By D. B. Gain: The RTL SDR V3
By D. B. Gain: The RTL SDR V3
Radio Hardware;
RTL-SDR Blog on the internet recently touted a new RTL dongle, one with improved performance
over run of the mill RTL dongles. The fine folks at RTL SDR Blog were kind enough to send me a
sample RTL SDR V3 dongle and antenna kit, this is the dongle used throughout the review.
If you didn't know, the RTL USB dongles are originally manufactured for VHF/UHF digital TV
broadcast reception via a pc and were found to be hackable for general radio pursuits such as
scanning, trunked radio system tracking, APCO25 and other digital modulation scheme decoding, as
well as other radio endeavours in the 25 to 2000MHz range. You plug the dongle into a handy USB
port, fire up the control app (after installing a driver and/or dll for the dongle perhaps), and off you
go into the vast new world of SDR radio. A world rich in Software Defined Radio and all it entails.
Normally one wants more than an 8bit ADC to get sufficient gain and dynamic range on HF (or
anywhere else), somewhere along the lines of 14bits and up will do very well, 24bits would get you a
theoretical 144dB dynamic range (dynamic range is the ability to copy weak signals undistorted in the
presence of much stronger nearby and not so nearby signals, very similar to being able to hear a
whisper spoken to you while standing next to a jet engine at full afterburner), but a inexpensive 24bit
chip that can handle HF has yet to enter the hobbyist market, so the 8bit chips are the price leader.
These higher bit ADC chips exist in the marketplace, but typically in the form of military and
government surveillance receivers that come with high prices. Oh by the way, there's a 24bit or
better ADC sitting in the CD/DVD player of your PC.
Some of the consumer grade SDR radios aimed at the HF market are of the 12 to 16bit range and
these do nicely in the crazy HF environment, especially when one gets to 14bits and above.
The jump from 8bits to 12bits is on the order of 6 times the price of the 8bit V3, but the jump from
12 to 14 bits is around three times the price or more of the 12bit SDRs. The way I look at it, the jump
from 8 to 12bits gets you about 15 to 20dB dynamic range increase, which is very noticeable and
useful and is a game changer compared with 8bits. The performance difference between a 12bit ADC
and a 14bit ADC is closer than the performance difference between an 8bit ADC and a 12bit ADC. I
guess what I'm really waiting for is the time when future 32bit chips are as cheap as the 8bit ones of
today, however long that may take.
The physical properties of said dongle are about a thumb's length and half thickness of a thumb,
with a USB connector on one end and an antenna port on the other.
The dongle comes with a mag mount antenna base with attached coax and SMA connectors and
two telescopic rod antennas, one of impressive stature. The V3 can also be purchased by itself for a
bit less money.
The new RTL dongle sports an shiny aluminium case to keep signals in or out of the SDR as the case
may be, said case also acting as a nice heat sink, and has a measure of noise and interference
reduction in the circuit board itself compared to previous RTL dongles, especially with regard to the
USB line and switching mode power supply, very welcome advances in the dongle art. Typically these
dongle have a plastic case and are thusly prone to allowing interfering signals from the pc and other
in home devices as well as the local RF environment to spoil the signals before they enter the pc
through the dongle and SDR app. Another advance is the presence of a TCXO, or Temperature
Controlled Crystal Oscillator.
Co-channel FAX on S4285 Signal
The TCXO reduces the oscillator drift that arises normally from internal component temperature
variations due to operation and environmental factors by keeping the crystal that the reference
oscillator is derived from at a more constant temperature, negating drift after a few minutes of
warmup. I can vouch for the RTL V3 TCXO's abilities to keep you on frequency with little or no drift,
at least on HF. The dongle also sports a SMA antenna connector, one will have to obtain an SMA to
whatever coaxial connector one uses. I obtained a SMA to UHF adapter, lightweight so as to keep
stress off the antenna connector in the RTL V3.
PC Hardware;
I used a PC with a 4 core Intel CPU, Intel mobo, 4gb ram, and 32bit Win7 OS. The onboard HIDEF
sound system handles the sound routing via the stereo mixer. The particular sound system has a max
sampling rate of 48KHz but that works out just fine for HF decoding apps and general HF listening.
