factors FITS

J2 Card edge connector 46.0000 1.5596 71.7438

J3 3-pin Jumper 46.0000 1.5596 71.7438
M1 AO3162 FET 12.0000 3.8624 46.3490
M2 BSS126 FET 12.0000 3.7735 45.2820
M3 AO3162 FET 12.0000 3.8624 46.3490
J1 SiPM connector 7.0000 1.5596 10.9175
C28 10 nF capacitor 2.0000 4.9017 9.8034
C24 100 nF capacitor 2.0000 4.4136 8.8271

Table 3.3.2. Parts on the Board having the Highest Predicted Failure Rates

As described in Section 3.2, the MTTF decreases by a factor of 17.6 when the
environment factors for harsh condition AUC are incorporated in the calculations. This is
significant, and if taken literally, would imply that the reliability of the board would miss
the performance goal.

The consideration of the effect on the MTTF under harsh environmental conditions
was introduced earlier. The handbook defines “AUC, Airborne Uninhabited Cargo.” as,
“Environmentally uncontrolled areas which cannot be inhabited by an aircrew during
flight. Environmental extremes of pressure, temperature and shock may be severe.
Examples include uninhabited areas of long mission aircraft.” There are some similarities
with this definition to the Mu2e experiment. Certainly, the front-end electronics will be
inaccessible during the running of the experiment, where access is expected approximately
once per year for maintenance. The experiment will have wide variation in air
pressure/vacuum. Temperature swings will be present, although likely not to the level
experienced in uninhabited aircraft. There should not be much mechanical vibration or
shock in the Mu2e detector, although the detector train will move, albeit with much smaller
forces and velocity. However, a significant environmental factor for the Mu2e front-end
electronics is radiation damage, which is not mentioned in the definition of AUC, even
though there is a radiation effect in aircraft. The thinning of the atmosphere results in a
higher flux of cosmic rays, resulting in a higher radiation dose, albeit small, compared to
ground-based operation. Even so, the most extreme dose and fluence levels in aircraft are
orders of magnitude smaller than in Mu2e. So, there are similarities between the two
environments, but differences as well. Again, given the uncertainties in the definition of
associated environmental acceleration factors for the different types of electronic
components, consideration of the AUC environment provides a sense of how the MTTF is
affected by harsh environmental conditions. That said, the results from the inclusion of
the environmental acceleration factors should be regarded as illustrative only, with
uncertain expectation that it will match realistic operating conditions concerning reliability.
The exercise does serve though to underscore the importance of understanding the radiation
tolerance of the electronic components in a front-end design.
 51

3.4. Interpretation of this Analysis

As stated above, military handbook MIL-HDBK-217F was developed to be a

reference, providing a methodology for calculating the reliability of electronic boards. It
is inherently an estimation tool. To make this intent clear, the first sentence of the
handbook states, “This handbook is for guidance only and shall not be cited as a
requirement.” It continues, “The handbook is intended to be used as a tool to increase the
reliability of the equipment being designed.” It is in this spirit that the analysis and results
presented in this document are to be interpreted.

The handbook is not without caveats, limitations, and other issues, some of which
are identified below.

• The handbook tends to generalize reliability aspects across broad ranges of

manufacturers, manufacturing processes for a given type of component, and
differences that can occur in manufacturing lots or batches. The methodology in the
handbook was developed from the collective experience of the developers, which
included reliability measurements on certain components, as well as general
operating experience over some range of projects and applications. Much of the
background is referenced in mil-spec documents. As such, the analysis
incorporates general trends, but does not represent the reliability of the actual parts
used.

• The handbook is quite dated. The first edition was released in 1961, around the
advent of the marketing of transistors. The original basis for the handbook was
experience with vacuum tube electronics. There have been several releases since
that first version, attempting to include parts and developments in technology as
they evolved. The latest version, MIL-HDBK-217F, was released in 1991. Since
this time, there have obviously been additional major advances in integrated circuit
technology nodes (fabrication feature sizes.) For example, in 1991, the CMOS 0.35
um technology node began production. Today the smallest technology node in
production is 5 nm, a factor of 70 in feature size, or 1400 in area, which matches
the predictions of Moore’s Law [23-24]. New IC fabrication companies have come
and gone in this time. The 1991 handbook makes virtually no reference to
technology nodes. In another example, the majority of electronic components
manufactured today are surface mount, in a variety of sizes, packages, and
materials, but in the 1991 handbook, surface mount parts tend to be lumped into
single categories, although some discrimination is provided for power ratings.
Many important differences in surface mount components and packages that are
known today to affect reliability have been neglected. While it appears that the
Dept. of Defense will not be issuing any more updates to the handbook, the VITA
Standards Organization (VSO) [25] has produced updates [26], which some
reliability analyses have been incorporating. These updates from VITA have not
been included in this analysis.
 52

• The handbook methods are based upon elementary reliability concepts, which do
not take into account newer developments in reliability analysis. Some of these can
be found in [27-30]. In addition, a recent thrust has been in the study of “physics
of failure” [31]. The analysis described herein is based entirely on the methodology
set forth in the handbook, and does not include any of these modern aspects to
reliability theory. An overview of the foundation for this analysis is provided in
the Appendix.

As a result, the methodology in this handbook might appropriately be regarded as

a tool for evaluating design and parts choices in a relative way rather than predicting the
reliability performance form an absolute perspective. Despite these limitations, the
handbook is widely used in the design and vetting of electronic systems used in military
and aerospace applications, where a premium is placed upon high-reliability design and
operation due in part to personnel safety, but also in consideration of the situation where
accessibility is limited, making it difficult or impossible to service the electronics when it
fails. In high-risk applications such as aerospace, the component testing regimen often
incorporates aspects of the handbook as specifications for qualifying parts for use. For
electronics systems where reliability is an important aspect of performance, the need for
an evaluation framework like this is evident. Indeed, the handbook has also found
extensive use in commercial electrical systems where reliability is an important part of the
performance goals. Reliability is certainly important for on-detector electronics in HEP
experiments that have limited access and “harsh” environmental conditions. The analysis
described herein can thus be viewed in the context of standard reliability calculation
techniques used in the broad electronics community, but with the caveats and limitations
described above.
 53

4. Appendix I – Overview of Reliability Analysis Methodology

The following overview comes from [32]:

In general, there are three categories of failures in electronic systems, and

correspond to three periods of lifetime:

1. Early failure, also known as the Infant Mortality period. This type of failure
happens early in the lifetime of a component, and is usually caused by
manufacturing defects.
2. Useful lifetime, also known as the period of constant failure rate. In this period,
failures are random, but occur with an overall constant rate.
3. End of life, also known as the Wear-Out period.

Taken together, these periods comprise the “bathtub curve,” as shown in Fig. 4.1.1 [36].
Generally speaking, this note is concerned with calculating the (constant) failure rate for
the useful lifetime period.

Fig. 4.1.1. Bathtub Curve, showing the three types of failures and their associated
periods. (Courtesy of Wikipedia [26].)

In general, failures of electronic components and systems during their useful

lifetime tend to have an exponential form [22-23]. A plot of the cumulative number of
failures as a function of time, represented as a fraction of the total number of a given
component type, typically looks like that in Fig. 4.1.2, and has the form:

=1− , 0≤t≤∞ (4.1)

where λ is a constant.
 54

Fig. 4.1.2. Typical distribution of the cumulative failures of a given type of

electronic component as a function of time, expressed as a fraction of
the total population.

The quantity F(t) is called the Cumulative Distribution Function (CDF). The CDF can be
interpreted as:

1. F(t) is the probability that a random component of a specific type and value in a
system fails by time t; or
2. F(t) is the fraction of all like components in a system fail by time t.

Note that while the point at which F(t) = 1 effectively represents the failure of all
components of a particular type in the entire system, this is distinctly different from the
“wear out” failure rate shown in the bathtub curve of Fig. 4.1.1. Wear out comes from
fatigue from use, whereas the failures in the useful lifetime is considered to be random
events related to imperfection in the manufacturing process. Typically, parts wear out
much sooner than the random failures would deplete the population.

It is useful to consider the probability of a failure occurring as a function of time.

The Probability Density Functions (PDF) is defined as:
b
λ∗
b
, 0≤t≤∞ (4.2)
 55

where λ is a constant, which will be discussed shortly. A plot f(t) is shown in Fig. 4.1.3,
for the case where λ = 1E6. The quantity f(t) dt represents the fraction of failure times in
interval dt.

Fig. 4.1.3. Plot of a PDF having an exponential distribution

The Reliability function (also called the Survival function), is defined as:

R(t) = 1 - F(t) (4.3)

The Reliability function can be interpreted as:

1. R(t) is the probability that a random component of a specific type and value in a
system will still be operating after t hours; or
2. R(t) is the fraction of all like components in a system that will still be operating
after t hours.

At a given time τ, some number of failures will have occurred in the system. The
probability of failure in the next ∆τ of time is expressed as a conditional probability, the
probability of failure in the next ∆τ of time given that number of components that have
survived to time τ:
 56

Ž •• ‘• Ž •
G G Š ∆‹ | •• ‹ =
V •
(4.4)

Of interest in reliability analysis is the rate of failure, also known as the hazard rate
or instantaneous failure rate. This is denoted as h(t), and is defined as:

Q“ Ž •• ‘• Ž • •
ℎ =
•→H ‘• V • V
= (4.4)

For the case where f(t) is an exponential as represented in (4.2), the hazard rate reduces
to:

ℎ = (4.5)

The units of λ are in “number of failures per unit time,” which is a failure rate.
Electronic component manufacturers often express failure rates in terms of the number of
failures in 1E9 hours, which is called “failures in time” or FITs.

In the general case, h(t) can vary as a function of time. Of interest is the Average
Failure Rate, or AFR. Over a time period t2 – t1, this is defined as:
˜
—˜ = e b
– i − = 6
(4.6)
= 6

Again, for the case where f(t) has an exponential form as shown in (4.2), the AFR is:

– i − = (4.7)

Thus, assuming an exponential form of the Cumulative Distribution Function for the
failures of electronic components, the FITs value will be an indicator of the average
failure rate of the components in the system.

Of interest in reliability analysis is the Mean Time to Failure, of MTTF. This is

related to the hazard function as:

„cc = (4.8)

Note that MTTF is for the case where components fail in a system and are not replaced
(immediately). This would be the situation for a detector in which access to perform
repairs is infrequent. This should not be confused with the term Mean Time Between
Failures, or MTBF, in which components are replaced as they fail. These terms are often
used interchangeably, although their meaning is different.

