Chapter 3

Measurement System Behavior

Part 1

Material from Figliola and Beasley, 2000, 3ed.

Measurement System Behavior

Topics:
„ General Systems Model
„ Transfer Functions
„ Phase Linearity
„ Multiple Function Inputs
„ Coupled Systems

Definitions:
1. Each component of a measurement
system has its own response to a
given input.
2. The overall system response is
affected by the response of each
stage in the system.
3. To reconstruct an input signal, we
need to know the amplitude,
frequency, and/or general shape of
the signal.

1
Definitions:
4. Behavior of MS can be defined by a
few special inputs which will give us
the input/output relationship. This
allows us to calibrate the system with
a restricted number of tests.
5. MS response to input must be fast
enough to pick up dynamically
changing conditions.

Dynamic Measurements
The behavior of measurement systems to a few
special inputs will define the input-output signal
relationships necessary to enable the correct
interpretation of the measured signal.
Only a few measurement system characteristics
are needed to define the system response, so
that dynamic calibrations can be restricted to a
few specific tests.
The results of these tests can be used to judge
the suitability of a particular system for a
measurement.

MS as a Black
Box that operates
on the input to
give us an
output. We use
y(t) to infer F(t).

1. The input is represented by some function

that acts on the MS
2. A mathematical model is used to represent
MS
3. We observe MS response and do an
analytical calibration.

2
General Model of MS:
nth order linear differential equations with
forcing function
n−1
a n ( d y / dt ) + a n − 1( d y / dt ) + ...
n n n−1

...+ a 1( dy / dt ) + a 0 y = F ( t )

• Coefficients a0, a1,…, an are unique to MS

General Model of MS:

The forcing function can be generalized
into the mth-order form:
F (t )= bm (d m x /dt m )+ bm−1 (d m −1 x /dt m −1) + ...+ b0
m≤ n

• Coefficients b0, b1,…, bm represent

physical system parameters

General Model of MS:

Most practical systems can be modeled
by zero, first, or second order linear
differential equations.
Modeling is used to help select the
specific instrument and to help design
the calibration test.
Exact I/O relationship is determined by
calibration.

3
Example
In Figure 3.3, a seismic accelerometer
is shown.
Various configurations of this
instrument are used in seismic and
vibration engineering to determine the
motion of large bodies to which the
accelerometer is attached.

Example
As the small accelerometer mass reacts
to motion, it places the piezoelectric
crystal into compression or tension,
which causes a surface charge to
develop on the crystal.
As the large body moves, the mass of
the accelerometer will move with an
inertial response.

4
Example
The stiffness of the spring, k, will
provide a restoring force to move the
accelerometer mass back to equilibrium
while internal frictional damping, c, will
oppose any displacement away from
equilibrium.

Example
Solving Newton’s second law for the free
body yields the second-order linear, ordinary
differential equation:

⎛d y⎞ ⎛ dy ⎞ ⎛ dx ⎞
2
m⎜ 2 ⎟ + c⎜ ⎟ + ky = c⎜ ⎟ + kx
⎝ dt ⎠ ⎝ dt ⎠ ⎝ dt ⎠
The forces developed as a result of the
velocity and displacement of the body
become the inputs to the accelerometer.

Zero-Order System
Simplified Model:
„ a0y=F(t) where F(t)=b0X is a forcing function
„ y=KF(t) where K=1/a0 which is static sensitivity
In zero-order, the MS responds to the
input immediately/instantly. It is a system
of choice for static inputs or static
calibration.
Valid for dynamic signals only at
equilibrium.
“K” is determined by calibration; it is the
slope of the calibration curve.

5
Pencil Type Pressure Gauge

ΣF=0 Ky=F – Fatm

y=(A/K)(P – Patm)
(P – Patm) represents pressure relative to
atmosphere
Κ=A/K = static sensitivity

Pencil Type Pressure Gauge:

The spring force resists the pressure on

the piston. Deflect it until the spring
force is in equilibrium with the pressure.

The displacement “y” is a

representation of pressure and is
presented on the scale of the piston
rod.

First Order
Measurement systems that have a
storage element cannot respond
instantly to a change in input. The
temperature of a bulb thermometer
changes with time until equilibrium is
reached. This causes the time delay. It
is modeled by first order diff equations.

6
First Order:
A system with storage or dissipative capacity,
but negligible inertial forces can be modeled
as 1st order.
a1 y + a 0 y = F ( t ) y = ay / dt
o o

Dividing by a0

τ y + y = KF (t )
o
τ = a1 / a 0
τ = time constant, provides a measure of the
speed of system response and is an
important specification in measuring dynamic
input signals.

First Order:
• Response to Step Input
Step AU(t)=0 t≤0- A=amplitude
AU(t)=A t≥0+ u(t) = unit step

τ y + y = KAU (t ) = KF ( t )
o
letting y(0)=y0

• Physically, this function describes a sudden change in

the input signal from a constant value of one
magnitude to a constant value of some other
magnitude, such as a sudden change in temperature,
pressure, or loading.

Solution for t ≥ 0+

y(t) = KA + (y0 - KA) e-t/τ

(time response) (steady response) (transient response)

„ y(t) is the output indicated by the display

stage and represents the time variation.
It represents the time variation of the
output display of the measurement
system if an actual step change were
applied to the system.

7
 Solution for t ≥ 0+
• y(t) = KA + (y0 - KA) e-t/τ
• The second term is known as the transient
response of y(t) since, as t goes to infinity, the
magnitude of this term eventually reduces to
zero.
The first term is known as the steady
response since, as t goes to infinity, the
response of y(t) approaches this steady
value.
The steady response is that portion of the
output signal that remains after the
transient response has decayed to zero.

Solution for t ≥ 0+
Transient response: goes to zero as tÆ∞
Steady response: what remains after
transientÆ0

Fig 3.5- step Fig 3.6- 1st order response to step

Error Fraction
Rewrite response
Γ (t ) = ( y(t ) − y∞ ) / ( y0 − y∞ ) = e − t / τ
Γ (t ) → 0 as t / τ = ∞
Γ (t ) = 1 at t / τ = 0
at t=τ Γ=.368 and response is
63.2%
at t=2.3τ Γ=.10 and response is
90%
at t=5τ Γ=.007 and response is
99.3%
τ=2.3τ is called rise time and is
often specified in literature
(90% response)

8
Determination of τ
1. Time constant determined
by exciting the system with
a step input and recording
the response.
2. The Approach:
1. Recording from t=0 to
steady state
2. Plot as error fraction versus
time on semi-log
3. Plot is equivalent to
lnΓ=2.3log Γ=-(1/τ)t
4. Fit curve through data
5. Slope m=-1/τ

Method’s Advantages:
System varies from true 1st order

Variations show up on plot

Curve-fitting tends to eliminate the influence

of any one bad data point

Do not need to accurately determine Γ=0.368

and Γ=1.0 points