The goal of this analysis is to calculate the value of the hazard function for an
entire printed circuit board. A printed circuit board typically has many components on it,
 57

each having their own hazard function. The probabilities of failure (or survival) of the
different components must be combined in order to get the overall probability of failure
(or survival) for the entire board. For two events, A and B, the probability of them
occurring simultaneously (intersection of event spaces) is given by conditional
probability:

C M⋂
–|™ =
C
(4.9)

Rearranging:

– ⋂™ = ™ ∗ –|™ (4.10)

If the events are independent, then:

–|™ = – (4.11)

Then:

– ∩ ™ = ™ ∗ – (4.12)

For a simple printed circuit board that has two different components on it, A and B,
having probabilities of survival P(A) and P(B) respectively, the probability of them both
surviving as a function of time t is obtained by multiplying the two probabilities together
to get the overall probability. This assumes that their failures are independent of each
other, that the failure of one does not cause the failure of the other. For simple reliability
calculations, this is what is generally assumed. If component A have a hazard rate of λA,
and component B have a hazard rate λB, then the combined probability for survival as a
function of time t is given by:

œ…. = • ∗ " = [ •• " ∗ ]

(4.12)

Extrapolating to a printed circuit board containing M components, each with hazard rates
λ1, λ2, …λM respectively, the hazard rates are added together to give an overall hazard
rate for the board λBD:

= ∑Ÿ (4.13)

Once the overall hazard rate for a board is known, the probability of having a
board failure as a function of time is given by Cumulative Distribution Function:

= = 1− "# ∗ (4.14)

For a system having N identical boards, the CDF can be used to calculate the
number of expected failures of boards in the system:
 58

# = ,∗ = ,∗ 1− "# ∗ 6 (4.15)

For any given component on a board, there may be several factors that contribute
to the hazard rate. Examples include temperature, mechanical or electrical stress, overall
quality of the part, the environment, etc. One could model this as individual hazards or
failure mechanisms. Assuming that the failure mechanisms are independent, the overall
survival probability would be given by the product of the individual probabilities for each
failure mechanism, i.e., the net probability that all independent failure mechanisms will
not occur at time t. Rather than doing this, the approach used in MIL-HDBK-217F is to
define a base hazard rate, λb, and then define multiplicative factors that are functions of
the particular failure mechanism. The resulting hazard rate, λp, has the form,

λp = λb * π1 * π2 * … * πK , for K failure mechanisms (4.16)

Handbook MIL-HDBK-217F describes the different failure mechanisms, and provides

guidance on how to calculate the πi factors for each type of component.
 59

5. References

[1] “Reliability Prediction of Electronic Equipment,” MIL-HDBK-217F, Washington,

DC 20301: Department of Defense, December 1991.
[2] R. Ray, et al., “Mu2e Technical Design Report,” Mu2e Internal Report, DocDB
4299, Mar. 11, 2015.
[3] N. Atanov, et al., “The Calorimeter Final Technical Design Report,” Mu2e
Internal Report, DocDB 8429, Dec. 28, 2016.
[4] L. Lucchesi, “Test of the Calorimeter Front End Unit,” Mu2e Internal Report,
DocDB 14556, Nov. 28, 2017.
[5] G. Corradi, et al., “Technical description of the Calorimeter FEE Boards,” Mu2e
Internal Note, DocDB 29910, Nov. 18, 2019.
[6] G. Corradi, et al., “Technical Specification of Calorimeter FEE Boards and
Cables,” Mu2e Internal Note, DocDB29817, Nov. 18, 2019.
[7] G. Corradi, et al., “Summary of FEE Rad Test,” Mu2e Internal Note, DocDB
24486, Nov. 15, 2019.
[8] G. Corradi, et al., “Temperature Trend in Calorimeter FEE Boards,” Mu2e
Internal Note, DocDB 29814, Nov. 11, 2019.
[9] G. Corradi, et al., “Design of the Calorimeter FEE and Mezzanine Boards,” Mu2e
Internal Note, DocDB 27589, Jul. 17, 2019.
[10] G. Corradi, et al., “Neutron and TID Test of Calorimeter FEE,” Mu2e Internal
Note, DocDB 22903, Dec. 11, 2018.
[11] G. Corradi, et al., “Mu2e Calorimeter Development Electronics FEE Review,”
Mu2e Internal Note, DocDB 14634, Dec. 4, 2017.
[12] G. Corradi, et al., “CD2 EMC Front End Electronics,” Mu2e Internal Note,
DocDB 4686, Oct. 11, 2014.
[13] F. Abe et. al., "The CDF Detector: An Overview," Nucl. Instrum. Meth., vol.
A271, 1988, pp. 387-403.
[14] Bin Lu, Luke W. Mo, Thomas A. Nunamaker, “The Cockcroft-Walton
photomultiplier tube base and the Ethernet high voltage controller,” Nuc. Inst.
Meth. A, Vol. 313, Issues 1–2, 1992, pp. 135-141, ISSN 0168-9002,
https://doi.org/10.1016/0168-9002(92)90089-M.
 60

[15] A. Byon-Wagner, K. Byrum, J.W. Dawson, G. Drake, C. Drennan, G.W. Foster,

W.N. Haberichter, J. Hoff, S. Kuhlmann, M.A. Lindgren, L.J. Nodulman, J.
Proudfoot, J.L. Schlereth, and J.Y. Wu, "The Shower Maximum Front-End
Electronics for the CDF Upgrade," IEEE Trans .Nucl. Sci., vol. 49, 2002, pp.
2567-2573.
[16] S. Fourletov, “Straw tube tracking detector (STT) for ZEUS,” Nuc. Inst. Meth. A,
Vol. 535, Issues 1–2, 2004, pp. 191-196, ISSN 0168-9002,
https://doi.org/10.1016/j.nima.2004.07.212.
[17] T. Cundiff, J. W. Dawson, L. Dalmonte, G. Drake, T. Fitzpatrick, W. Haberichter,
D. Huffman, W. Luebke, C. Nelson, D. Reyna, J. L. Schlereth, P. Shanahan, J. L.
Thron, and M. Watson, "The MINOS Near Detector Front End Electronics,"
IEEE Trans. Nucl. Sci., vol. 53, 2006, pp. 1347-1355.
[18] H. P. Beck, M. Abolins, et al., “Performance of the Final Event Builder for the
ATLAS Experiment,” IEEE Trans. Nucl. Sci., vol. 55, 2006, pp. 177-181.
Presented at the 15th IEEE 2007 NPSS Real Time Conference, Batavia, IL, Apr.
29-May 4, 2007.
[19] F. Tang, et al., “Design of the Front-End Readout Electronics for the ATLAS Tile
Calorimeter at the LHC,” IEEE Trans. Nucl. Sci., vol. 60, 2013, pp 1255-1259.
[20] G. Drake, et al., “Design of a New Switching Power Supply for the ATLAS
TileCal Front-End Electronics,” J. of Inst., vol. 8, 2013, article C02032.
[21] D. L. Adams, et al., “Photon Detector System Timing Performance in the DUNE
35-Ton Prototype Liquid Argon Time Projection Chamber,” J. of Inst., vol. 13,
2018, article P06022.
[22] G. Drake, “Radiation Tolerance requirements for the Mu2e Front End
Electronics,” Mu2e Internal Note, DocDB 11288, June, 21, 2017.
[23] “Technology Node – Wiki Chip,” https://en.wikichip.org/wiki/technology_node
[24] “Moore’s Law,” https://en.wikipedia.org/wiki/Moore%27s_law
[25] VITA Standards Organization, https://www.vita.com/.
[26] “VITA 51.1 - Reliability Prediction MIL-HDBK-217 Subsidiary Specification,”
13th Edition, 2008, with Reaffirmation Notice , 2018.
[27] A. Mosleh, Y.H. Chang, “Model-Based Human Reliability Analysis: Prospects
and Requirements,” Reliability Engineering & System Safety, vol. 83, num. 2,
2004, pp. 241-253, ISSN 0951-8320, https://doi.org/10.1016/j.ress.2003.09.014.
[28] H.-Z. Huang, “Reliability Analysis Method in the Presence of Fuzziness Attached
to Operating Time, Microelectronics Reliability, Volume 35, Issue 12, 1995, pp.
1483-1487, ISSN 0026-2714, https://doi.org/10.1016/0026-2714(94)00173-L.
 61

[29] G. Levitin, “Computational Intelligence in Reliability Engineering: Evolutionary

Techniques in Reliability Analysis and Optimization,” Volume 39 of Studies in
Computational Intelligence,” Springer Science & Business Media, 2006, ISBN
3540373675, 9783540373674, 398 p.
[30] Jiang, C., Lu, G.Y., Han, X. et al., “A New Reliability Analysis Method for
Uncertain Structures with Random and Interval Variables,” Int J Mech Mater
Des, vol. 8, 2012, pp. 169–182, https://doi.org/10.1007/s10999-012-9184-8
[31] M. Pecht and A. Dasgupta, "Physics-of-Failure: An Approach to Reliable Product
Development," IEEE 1995 International Integrated Reliability Workshop, Final
Report, Lake Tahoe, CA, USA, 1995, pp. 1-4.
[32] P. Tobias, D. Trindade, “Applied Reliability,” Van Nostrad, 1986, ISBN 0-442-
28310-5.
[33] MENDENHALL, W. & HADER, R. J. (1958), "Estimation of parameters of
mixed exponentially distributed failure time distributions from censored life test
data", Biometrika, 45, 1958.
[34] Ansell, J., & Phillips, M. (1989), “Practical Problems in the Statistical Analysis of
Reliability Data,” Journal of the Royal Statistical Society. Series C (Applied
Statistics), 38(2), 205-247. doi:10.2307/2348057
[35] S. Giovannella, S. Miscetti, “ Mu2e Calorimeter Operational Requirements,”
Mu2e Internal Note, DocDB 22786, V7, Feb. 11, 2020.
[36] D. Smith, “Reliability, Maintainability and Risk - Practical Methods for
Engineers,” Elsevier, 9th Ed., 2017, ISBN 978-0-08-102010-4.
 Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Contents lists available at ScienceDirect

Nuclear Inst. and Methods in Physics Research, A

journal homepage: www.elsevier.com/locate/nima

Assessment of operational availability for the PIP-II Superconducting Radio

Frequency linear accelerator facility
Arun Saini a ,∗, Ram Prakash a,b,c , Joseph D. Kellenberger a
a Fermi National Accelerator Laboratory, Batavia IL 60510, USA
b
Raja Ramanna Centre for Advanced Technology, Indore 452013, India
c
Homi Bhabha National Institute, Anushaktinagar, Mumbai 400094, India

ARTICLE INFO ABSTRACT

Keywords: Operational availability is a critical performance measure for an accelerator facility in modern time. A
SRF accelerator facility high availability enables the facility to serve a wide range of users simultaneously. Consequently, besides
Operational availability pure accelerator physics considerations, newly proposed accelerator facilities account for the availability and
Down-time
reliability aspects in the design phases. It allows incorporation of appropriate mitigation strategies for the most
vulnerable systems in the machine and therefore, minimizes unscheduled interruptions during the operation.
This paper lays out a methodology for the availability assessment of the complete particle accelerator facility
and presents an initial assessment of the availability of the newly proposed Proton Improvement Plan-II (PIP-II)
accelerator facility at Fermilab. The paper describes a comprehensive reliability model of the PIP-II facility that
comprises not only 800 MeV linear accelerator (linac) system but also essential utility systems in the form of
cryogenic, water, power and air systems. The paper details estimations of the availability of the PIP-II facility
for two operational modes i.e. the nominal operational mode featuring 800 MeV beam and critical operational
mode involving operation with the lowest objective beam energy of 600 MeV.

1. Introduction The SRF technology brings in multiple advantages to an accelerator fa-

cility. It enables not only a high duty beam operation but also facilitates
Availability analyses have been a standard protocol in industries a high accelerating gradient in the cavities. With all numerous benefits,
where the operational costs are taken into account at design level the SRF technology also brings in an adverse feature in terms of the
of a new product. However, practice of the reliability engineering in additional complex systems (cryostat, cryo-plant, cryogen distribution
research infrastructures, that are usually driven by a fixed construc- etc.) needed for its implementation to an accelerator facility. Fur-
tion cost, is relatively new. The late introduction of the reliability thermore, repair or replacement of a malfunctioned superconducting
component is both expensive and time consuming. Restoring a nominal
engineering in the particle accelerator design is mainly due to a very
accelerator operation after an interruption also takes time. This in turn,
complex nature of the machine. Every particle accelerator is unique in
reduces overall availability of an accelerator. Consequently, modern
the design and operation. Thus, input data required for such analyses
SRF accelerator facilities are sighting importance of the availability
are usually limited and specific to a given system. This adds uncer-
and reliability analyses [5–8] to reduce operational cost of the unre-
tainty against realization of the reliability engineering aspect in the
liable accelerators. To assure a reliable operation with the minimal
accelerator design. However, this trend has been changing lately and
unscheduled interruptions, the reliability engineering aspects need to
the modern accelerator facilities are anticipating importance of the
be considered from the design phases of the SRF accelerators. Per-
reliability engineering in the accelerator design. forming the availability analysis at various stages of the design enables
Most of newly proposed accelerator facilities around the world such identification of critical components with a higher probability of failure
as Linac Coherent Light Source-II (LCLS-II) [1], European Spallation as well as prediction of the unscheduled down time during operations.
Source (ESS) [2], Indian Spallation Neutron Source (ISNS) [3], China This in turn, may allow developing a mitigation strategy for critical
Accelerator Driven System (CADS) [4], etc. are based on the Super- components, appropriate allocation of redundancy and, resources for
conducting Radio Frequency (SRF) technology. Recent advancements in spare and replacement parts.
the SRF technology make its usage more practical and cost effective for In the past, the availability analysis for the particle accelerators
large accelerator facilities as well as for the commercial applications. was often carried out either for certain sub-systems of the machine

∗ Corresponding author.
E-mail addresses: [email protected] (A. Saini), [email protected] (R. Prakash).

https://doi.org/10.1016/j.nima.2020.164874
Received 14 April 2020; Received in revised form 16 November 2020; Accepted 16 November 2020
Available online 19 November 2020
0168-9002/© 2020 Elsevier B.V. All rights reserved.
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Table 1
Design specifications for operational beam parameters of the PIP-II linac.
Parameter Magnitude Units
Final beam energy 800 MeV
Beam pulse repetition rate 20 Hz
Beam pulse length 0.55 ms
Average CW beam current 2 mA
Final 𝜀z <0.4 mm-mrad
Final 𝜀t ≤0.3 mm-mrad

𝜀z normalized RMS longitudinal emittance; 𝜀t normalized RMS transverse Fig. 1. Block diagram representation of the PIP-II Linac. Red coloured blocks represent
emittance. the warm sections (RT) whereas the blue blocks represent superconducting sections (SC)
operating at 2 K. Normalized design velocity (𝛽) of the cavity in each section is also
shown. (For interpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)
(e.g. cryo-plant, RF system etc.) or considering a simple form of the
major beamline elements [9–13]. For this reason, the paper lays out
a methodology for the availability assessment of the complete particle Table 2
Optics elements and transition energy in each section of the PIP-II SRF linac.
accelerator facility. It describes a comprehensive reliability model for
Section CM Cav/Mag Operating Energy
the availability assessment of the Proton Improvement Plan-II (PIP-
per CM Frequency (MeV)
II) SRF accelerator facility [14] that includes not only the accelerator
HWR 1 8/8 162.5 MHz 2.1–10
components but also essential utility systems in terms of the water, SSR1 2 8/4 325 MHz 10–32
air, cryogenic and power systems. Furthermore, the model implements SSR2 7 5/3 325 MHz 32–177
the accelerator components in their detail composition that implies an LB 9 4/1a 650 MHz 177–516
accelerator component is described with its essential auxiliary systems. HB 4 6/1a 650 MHz 516–833
For an instance, an accelerator cavity in the model is implemented with a
Normal conducting quadrupole doublet.
its power coupler, frequency tuner and RF power source. Thereafter, the
paper discusses studies for the PIP-II facility that lead to finding of the
most critical section determining the unavailability budget of the PIP-II these families, the linac is segmented into five SRF sections named
facility. Lastly, the paper converses the input data sensitivity analysis as, Half Wave Resonator (HWR) [20], Single Spoke Resonator (SSR)
assessing impact of a spread in the reliability input data on the model 1 & 2 [21,22], and Low Beta (LB650) and High Beta (HB650) [23].
prediction and, validation of the model methodology using a reference Table 2 highlights configuration of each section and includes details of
model of the existing operational accelerator facility. a number of cryomodules (CM), focusing magnets and cavities as well
The paper is organized in seven sections. Section 2 provides an as operating frequency of cavities and their accelerating ranges. Note
overview of the PIP-II SRF linear accelerator whereas Section 3 in- that, superconducting solenoid magnets are used in the HWR, SSR1 and
troduces key definitions and concepts of the availability analysis for SSR2 sections whereas normal conducting (NC) quadrupole magnets
an accelerator system. Section 4 discusses preparation of the PIP-II arranged in doublet configuration are utilized in the LB650 and HB650
accelerator facility model and describes components selection criteria, sections for the transverse beam focusing.
operational modes and the high-level functional block diagram of the The linac optics has been carefully designed to deliver a high-quality
facility. Section 5 converses results of the availability analyses while beam at the Booster entrance. Fig. 2 shows the accelerating voltage and
Section 6 presents a sensitivity analysis and the model benchmarking output energy at each cavity along the linac for the baselined optics.
with an operational accelerating facility. The paper concludes with a Detailed description of the linac architecture and its optics design has
summary in Section 7. been presented elsewhere [24].

2. PIP-II SRF linac accelerator facility 3. Availability formalism for accelerators

Fermilab is planning to perform a systematic upgrade to its existing There are many good text books [25,26] dedicated to the reliability
accelerator complex to support a world leading neutrino program. A engineering theory. For the comprehension of this article, this section
comprehensive roadmap named ‘‘Proton Improvement Plan (PIP)’’ has introduces necessary theory and, discusses how it is applicable in the
been established. The second stage of the Proton Improvement Plan framework of accelerators.
comprises construction of a new superconducting linear accelerator The failure rate (𝜆) of a component through its life span usually
(linac) capable of accelerating a 2 mA H− ion beam up to 800 MeV follows a bath-tub distribution as shown in Fig. 3. Initial portion of
in a continuous wave (CW) regime. However, the initial operational the bath-tub curve is called the burn-in period that consists of a high
goal is to deliver a 1.1% duty factor pulsed beam to the existing Booster failure rate due to the infant mortality. Similar behaviour is observed
synchrotron [15]. The PIP-II accelerator facility aims at the operational at the end of the curve due to deterioration of components. This period
availability of 90% over a fiscal year [16]. Table 1 summarizes the most is defined as the wearing-out period. Between these two regions, a
relevant operational beam parameters of the PIP-II linac. system has a useful life period which consists of a relatively lower
A schematic of the SRF linac’s architecture is shown in Fig. 1. It and constant failure rate. Assuming, accelerators also follow the bath-
is composed of a warm front-end and an SRF accelerating section. The curve analogy. The burn-in period is then referred to the commissioning
warm front-end comprises an H− ion source (IS) capable of delivering a period when the accelerators are being actively tuned and tested to
15 mA, 30 keV, DC or pulsed beam, a 2 m long Low Energy Beam Trans- deliver operational parameters. A wear-out period for the accelerators
port (LEBT) line [17], a 162.5 MHz, CW Radio Frequency Quadrupole is the period when an upgrade or replacement is needed to maintain
(RFQ) [18] that accelerates the beam to 2.1 MeV and a 13 m long its operational performance. In this paper, the main emphasis is on the
Medium Energy Beam Transport (MEBT) line [19] that includes variety useful period of an accelerator which can be interpreted as its nominal
of diagnostic devices and a chopper system capable of generating operational period. In subsequent sections, the availability model is
an arbitrary bunch pattern before the beam is injected into the SRF solved using the assumption of a constant failure rate of components.
section. Note that, the assumption not only justifies the bath-tub analogy but
The MEBT is followed by the SRF linac that uses five families also permits solving the model analytically which otherwise becomes
of SRF cavities to accelerate the beam up to 800 MeV. Based on too cumbersome to solve analytically for the large systems.

2
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Fig. 4. Evolution of operating states of a binary system with time.

3.1. General formalism

Fig. 4 illustrates a system which has only two operating states

i.e. working state and failed state. The length of time for which the
system keeps a working state is called Time to Failure (TTF) whereas
the time taken to repair the system after a failure is termed as Time to
Repair (TTR). The time between successive failed states is quantified
as Time Between Failure (TBF). These times are collectively called
the characteristic times of a system. In a n-component system there
are several ways a system might fail and be repaired. Thus, it is
more appropriate to determine the characteristics times from mean of
respective distributions.
The Mean Time Between Failure (MTBF) is then expressed as:

𝑀𝑇 𝐵𝐹 = 𝑀𝑇 𝑇 𝐹 + 𝑀𝑇 𝑇 𝑅; (1)

where Mean Time To Fail (MTTF) is the statistical average of operating

time and Mean Time To Repair (MTTR) is the statistical average of the
repair time for a system. In case of a constant failure rate (𝜆), MTTF
for a non-repairable system can be written as:
Fig. 2. (a) Accelerating voltage and (b) output energy along the PIP-II linac for the 1
nominal optics. Note that, the bunching cavities in MEBT operates at −900 synchronous 𝑀𝑇 𝑇 𝐹 = . (2)
𝜆
phase and therefore provides no acceleration. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.) The reliability of an accelerator is defined as the probability that it
does not fail in a given mission time whereas the availability (𝐴) is
proportion of its ‘‘up time’’ to the total operational time over a defined
operation period. It can be quantified as:
𝑀𝑇 𝑇 𝐹
𝐴= . (3)
𝑀𝑇 𝑇 𝐹 + 𝑀𝑇 𝑇 𝑅
One can conclude from their definitions that the mean time between
the failure and the failure rate are measures of the reliability. To obtain
a high reliability and availability, an accelerator must avoid repetitive
failures and long down time after a failure occurrence. Thus, decreasing
the mean time to repair is one of the main design considerations.
Common strategies to minimize MTTR include frequent monitoring
of accelerator-systems to identify probable issues before failure occur-
rence; proper distribution of diagnostic devices to minimize diagnostic
Fig. 3. Evolution of the characteristic failure rate function of a system over a length time, appropriate allocation of spares to reduce logistics and minimiz-
of time. ing the replacement time; establishing a dedicated team of experts to
perform quick repairs, etc. It is clear that improvement in MTTR is
achieved at the expense of an increase in overall cost of the facility.
The cumulative experience with existing operating accelerator facil- A balance must be obtained between objective MTTR and the resulting
ities suggests a gradual degradation in performance of the accelerator cost.
components over a period of time. For instance, surface contaminations The foremost step in estimation of the availability of an accelerator
of the SRF cavities may reduce the maximum available accelerating gra- is to obtain availabilities of individual components using their MTTF
and MTTR input data. Then, the next step involves acquiring infor-
dient. These gradual degradations in the operational performance over
mation of the component functional roles in the accelerator-system.
time are called parametric drift failures in the reliability engineering.
This system-component functional relationship is often expressed in the
Adding a safety margin in the operational parameters and using new
form of a Reliability Block Diagram (RBD) where each component is
advances in the accelerator technology, such as plasma processing of represented in the form of a block. Fig. 5 shows the most common
the SRF cavities [27], and allowing others to address the parametric system-component functional relationship. In this example, the com-
failures in the accelerators at some extent. Thus, in this article, it is ponents are connected in a series (failure of a component leads to a
assumed that a component has only two states of operation either a failure of the overall system, similar to the logical AND gate analogy)
nominal working state or a failed state. and in parallel arrangements (failure of a component will not lead to

3
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

• Components involved in the high current operations e.g. modula-

tors.
• A larger set of commercial components as they might not be
designed for high reliability.

Note that, components exhibiting above characteristics are relatively

more vulnerable to failures and therefore, drive the overall availability
of the PIP-II facility.

4.2. Model assumptions

The model uses the following assumptions to compute the availabil-

ity of the PIP-II facility:

Fig. 5. A reliability block diagram representing common system-component functional • As also mentioned earlier, each component in the model possesses
relationship in a complex system. only binary states of the operation i.e. either operating nominally
or failed. The component can migrate any state independent of
its history of the operation.
a failure of the overall system until all parallel connected components • A component exhibits a constant failure rate during its operation.
get failed, similar to the logical OR gate analogy). Note that series and • Each component fails at a random time with an exponential
parallel connections are a limiting case of the ‘‘k out of n’’ system where distribution determined by its MTBF. Two simultaneous failures
the availability of a system with n identical components is obtained as are prohibited in the model. Those uncorrelated component fail-
following: ures are then represented by the Markov chains [26] and solved
∑
𝑘 analytically to evaluate the system availability.
𝑛!
𝐴(𝑡)𝑠𝑦𝑠(𝑟≤𝑘) = 𝐴(𝑡)𝑛−𝑟 (1 − 𝐴 (𝑡))𝑟 (4) • When a component fails, it leads to the system failure (unless
𝑟! (𝑛 − 𝑟)!
𝑟=0 fault-tolerances are specified) resulting in an unscheduled accel-
where r is number of failures, k is maximum allowable failures and erator shut-down. A temporary component failure such as one
A(t) is the availability of the component at time t. When k=0, all resulting from quenching of an SRF cavity or magnet is not treated
components are connected in series while for 𝑘 = 𝑛 − 1, all components as a failure in the model.
are connected in parallel. • The model assumes components meet their design specifications
In an accelerator, a variety of component-system functional rela- and the system is maintained to its best operable condition. Thus,
tionships such as series, parallel, standby, redundant connects etc., the model does not incorporate manufacturing errors, human
might exist simultaneously. A list of formulae for the system availability errors and environmental errors. Additionally, implications of the
and reliability with such configurations has been presented in the drift failures or degradation in performance of components are
Appendix A.1. not included in the model.
• The model implements only corrective maintenance. It implies
4. Availability assessment model for PIP-II the fault detection time, logistic time at various stages of repair,
tuning etc. are excluded. As soon as a failure is detected, the
A comprehensive availability assessment model of the PIP-II accel- maintenance process is launched. After a repair, the component
erator facility in form of the high-level functional block diagram is is treated ‘‘as good as new’’. Thus, resulting availability of the
developed to compute its availability. This section details preparation system is called inherent availability. Note that, the availability
of the model and delineates assumptions and guidelines used to build in this paper is always attributed to the inherent availability.
the model. • The model is further simplified with the assumption that the
facility transits from a no-beam state after a failure to the nominal
4.1. Component selection beam state as soon as a repair is completed.
• A mission time of about a year, equivalent to eight thousand
It is evident that an accelerator comprises numerous components
operational hours, is assumed for the availability analysis of the
and dependent systems. Many of these components need additional
PIP-II accelerator facility.
auxiliary elements to execute their nominal function. For instance, an
accelerating cavity assembly in the beamline is comprised of several
4.3. Operational modes
auxiliary elements such as, power coupler to feed RF power; a me-
chanical tuner to tune its resonant frequency etc. This in turn, adds
A system can require to operate in different modes. These opera-
another layer of elements in the model. Consequently, the model of
tional modes define the system-component functional relationship and
an accelerator facility becomes very large and cumbersome. In order
to resolve this issue, a component-selection criterion was applied to therefore, a failure pattern of the system. Consequently, the system op-
the PIP-II model. A component features any of following characteristics erational availability may vary from one operational mode to another.
is included in its detailed composition as practically permissible while Thus, it is essential to establish operational modes of a system before
preparing the model of the PIP-II facility. estimating its availability. In this article, the availability of the PIP-II
accelerator facility is evaluated for two operational modes named as
• Components having moving parts such as vacuum pumps, cavity the nominal operational mode and critical operational mode.
tuners etc.
• Components operating in pulsed mode such as high voltage 4.3.1. Nominal operational mode
switches, kicker system in the MEBT etc. In the nominal operational mode, the PIP-II facility delivers
• Components that are involved in thermal cycling processes e.g. 800 MeV beam to the Booster synchrotron with the design specifica-
heat exchangers for low conductivity water (LCW). tions listed in Table 1. Note that, the baseline configuration of the
• Components containing a high stored energy, e.g. RF cavities and SRF linac has been designed to accelerate the beam up to 833 MeV.
magnets etc. This additional energy provides a safety margin to achieve the nominal

4
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

operational mode. It has been shown elsewhere [28,29] that the SRF
linac optical design is sufficiently robust to tolerate a failure of optical
element in each SRF section without conceding the design specifica-
tions. Consequently, the nominal operational mode can be achieved in
two ways. The first nominal operational scenario, termed as no-failure-
permit in this paper, involves all optical elements are operating with
their design parameters. In this configuration, any component failure
will produce a complete system failure. The second scenario is named
as the fail-tolerance operation that permits a faulty/malfunctioned
accelerating cavity in each SRF section (HWR, SSR1, SSR2, LB650 and
HB650). It implies that the facility would keep operating even after
a failure of the SRF cavity in each section. Note that, a repair or
replacement of an element in cryogenic environment requires relatively
a longer time in comparison to repair of a normal-conducting element.
Consequently, the fault-tolerances in the availability estimate have
been included only in SRF sections. This choice for the analysis does not
infer the fault-tolerance capability of the normal-conducting sections
and, allows a conservative estimate of the availability. It is worth to
Fig. 6. Components with the minimum MTTFs in the PIP-II model. The colour of the
mention that a conservative assessment is beneficial at the design phase bar represents the component’s association with respective assembly or section. For
where a number of factors (human errors and environmental impacts) instance, red coloured bar shows the MTTF of components in the ion-source assembly.
are relatively less known. (For interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)

4.3.2. Critical operational mode

The lowest permissible beam energy out of the linac, at which the
PIP-II facility could sustain an operation, is specified to be 600 MeV.
This energy is called the critical threshold energy below which Booster
synchrotron operation becomes incompatible due to excessive beam
losses. Availability assessment of the PIP-II facility is also performed
for this mode where the linac delivers 600 MeV beam with same rest
of specifications as listed in Table 1.

5. Availabilty assessment for PIP-II facility

5.1. Input data

The reliability input data, MTTF and MTTR, for components are ac-
quired from various sources including educated guess from the subject
experts, operational experience with similar components at Fermilab
as well as existing accelerator facilities, and from prototype tests. The
beam commissioning of the PIP-II front-end at the Proton Improvement
Plan-II Injector Test (PIP2IT) facility [30] also provided a useful in-
formation about operational reliability of the PIP-II components such
as ion-source, magnet power supplies etc. A few components were Fig. 7. Components with the maximum MTTF in the PIP-II facility model. The colour
of the bar represents the component’s association with respective assembly or section.
commercially available and therefore, corresponding data were readily For instance, red coloured bars represent components in a superconducting (SC) cavity
available. In addition, a few references [5–13] were also used to obtain assembly. (For interpretation of the references to colour in this figure legend, the reader
data that were unavailable otherwise. is referred to the web version of this article.)
Fig. 6 shows most vulnerable components in the PIP-II accelerator
facility model. It can be noticed from Fig. 6 that components in the ion
source assembly possess the minimum MTTF that are followed by the mitigation strategy involves replacing the faulty cryomodule with a
compressor in the air utility system. Fig. 7 shows the most robust and fully-functional spare cryomodule. Then, repair of the faulty-element
reliable components of the PIP-II facility model that have longest MTTF. in the cryomodule is carried out in parallel without affecting the
Note that, a high MTTF implies less frequent failures of the component. accelerator operational time. This strategy restricts the repair time of a
It can be noticed from Fig. 8 that the high voltage transformer in superconducting element to only about a month.
the electrical power grid and the SRF cavities acquire longest MTTR in
the model. Based on previous experience at Fermilab, experts suggest 5.2. High-level functional diagram for the PIP-II facility
that a repair/replacement of such transformer could take up to full two
weeks. Considering an eight-hours work shift per day, the repair time is As a next step for the availability assessment, a high-level functional
then estimated to more than 1000 h (24 × 3x14 >1000 h). Because of block diagram model of the PIP-II facility was developed. The facility,
this, the PIP-II facility envisions two power lines. Electric-power loads as shown in Fig. 9, was modelled in two main parts: Utility systems and
is swiftly shifted from one line to another in case of a failure. A repair is linac systems.
then performed in parallel without a long interruption. Also note that,
repair of an SRF cavity may need warming of the cryomodule from a 5.2.1. Utilities systems
cryogenic temperature to the room temperature, taking cryomodule out A utility system in the model indicates a central facility of the core
from the accelerator tunnel and then, dismantle it to replace/repair the supply essential to operate an accelerator such as a cryo-plant to supply
faulty cavity. It could result in a long unscheduled down time spanning the cryogen for the SRF cavities. The model incorporates four utility
over several months. To minimize this time at the PIP-II facility, the systems that are subsequently discussed in detail.

5
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Table 3
Components and their functions in the respective packages in the HWR cryomodule.
Component Function
Cavity package
Cavity Acceleration, longitudinal beam
focusing
Tuner Tune cavity resonant frequency
Power coupler Feeding RF power to cavity
Interlock sensors and electronics Sensors and electronics
Low Level RF RF control and instrumentation
Solid state Amplifier (SSA) RF power source
RF control package: SSA control and timing
SSA controls RF controls to SSA
SSA timing Timing to the SSA
Magnet assembly package: Solenoid magnets assembly
Magnet power supply Power supply to solenoids
Magnet Transverse focusing of the
beam.
Magnet instrumentation Control system
Fig. 8. Components with the longest MTTR in the PIP-II facility model.
Steering assembly package
Steering Magnet Beam Trajectory Correction
Steering Power Supply Magnet power supply
• Electrical-Power System: The PIP-II accelerator facility envisions two
Vacuum system package
electrical-power substations where one of the substations is available
Vacuum Valves Maintain vacuum
in the standby mode. In an event of failure, the power-load is swiftly Vacuum. Pump Creating vacuum in the
shifted to the standby substation. The model includes major electrical beamline
components such as transformers, switchgears, fuses, circuit breakers Vacuum pump power supply Powering the vacuum pump
and cables. The most vulnerable component in the electrical system Local cryogenic system package
is the Vacuum Circuit Breaker (VCB) which exhibits a higher failure Local cryogenic system Cryogenic distribution, cryostat
rate. Because of this, four out of every eight VCBs are redundant in the structure and control
electrical system. Note that, the model does not incorporate the power
generating system but only the supply system.
• Cryo-plant System: A cryo-plant supplies the cryogen necessary to 5.3. Case study of availability assessment for HWR section
maintain cryogenic temperature of the superconducting cavities. The
main components of the cryo-plant included in the model are the cold
In order to illustrate how the availability assessment is performed,
compressors, turbines, expanders, warm compressors and, associated
this section discusses a detailed case study for the HWR section and
control systems. The warm compressors are the most susceptible to
describes the methodology applied to evaluate the availability of the
failures among the cryo-plant components.
complete PIP-II facility.
• Low Conductivity Water (LCW) System: It delivers water to maintain
The HWR section is the first SRF section in the PIP-II linac. As
the operating temperature of normal conducting water-cooled elements
shown in Table 2, it consists of one cryomodule that comprises eight
such as the RFQ. The LCW system includes circulating pumps, heat
solenoid magnets and same number of HWR cavities. Each solenoid
exchangers, gauges, transducers, flow meters and, valves. Among those
magnet includes the steering magnets to correct the beam trajectories
components, the circulating pumps are more often involved in the
in horizontal and vertical planes. Those beamline elements further
failures. Consequently, the LCW system of the PIP-II facility includes
need auxiliary components to execute their nominal operation. Thus,
a redundant unit per three circulating pumps.
it is more appropriate to describe an essential element in terms of the
• Compressed Air System: The air system supplies compressed air for package including all supporting components. The cryomodule model
cooling of the radiation-cooled components, actuation and control of is then represented using six packages: cavity, RF control, magnet
pneumatic valves etc. Two main components of the air-system are the assembly, steerer assembly, vacuum system and local cryogenic sys-
compressor, and dryer. Each of them has a redundant unit in the model. tem packages. Table 3 lists major components and their functions in
respective packages for the HWR cryomodule.
5.2.2. Linac system Availability assessment for the HWR cryomodule is performed for
The model includes a detailed description of the accelerator system. two operational modes: no-failure-permit and a cavity-fail-tolerance.
Along with the SRF linac (described in Section-II), details of the Beam In a no-failure-permit mode, failure of any component leads to failure
Transfer Line (BTL) [31] were also included in the model. The BTL of the complete HWR cryomodule whereas in, a cavity-fail-tolerance
line is used to transport the beam from the end of the SRF linac to mode, the cryomodule keeps operating even after failure of one out
the Booster entrance. It is about 350 m long and mainly composed of of any eight SRF cavities. Fig. 10 illustrates the functional block dia-
normal conducting quadrupole and dipole magnets. grams of the HWR cryomodule describing logical connections among
As shown in Fig. 9, the utility systems are connected to the linac element packages for two operational modes. In the no-failure-permit
in a series configuration. It implies failure of any functional blocks mode, all elements packages are connected in the series configuration
will shut-down the complete facility. After establishing the component- (Fig. 10(a)). In a cavity-fail-tolerance mode (Fig. 10(b)), all element
system functional relationship, the PIP-II accelerator facility model packages are connected in series with the cavity packages that are
was incorporated in a Python-based program. The program has been configured in seven out of eight arrangement.
developed at Fermilab to automate the availability assessment. It not After establishing the functional diagram for the HWR cryomodule,
only computes availability of the complete facility but also for the next step involves computing availability of individual component in
individual section and component. This feature facilitates finding the an element package using input data of MTTF and MTTR in Eq. (3).
most vulnerable section determining overall availability of the facility. Table 4 shows availabilities of components in the cavity and magnet

6
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Fig. 9. High level functional diagram for the PIP-II accelerator facility.

Fig. 10. Functional block diagram for the HWR cryomodule for two operational modes :(a) no-failure-permit and (b) a cavity-fail-tolerance.

packages. Then, using the knowledge of components logical connec- where 𝐴𝑝 is the element package availability, 𝐴𝑖 is the availability of
tions in an element package, availability of the package is evaluated. 𝑖th component in the package and N is total number of components in
∑
Since components are connected in series configuration in the packages, a package. Similarly, failure rate of the 𝜆𝑝 = 𝑁𝑖=1 𝜆𝑖 element package
availability of a package is obtained using equation: is computed as:
∏
𝑁 ∑
𝑁
𝐴𝑝 = 𝐴𝑖 (5) 𝜆𝑝 = 𝜆𝑖 (6)
𝑖=1 𝑖=1

7
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Fig. 11. HWR cryomodule is modelled using six essential element packages that are connected in a series configuration with the cavity package. (a) Combined availability of each
essential package and (b) availability of the full HWR cryomodule for two operational modes i.e. no-failure-permit (Case 1) and a cavity-fail-tolerance (Case 2).

where 𝜆𝑝 is the failure rate of an element package and, 𝜆𝑖 is the failure

rate of individual components connected in a series configuration. As
shown in Table 4, the failure rate of the cavity and magnet packages
are 3.45E−05 and 1.2E−05 per hour respectively. Thereafter, the com-
bined availability (𝐴𝑐𝑝 ) and Mean Time Between Failure 𝑀𝑇 𝐵𝐹𝐶𝑃 of
the packages are obtained after accounting for total number of the
respective package and logical arrangement among them in the HWR
cryomodule. It can be noticed from Table 4 that the combined avail-
ability of the cavity package in the HWR cryomodule was computed
to be 99.79% for the no-failure-permit mode and 99.99% for a cavity-
fail-tolerance mode. The combined availability of the solenoid magnet
assembly package was obtained to be 99.989%. Similarly, combined
availabilities of the rest of the packages in the HWR cryomodule were
computed. Fig. 11(a) shows the combined availability of all element
package in the HWR cryomodule. Note that, without a fail tolerance, Fig. 12. Availability of each section of the PIP-II facility for two variants of the nominal
the cavity package offers the least combined availability. Since all operational modes i.e. no-failure-permit mode (blue) and fail-tolerance mode (orange).
element packages are connected in a series configuration with the Note that the fail-tolerance was applied only to the SRF section. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version
cavity package (as shown in Fig. 10), availability of the full HWR
of this article.)
cryomodule is simply obtained from the product of their combined
availabilities as depicted in Fig. 11(b). Resulting availability of the
HWR cryomodule was obtained to be 99.69% for the no-failure-permit
sectional block (shown in Fig. 9) of the PIP-II facility model for both
mode that increases to 99.90% for a cavity-fail-tolerance mode.
nominal operational modes. In addition, it highlights the least available
To benchmark this calculation, availability assessment of the HWR
components/system-units in the respective sections. It is apparent from
cryomodule for the no-failure-permit mode was performed using a
Table 5 that the transfer line possesses the least availability among all
trial version of commercially available Monte-Carlo simulation pack-
sections. Note that, the transfer line is about two times longer than the
age BlockSim [32]. The results were in good agreements with our
SRF linac and, mainly composed of conventional normal conducting
estimation as shown in Appendix A.2.
quadrupole and dipole magnets. The power supplies of the magnets
exhibit a relatively higher failure rate with MTTF of ∼4E+04 h that
5.4. Availability assessment of the PIP-II facility for nominal operational
brings down the availability of the section.
mode
Fig. 12 shows a comparison of availabilities obtained from two
nominal operating modes. It is evident from Fig. 12 that availabilities
Availability of the full PIP-II facility for the nominal operational
of the SRF sections substantially improve in the fail-tolerance mode.
mode was modelled using the same methodology applied to the HWR
To make it more suggestive for practical purposes, the PIP-II sec-
section. Note that, as discussed earlier in Section 4.3.1, there are two
tional blocks are grouped into three major systems i.e. Utility, NC
variants of the nominal operational mode i.e. no-failure-permit and
and SRF linac. Table 6 lists the availability and MTBF of each major
fail-tolerance mode. In subsequent availability assessment, the fail-
system. The SRF linac exhibits the lowest availability of 95% for the
tolerance mode includes failure of an accelerating cavity in every SRF
no-failure-permit mode that increases to 99% after applying a cavity
section.
fail tolerance in every SRF section. Then, availability of the full PIP-II
The availability results show that an ion source offers the least
facility, computed from a product of the availability of every section,
availability of 89.08%. It is mainly because of the fact that the ion
was found to be 89.2% and 93.0% for the no-failure-permit and fail-
source requires the filament replacement for every three hundred hours
tolerance modes respectively. Again, all sections were connected in a
of operation. Consequently, it creates a bottleneck on availability of the
series configuration in the PIP-II model (Fig. 9). It should also be noted
complete facility. To improve the ion source availability and therefore,
that the facility exhibits a higher MTBF of 74.5 h in the fail-tolerance
for the complete facility, an additional ion source is installed in the
mode in comparison to 62.5 h of the no-failure-permit mode. The MTTR
standby configuration. In this arrangement, an ion source is always
of the PIP-II facility was computed using following equation:
available for operation while others get repaired. This in turn, improves
the ion source availability to 98.67%. Table 5 lists availability of each 𝑀𝑇 𝑇 𝑅 = 𝑀𝑇 𝐵𝐹 − 𝐴 ∗ 𝑀𝑇 𝐵𝐹 (7)

8
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Table 4
Availability of the cavity package and the solenoid magnet assembly package in the HWR cryomodule.
Component MTTF (T) 𝜆 (ℎ−1 ) MTTR 𝐴𝑖 𝑀𝑇 𝐵𝐹𝑐𝑝 𝐴𝑐𝑝
(h) (h) (%) (h) (%)

Cavity package
Cavity 8.76E+08 1.14E−09 776 99.999 Case 1: No-failure-permit mode
Tuner 1.00E+06 1.00E−06 216 99.978 8 cavity packages 8 cavity packages
Coupler 1.00E+07 1.00E−07 0.5a 99.999 in series in series
( )8
Interlock sensors 1.00E+05 1.00E−05 1 99.999 𝑀𝑇 𝐵𝐹𝐶𝑃 = 8𝜆1 𝐴𝐶𝑃 = Ap
𝑝
= 3623.19 𝐴𝐶𝑃 = 99.79

Interlock electronics 1.00E+05 1.00E−05 1 99.999 Case 2: A Cavity-Fail-Tolerance

Solid state amplifier 2.98E+05 3.36E−06 6 99.997 7 out of 8 cavity 7 out of 8 cavity
(SSA) packages packages
( )8
SSA Low Level RF 1.00E+05 1.00E−05 1 99.999 𝑀𝑇 𝐵𝐹𝐶𝑃 = 𝐴𝐶𝑃 = Ap +
∑7 ∏7 ( ) ( )7
1 1
𝜆𝑝 = 𝑖 𝜆𝑖 𝐴𝑝 = 𝑖=1 𝐴𝑖 + 71 8 × Ap ×
𝜆 p 8 ( )
𝜆𝑝 = 3.45E − 05 1 − Ap
𝐴𝑝 = 99.97 𝑀𝑇 𝐵𝐹𝐶𝑃 = 𝐴𝐶𝑃 = 99.999
763.98
Solenoid magnet assembly
Magnet power 1.00E+06 1.00E−06 2 99.992 8 Solenoid magnet 8 Solenoid magnet
supply assemblies are in assemblies are in
Magnet 1.00E+06 1.00E−06 792 99.999 series series
( )8
Magnet controls 1.00E+05 1.00E−05 2 99.998 𝑀𝑇 𝐵𝐹𝐶𝑃 = 8𝜆1 = 𝐴𝑐𝑝 = Ap
∑3 ∏3
𝑝

𝜆𝑝 = 𝑖 𝜆𝑖 Ap = 𝑖=1 𝐴𝑖 10416.67 𝐴𝑐𝑝 = 0.999

𝜆𝑝 = 1.2E − 05 Ap = 99.989

a It
is assumed that the coupler MTTR is the time needed to restore accelerator operation after detuning the cavity. Major coupler repairs are
accounted in the cavity MTTR.

Table 5
Availability of the functional blocks of the PIP-II facility for two nominal modes. The component with the least availability in respective section
is also listed.
Section Availability (%) Component with lowest availability in the section.
No-Failure- Fail-tolerance Component name Availability
Permit mode mode (%)
1 Electrical power system 98.79 98.79 Electric wire 99.22
2 LCW central system 99.88 99.88 Pressure gauge 99.91
3 Cryo-plant system 99.07 99.07 Warm compressors 99.82
4 Compressed air system 99.99 99.99 Compressor 99.99
5 Ion source 98.67 98.67 Individual ion source 89.08
6 LEBT 99.93 99.93 High voltage switch 99.95
7 RFQ 99.58 99.58 LCW—distribution (RFQ) 99.70
8 LCW—distribution 99.89 99.89 Circulating pump 99.91
9 MEBT 99.57 99.57 Magnet power supply chain 99.80
10 HWR 99.69 99.90 Solenoid magnet 99.91
11 SSR 1 99.40 99.90 Solenoid magnet 99.91
12 SSR 2 98.50 99.72 Solenoid magnet 99.78
13 LB 650 98.49 99.76 Quadrupole magnet package 99.85
14 HB 650 98.89 99.89 Quadrupole magnet package 99.87
15 Transfer line 98.27 98.27 LCW distribution (Transfer 99.09
line)

It results in the MTTR of 6.8 and 5.2 h for the no-failure and fail- Table 6
tolerance modes respectively. Availability and MTBF allocation by category for two operational modes of the
PIP-II linac facility.
The operational statistics of the existing accelerator facility corrob-
No-failure-permit mode Fail-tolerance-mode
orates that the target availability of 90% is well within reach of the
MTBF A MTBF A
modern technology. The Spallation Neutron Source (SNS) accelerator
(h) (%) (h) (%)
facility at Oak Ridge [33] has been reporting an availability of 90%
Utility system 1881.2 97.6 1881.2 97.6
since 2011 [34,35]. The proposed ESS facility also targets the facility
NC Linac system 127.8 96.1 127.8 96.1
availability of at least 90% over a calendar year [7]. This in turn, SRF Linac system 130.9 95.1 197.8 99.2
confirms feasibility of the PIP-II availability target. It is apparent from PIP-II facility 62.5 89.2 74.5 93
Table 6 that the PIP-II accelerator facility can deliver the target avail-
ability of 90% over a fiscal year in both operational modes. However,
the analysis also corroborates that an additional improvement in the
availability can be achieved through gaining a capability of operation

9
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Fig. 13. Distribution of the down-time hours by sections of the PIP-II facility operating in (left) no-failure-permit and (right) fail-tolerance modes. Note that, fail-tolerance of a
cavity per section was applied only to the SRF sections of the facility.

in a fail-tolerance mode. This is why the baseline design of the PIP-II Table 7
Availability of the PIP-II facility for two cases of the critical
linac [36] has adopted a cavity fault-tolerance in every SRF section. In
operational mode.
addition to a local energy correction, allocation of a spare cavities per
Section Total number Spare A(%)
section enables optics tuning in case of malfunctioned elements which of cavities cavities
is otherwise not possible if spare cavities are located at the end of linac.
LB650 36 11 93.35
At times it is more practical to describe the unavailability in term HB650 24 15 93.34
of the down-time that can be estimated using following equation:

𝐷𝑜𝑤𝑛𝑇 𝑖𝑚𝑒 = (1 − 𝐴) 𝑇𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 (8)

(as shown in Table 5) is the least available unit in both LB650 and
where, 𝑇𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 is the total operational mission time. Based on the
HB650 sections that determines overall availability of these sections.
operational mission time of 8000 h (excluding scheduled maintenance),
Consequently, additional fail-tolerances of the SRF cavities bring in
the down-time of each section of the PIP-II facility was estimated. As
only a little impact on the availabilities of these sections and therefore,
shown in Fig. 13, the BTL section imposes the maximum unscheduled
on the availability of the complete facility which is primarily governed
down-time of about 138 h to the PIP-II facility. This is mainly because
by the least available BTL and ion source sections.
of the fact that the BTL is the longest section of the facility with the
length of around 350 m. The second largest contributions of 120 h come
6. Sensitivity analysis and benchmarking of the model
from the LB650 and SSR2 SRF sections. The SRF sections enforces a
collective down-time of over 400 h. However, this time shrinks to 65
6.1. Input data sensitivity analysis
h in the fail-tolerance operating mode. The second largest contribution
of 106 h comes from the ion source in this mode of operation.
The quality of input data is the most crucial aspect of the availability
analysis that needs to be assured to obtain a meaningful outcome from
5.5. Availability assessment of PIP-II facility for critical operational mode the analysis. On the contrary, because of the first-of-a-kind nature of
every new accelerator, there are uncertainties involved with MTTF and
In the critical operational mode, the PIP-II facility operates to de- MTTR data of the components. In order to attain an adequate level of
liver the beam at 600 MeV to the Booster Synchrotron. The difference of confidence in outcome of the availability analysis, a sensitivity analysis
200 MeV from the nominal energy is modelled by turning off additional was performed to understand the impact of a spread in the input data
SRF cavities in the linac. These cavities are treated as the spare cavities. on the PIP-II availability.
Since the energy gain per cavity varies substantially along the linac It is evident that MTTF of a component is usually several order
(Fig. 2), there are several combinations to obtain the total number of higher than its MTTR. In some cases (such as for the cavities), MTTF
the spare cavities needed to downscale the beam energy from 800 MeV of a component could be higher from its life span. Consequently, MTTF
to 600 MeV. These combinations define states of the critical operational data may possess a relatively higher uncertainty in comparison of the
mode. MTTR data especially for the large values due to a lack of the failure-
It is well known that most of the beam dynamics issues in an ion rate statistics for such components. In order to analyse implication
linac are associated with its low energy portion. To incorporate this of uncertainty with the MTTF data in the PIP model, all components
fact in the availability analysis, it was assumed that there were no having MTTF above hundred years (∼ 106 h) were reduced by a scaling
additional spare cavities in the HWR, SSR1 and SSR2 sections. For factor and then the facility availability was evaluated. Fig. 14 shows
further simplification, it was considered that all the spare cavities were the availability of the PIP-II facility in the fail-tolerance mode as a
located only in one section. Thus, the critical operational mode was function of the MTTF scaling factor. It is apparent from here that the
modelled for two cases representing all the spare cavities were located facility reaches to its target availability even after reducing the MTTF
either in the LB650 or HB650 sections. Table 7 lists the number of by a factor of ten. However, the availability degraded below 80% after
the spare cavities in the respective sections. Note that, transit time MTTF were scaled down by a factor of 50. It can easily be concluded
effect [37] has been included while evaluating the total number of the from Fig. 15 that the SRF linac system availability is more sensitive to
spared cavities in the respective sections. Table 7 lists the availability of the fluctuations in MTTF data in comparison to the Utility and NC linac
the PIP-II facility for two cases of the critical operational mode. It can systems. Still, it attains the availability above 90% even after applying
be discerned that the facility possesses about the same availability of the scaling factor of twenty-five to the MTTF data. In another approach
93% in both cases as in the fail-tolerance nominal operational mode to compute the least plausible availability of the PIP-II facility, it was
even after applying additional fail-tolerances in terms of the spare assumed that the operational lifetime of the machine was thirty years.
cavities. It is attributed to the fact that the quadrupole magnet package Accordingly, all the MTTFs beyond thirty years were reduced to 2.6×105

10
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Table 8
Number of packages in MB and HB sections of the SNS
SRF linac.
MB HB
Cavity package 33 48
Magnet package 22 24
Steerer package 22 24
Cryo-package 11 12
Transmitter 4 8
Modulator 3 4

Table 9
Operational availability of the SRF linac system of the SNS accelerator
facility in two operational modes.
No-failure-permit Fail-tolerance-
mode availability mode
Fig. 14. Variation in availability of the total facility with the MTTF scaling factor in availability
a fail-tolerance operational mode. (%) (%)
MB section 97.2 99.4
HB section 96.1 99.3
SRF linac 93.4 98.8

between adjacent cryomodules to provide transverse beam focusing.

The SRF cavities are powered individually using klystrons. Input data
(MTTF and MTTR) for klystrons and associated RF components were
obtained from Ref. [10] while the PIP-II input data were applied
wherever they were applicable. Then, based on their functions, com-
ponents in an SRF cryomodule of the SNS linac were grouped in the
element packages. Appendix A.3 provides a detailed composition and
individual availabilities of each element package. Table 8 lists number
of respective packages in MB and HB sections.
The functional block diagram of the SRF linac system was prepared
considering the element packages were connected to each other in a
Fig. 15. Availability of each major system in the PIP-II model as a function of MTTF series configuration. The full SRF linac availability was then computed
scaling factor in a fail-tolerance operational mode. (For interpretation of the references for the no-failure-permit and fail-tolerance operating modes. Note that,
to colour in this figure legend, the reader is referred to the web version of this article.) the fail-tolerance operating mode assumes one spare cavity in each
section that might fail without interrupting the SRF linac operation.
It can be observed from Table 9 that the model predicted the SRF linac
h (equivalent to thirty years) in the model. In this scenario, the facility system availability of 93% and 99% in the respective no-failure-permit
availability was found to be 71%. Availabilities of the utility and NC and fail-tolerance operating modes.
linac systems were found to be 88% and 95% respectively while the The operational statistics of the SNS SRF linac has been presented
SRF linac system obtained the availability of 85%. in Ref. [34,35] which shows availability of the linac has been about
It can be concluded from the sensitivity analysis that the PIP-II 98% since the fiscal year of 2011. It indicates the measured availability
model could even tolerate a spread of 96% to the MTTF data above is in agreement with the model predicted availability especially for
∼ 106 h without substantial impact on its target availability of 90%. the fail-tolerance mode where both matches within a percentage level.
Also, among all three major systems, the SRF linac system availability An increase of five points with respect to the availability in the no-
is affected most from the choice of the MTTF input data. However, even failure-permit mode may attribute to a conservative set of the reliability
in a conservative estimate, the model predicted the PIP-II facility would input data. However, it has been addressed elsewhere [35] that the
have an up-time of 70% of its total operational time with the SRF linac design energy of 1 GeV has yet to be achieved for a nominal beam
availability in high eighties. operation at the SNS accelerator facility. This is mainly because of
collective effects (field-emission, multi-pacting, heating etc.) limiting
6.2. Model benchmarking the operating accelerating gradient in the SRF cavities. In this case,
one can conclude that the fail-tolerance operating mode is a more
In order to validate the methodology developed for the PIP-II representative choice to describe the SNS, SRF linac operation and
availability model, a reference model of the operational SRF linac of therefore, the model predicted availability in this mode is in good
SNS accelerator facility at Oak Ridge was developed using the same agreement with the measured availability of the SNS linac.
methodology. The SNS SRF linac is an ideal choice for the reference The PIP-II linac baseline incorporates SNS SRF linac operational
model due to a close resemblance of its configuration and operational experience in its design. The PIP-II SRF cavities excludes usage of
parameters to the PIP-II, LB650 and HB650 sections. The SRF linac de- Higher Order Modes (HOMs) damper identified as roots of several asso-
sign information was obtained from Ref. [33–35]. It has been designed ciated problems (field emission, heating, etc.) limiting RF performance
to accelerate the beam from 180 MeV to 1 GeV using two families of of the accelerating cavities in SNS linac. A detailed study presented
SRF cavities. Accordingly, the SRF linac has been segmented in two elsewhere [38] concludes usage of HOMs dampers in SRF cavities are
sections: Medium Beta (MB) and High Beta (HB) sections. The MB futile for the PIP-II SRF linac involving operation with a low average
section includes eleven cryomodules where each cryomodule houses beam current of 2 mA. In addition, uncorrelated HOMs spectrums of
three medium beta SRF cavities. The HB section consists of twelve five families of the PIP-II SRF cavities, HOMs frequency spread due
cryomodules and, each cryomodule is composed of four high beta SRF to manufacturing errors and a lower HOMs impedances because of
cavities. There are normal conducting quadrupole doublets positioned non-relativistic nature of the beam, largely preclude most of the beam

11
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

instabilities induced by HOMs. Thus, elimination of HOMs damper in Appendix A

PIP-II SRF cavities is a preferable choice that lowers not only overall
capital cost but also allows avoiding a number of issues in SRF cavities.
This in turn, could improve RF performance of the SRF cavities and
A.1.
hence the complete PIP-II facility .

7. Summary For the comprehension of this article, this appendix lists standard
Reliability Engineering textbook formulae. Several of those formulae
The paper introduced a methodology to model the availability of were applied in this article.
the complete particle accelerator facility. A comprehensive reliability For of 𝑖th component, if 𝑟𝑖 = Reliability at any time t, 𝑎𝑖 =
model of the proposed PIP-II accelerator facility was developed that Availability, 𝜆𝑖 = failure rate and 𝜇𝑖 = repair rate, then we can obtain
included not only the accelerator systems but also essential supporting following formulae.
systems such as the central cryo-plant, electrical power systems etc. The
availability assessment of the PIP-II facility reveals that the ion source 1. Series Configuration of the components in a system:
is most vulnerable system with availability of only 88%. Consequently, ∏
the baseline of the PIP-II facility adopted an additional ion source a. Availability of the system 𝐴 = 𝑎𝑖 .
∏
configured in the standby mode. This arrangement increases the ion b. Reliability 𝑅 = 𝑟𝑖 .
source availability to 98.7%. The baseline design of the PIP-II SRF 1
c. MTBF = ∑ 𝜆 ,
𝑖 𝑖
linac also attributes a cavity fault-tolerance in every SRF sections that
d. Mean Time to Failure = (1 − 𝐴) ∗ 𝑀𝑇 𝐵𝐹 .
enables the facility to operate in the fail-tolerance mode. Furthermore,
the PIP-II integration and operation strategy plans for a fully functional 2. Parallel Configuration of the components in a system
spare cryomodule always available for each SRF section in inventory to
minimize a repair time of the superconducting elements and therefore, ∏
a. 1 − 𝐴 = (1 − 𝑎𝑖 ).
unscheduled down time of the facility. ∏
b. 1 − 𝑅 = (1 − 𝑟𝑖 ).
The availability of the full PIP-II facility in nominal operational
c. MTBF = (1−𝐴)1∑ 𝜇 .
mode was found to be 89% that increased to 93% after introducing 𝑖 𝑖

the fail-tolerance of a cavity in every SRF sections in the model. d. Mean Time To failure = (1 − 𝐴) ∗ 𝑀𝑇 𝐵𝐹
This corroborates that the baseline design of the PIP-II accelerator
facility is sufficiently robust to meet the target availability in both 3. k out of n systems: Assume that all the components have same
nominal operational modes. Moreover, availability of the PIP-II facility failure rate (𝜆) and repair rate (𝜇).
was computed for the critical operational mode featuring the facility ( )
∑𝑘 𝑛
operation at the minimum beam energy of 600 MeV. The availability a. 𝐴 = 𝑖=0 𝑎𝑛−𝑖 (1 − 𝑎)𝑖 where, k is maximum
𝑖
of the PIP-II facility in this mode was obtained to be 93%. An input data number of failure allowed in a system, n is total number
sensitivity analysis and the model validation using a reference model of of components.
( )
the SNS SRF linac generate an adequate level of confidence in the PIP-
b. MTBF = 𝜆1 1𝑛 + 𝑛−1 1
+ ⋯ + 𝑘1 for non-repairable systems.
II availability assessment that leads us further to initiate engineering
design of the PIP-II facility.
4. Standby (Cold): A standby component implies that the compo-
CRediT authorship contribution statement nent starts operating as soon as another component gets failed.
Two components in a system have same failure rate (𝜆) and
Arun Saini: Conceptualization, Methodology, Writing - original repair rate (𝜇) and one of the components is kept as standby
draft, Revising, Investigation, Data Curation, Supervision, Formal anal- mode, then reliability and MTBF of the system is expressed as
ysis, Visualization, Writing - review & editing. Ram Prakash: Method- below
ology, Investigation, Formal analysis, Software. Joseph D. Kellen-
berger: Software. a. Reliability 𝑅 = (1 + 𝜆𝑡) 𝑒−𝜆𝑡
2𝜇 2
b. MTBF = 𝜆2 + 𝜆2 (𝜆+2𝜇)
Declaration of competing interest
In general, when two components have different failure rate
The authors declare that they have no known competing finan- 𝜆1 &𝜆2 and repair rate is 𝜇1 and 𝜇2 , MTBF is then express as
cial interests or personal relationships that could have appeared to below ( )
influence the work reported in this paper. 1 1 𝜇1 1 1
MTBF = 𝜆1
+ 𝜆2
+ 𝜆2 𝜆2
− 𝜆 .
𝜆2 +𝜇1 + 𝜆2 𝜇2
1
Acknowledgements
A.2.
The authors are thankful to the large team of scientists, engineers
and technical staffs who provided key input data for this study. The
authors would like to express gratitude on a more personal level to A. See Fig. A.1.
Klebaner, A. Martinez, J. Holzbauer, D. L. Newhart and, J. E. Anderson
Jr. for their constructive suggestions and discussions that helped the A.3.
authors to enhance quality of the paper. The author also wishes to
acknowledge efforts of L. Serio and C. Adolphsen who reviewed this See Table A.1.
work and provided their invaluable feedback. The authors are also
grateful to Barbara Merrill and Dr. Priyanka Saini and Dr. Vyacheslav
Yakovlev for their invaluable time to proof-read the manuscript and
useful suggestions. Appendix B. Supplementary data
This manuscript has been authored by Fermi Research Alliance, LLC
under Contract No. DE-AC02-07CH11359 with the U.S. Department of Supplementary material related to this article can be found online
Energy, Office of Science, Office of High Energy Physics. at https://doi.org/10.1016/j.nima.2020.164874.

12
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

Fig. A.1. Availability of HWR cryomodule computed for no-failure-permit mode using BlockSim (blue coloured bars) and analytical model (saffron coloured bars). (For interpretation
of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table A.1
A detailed view of the element packages in the SRF cryomodule of the SNS linac.
Packages Components MTTF MTTR A(%)
Magnet package Magnet 1E+06 16
Power Supply 4.6E+04 2
Magnet Instrumentation 1E+05 2
Magnet package availability 99.99
Cavity package SRF cavity 8.7E+08 776
Tuner 1E+06 216
Coupler 1E+07 0.5
Interlock sensor 1E+05 1
Klystron 5E+04 4.5
Wave Guide 1.5E+05 3
Circulator 5E+04 3
Load 7.5E+04 3
LLRF 1E+05 2
Cavity package availability 99.93
Steering magnet package Steerer
Power supply 1E+06 2
Magnet instrumentation 1E+06 2
Steerer instrumentation 1E+05 2
Steering magnet package availability 99.99
Cryo package Vacuum valves 1E+07 8
Ion pump 1E+06 4
Ion pump power supply 1E+05 1
Local cryogenic distribution 5E+05 2
Cryo-package availability 99.99
Additional components Transmitter 2.26E+04 4 99.98
Modulator 5.6E+03 3 99.94

References [10] P. Tallerico, D. Rees, D. Anderson, An availability model for the SNS Linac RF
system, in: Proceedings of PAC2001, Chicago, IL, USA, 2001, MPPH112, pp.
1035–1037.
[1] J.N. Galayda, The LCLS-II: A high-power upgrade to the LCLS, in: Proceedings
[11] E.S. Lessner, P.N. Ostroumov, Reliability and availability in the RIA driver linac,
of IPAC2018, Vancouver, Canada, MOYGB2, pp. 18–23.
in: Proceedings of PAC2005, Knoxville, TN, USA, 2005, FOAC005, pp. 443–445.
[2] S. Peggs, et al., ESS Technical Design Report, 2013, http://inspirehep.net/record/
[12] G.W. Dodson, Accelerator systems RAM analysis, Talk in Accelerator Reli-
1704813?ln=en.
ability Workshop, 2002, http://www.esrf.eu/files/live/sites/www/files/events/
[3] A. Sharma, A.R. Jana, C.B. Patidar, M.K. Pal, N. Kulkarni, P.K. Hoyal, et al.,
conferences/2002/ARW/proceedings/MONPM/Dodson.pdf.
Reference physics design for 1 GeV Injector Linac and accumulator ring for
[13] M.J. Haire, Computation of Normal Conducting and Superconducting Linear
Indian spallation neutron source, arXiv:1609.04518 [physics.acc-ph].
Accelerator (Linac) Availabilities, ORNL, USA, Tech. Report, ORNL/TM-2000/93,
[4] Zhihui Li, Peng Cheng, Huiping Geng, Zhen Guo, Yuan He, Cai Meng, Huafu 2000, https://www.osti.gov/biblio/885853-yUWMiH/.
Ouyang, Shilun Pei, Biao Sun, Jilei Sun, Jingyu Tang, Fang Yan, Yao Yang,
[14] PIP-II Conceptual Design Report, 2017, http://pip2-docdb.fnal.gov/cgi-bin/
Chuang Zhang, Zheng Yang, Phys. Rev. ST Accel. Beams 16 (2013) 080101.
ShowDocument?docid=113.
[5] T. Himel, J. Nelson, N. Phinney, Availability and reliability issues for ILC, in: [15] E.L. Hubbard, Booster Synchrotron Report, 1973, https://lss.fnal.gov/archive/
Proceedings of PAC07, Albuquerque, New Mexico, USA, pp. 1966–1969. tm/TM-0405.pdf.
[6] L. Burgazzi, P. Pierini, Reliability studies of a high-power proton accelerator for [16] L. Merminga, PIP-II Global Requirements Document, FNAL, USA, PIP-II
accelerator-driven system applications for nuclear waste transmutation, Reliab. Document 1166-v8, ED0001222, 2020, https://pip2-docdb.fnal.gov/cgi-
Eng. Syst. Saf. 92 (2007) 449–463, http://dx.doi.org/10.1016/j.ress.2005.12.008. bin/RetrieveFile?docid=1166&filename=ED0001222%20PIP-II%20Global%
[7] E. Bargalló, R. Andersson, A. Nordt, A. De Isusi, E. Pitcher, K.H. Andersen, ESS 20Requirements%20Document%20GRD.pdf&version=8.
availability and reliability approach, in: Proceedings of IPAC2015, Richmond, [17] A. Shemyakin, M. Alvarez, R. Andrews, J.-P. Carneiro, A. Chen, R. D’Arcy,
VA, USA (2015), MOPTY045, pp. 1033–1035. B. Hanna, L. Prost, V. Scarpine, C. Wiesner, PIP-II injector test’s low energy
[8] J. Knaster, P. Garin, H. Matsumoto, Y. Okumura, M. Sugimoto, F. Arbeiter, P. beam transport: Commissioning and selected measurements, AIP Conf. Proc. 1869
Cara, S. Chel, A. Facco, P. Favuzza, T. Furukawa, R. Heidinger, A. Ibarra, T. (050003) (2017).
Kanemura, A. Kasugai, H. Kondo, V. Massaut, J. Molla, G. Micciche, S. O’hira, K. [18] S. Virostek, et al., Final design of a CW Radio Frequency Quadrupole (RFQ) for
Sakamoto, T. Yokomine, E. Wakai, Overview of the IFMIF/EVEDA project, Nucl. the Project X Injector Experiment (PXIE), in: Proc. NAPAC’13, Pasadena, CA,
Fusion 57 (10) (2017) 102016, http://dx.doi.org/10.1088/1741-4326/aa6a6a. USA, 2013, WEPMA21, pp. 1025–1027.
[9] R. Andersson, A. Nordt, E. Bargalló, Machine protection systems and their impact [19] A. Saini, C.M. Baffes, A.Z. Chen, V.A. Lebedev, L. Prost, A. Shemyakin, Design of
on beam availability and accelerator reliability, in: Proceedings of IPAC2015, PIP-II medium energy beam transport beam, in: Proc. of IPAC 2018, Vancouver,
Richmond, VA, USA, 2015, MOPTY044, pp. 1029–1032. Canada, 2018, TUPAF076, pp. 905–908.

13
A. Saini, R. Prakash and J.D. Kellenberger Nuclear Inst. and Methods in Physics Research, A 988 (2021) 164874

[20] Z.A. Conway, et al., IOP Conf. Ser.: Mater. Sci. Eng. 101 (2015) 012019. [31] A. Vivoli, J. Hunt, D.E. Johnson, V. Lebedev, Transfer Line Design for PIP-II
[21] M.H. Awida, et al., Development of low single-spoke resonators for the front Project, in: Proceedings of IPAC2015, Richmond, VA, USA, 2015, THPF119, pp.
end of the proton improvement plan-II at Fermilab, IEEE Trans. Nucl. Sci. 64 3989–3991.
(9) (2017) 2450–2464. [32] https://www.reliasoft.com/products/blocksim-system-reliability-availability-
[22] V. Roger, et al., Design Update of the SSR1 Cryomodule for PIP-II Project, maintainability-ram-analysis-software.
in: Proceedings of IPAC2018, Vancouver, Canada, 2018, WEPML019, pp. [33] S. Henderson, et al., The Spallation Neutron Source Beam Commissioning and
2721–2723. Initial Operations, ORNL, USA, Tech. Report, ORNL/TM-2015/321, 2015, https:
[23] A. Rowe, SRF Technology for PIP-II and PIP-III, in: Proc. SRF2017, Lanzhau, //info.ornl.gov/sites/publications/files/Pub56465.pdf.
China, 2017. [34] S.-H. Kim, R. Afanador, W. Blokland, M. Champion, A. Coleman, M. Crofford,
[24] A. Saini, Design considerations for the Fermilab PIP-II 800 MeV Superconducting et al., The status of the superconducting linac and SRF activities at the SNS,
Linac, in: Proc. of NA-PAC 2016, Chicago, USA, 2016, WEPOA60. in: Proceedings of the 16th International Conference on RF superconductivity,
[25] D.J. Smith, Reliability, Maintainability and Risk, Elsevier Ltd, 2011, http://dx. Paris, France, September 23–27, 2013, pp. 83–88, http://accelconf.web.cern.ch/
doi.org/10.1016/C2010-0-66333-4. AccelConf/SRF2013/papers/mop007.PDF.
[26] M. Raus, A. Hsyland, System Reliability Theory Models, Statistical Methods, and [35] S.H. Kim, et al., Overview of ten-year operation of the superconducting linear
Applications, second ed., John Wiley & Sons, Inc. accelerator at the Spallation Neutron Source, Nucl. Instrum. Methods Phys. Res.
[27] J. Upadhyay, D. Im, J. Peshl, M. Bašović, S. Popović, A.M. Valente-Feliciano, et A (ISSN: 0168-9002) 852 (2017) 20–32, http://dx.doi.org/10.1016/j.nima.2017.
al., Apparatus and method for plasma processing of SRF cavities, Nucl. Instrum. 02.009.
Methods Phys. Res. A 818 (2016) 76–81, http://dx.doi.org/10.1016/j.nima.2016. [36] M. Convery, et al., The PIP-II Preliminary Design Report, PIP-II Doc-
02.049. ument 2261-v33, 2020, https://pip2-docdb.fnal.gov/cgi-bin/ShowDocument?
[28] A. Saini, J.-F. Ostiguy, N. Solyak, V.P. Yakovlev, Studies of fault scenarios in SC docid=2261.
CW Project-X linac, in: Proceedings of NA-PAC2013, Pasadena, California, USA, [37] T.P. Wangler, RF Linear Accelerator, second ed., Wiley-VCH Verlag GmbH & Co.,
2013, MOPMA10, pp. 318-320. 2008.
[29] A. Saini, N. Solyak, V.P. Yakovlev, S. Mishra, K. Ranjan, Study of effects of failure [38] A. Sukhanov, A. Lunin, V. Yakovlev, M. Awida, M. Champion, C. Ginsburg, I.
of beamline elements and its compensation in CW superconducting linac, in: Gonin, C. Grimm, T. Khabiboulline, T. Nicol, Yu. Orlov, A. Saini, D. Sergatskiv,
Proceedings of IPAC2012, New Orleans, Louisiana, USA: 2012, pp. 1173–1175. N. Solyak, A. Vostrikov, Higher order modes in project-X Linac, Nucl. Instrum.
[30] P.F. Derwent, J.-P. Carniero, J. Edelen, V. Lebedev, L. Prost, A. Saini, A. Methods Phys. Res. A 734 (part A) (2014) 9–22, http://dx.doi.org/10.1016/j.
Shemyakin, J. Steimel, PIP-II Injector Test: challeng-es and status, in: Proc. of nima.2013.06.113.
LINAC’16, East Lansing, MI, USA, September 25–30, 2016, WE1A01.

14