As the max audio bandwidth I use for HF reception (and feed to decoder apps) is 10KHz or less,
48KHz is more than enough to satisfy the Nyquist criterion. Harry Nyquist said you have to sample a
given signal of interest at least twice its highest frequency to reproduce the sample digitaly without
loss or distortion. In regards to sampling analog things into the digital world we should listen to Harry
Nyquist, he knows what he's talking about. A soundcard typically takes analog signals presented to it,
runs them into an ADC (analog to digital converter chip), manipulates them digitally and then sends
them off to speakers or an application, via a DAC which is the reverse of a ADC, a digital to analog
converter.
The Barker and Williamson 30MHz lowpass filter also acts as a reactance moderator, somewhat
reducing the wildly swinging SWR as compared to when it is absent from the antenna system. It also
cuts off signals above 30MHz, w ich protects sensitive RTL dongle inners from unwanted radio
energies. Energies such as literal megawatts of FM and TV broadcast stations, pager systems, cellular
systems, and business/government communications that populate the V/UHF bands. These signals
would all impinge upon the hapless RTL dongle, greatly hindering HF performance if not for some
lowpass filtering.
The masses of AMBC stations in the 530 to 1700KHz Medium Wave range also can perturb a RTL
dongle if too much MW energy is placed at the input to the dongle. Here a highpass filter or
attenuator is in order, or a preselector.
I find that the use of a preselector or transmatch can reduce images and imd, but even at their best
the dongle still has issues with faint images and spurs, which is the nature of an 8 bit device. You
learn to live with it and tune around the worst if you can.
The antenna feed also sports a Polyphaser emp rated coaxial transient arrester. This two port
device, grounded via its casing, presents the radio side of the transient suppressor with capacitive
coupling of the coax center conductor to the rest of the antenna system, the antenna side sees an
high inductance to ground at the transient suppressor, which bleeds off static and yet passes rf. It
also has a gas discharge arrestor and is overall designed to handle emp events. This device alone cost
several times what the RTL SDR V3 dongle cost, much of the antenna system and PC hardware also
cost far more than the V3 dongle.
The TUN3 is no longer manufactured but can be found on the used market. A rather small black box
with two tuning knobs, one for band, the other for fine tuning of the resonating/matching capacitor,
it tunes from VLF to 30MHz and slightly above. It's intended for reducing images in single conversion
general coverage and AMBC receivers and does a pretty fair job of that in my estimation. It peaks a
desired signal nicely even on HF but quite spectacularly on AMBC and below while also reducing the
out of band rf. You can find schematics online for building your own preselector, from simple to
elaborate as you desire.
The MFJ Transmatch used in this review is also one no longer made, it's a simple and compact
aluminum box with three controls; one for the input side capacitor, the next for the inductance
which selects the band of use, and the last for the output variable cap. The back of the tuner has
input and output UHF sockets. As it's a T circuit it's highpass in nature, meaning it cuts off RF energies
below the frequency it's tuned to. That's a side benefit as its design goal was to match an high
reactance antenna system to the 50ohm antenna port found on HAM rigs. The T circuit offers about
the most tuning range as far as input reactance or Z from the antenna system. It's not a preselector
but it can reduce the RF energies present in the antenna system below the frequency it's tuned to,
and that is a benefit to many radios, not just the RTL SDR. You can find schematics online for building
your own Transmatch, from simple to elaborate as you desire.
Anyway, I notice SDR#, SDRuno, SDRConsole and CubicSDR have low Memory footprints in some
cases but are thread heavy (as in 20 plus threads), are CPU hogs, and not to my liking as far as
glitchyness or stability on my system in the case of SDR# and CubicSDR, but SDRuno was rock stable.
HDSDR wins here, even though it may take up twice as much ram as the other apps doing the same
job, HDSDR has never crashed or displayed glitchyness in use at my station. For example of a CPU
hog, SDR# used around 9 percent CPU time doing the same thing HDSDR did at 0 percent. If you have
high performance CPUs or multiple CPUs what's the deal about not liking the SDR app to use CPU
time?
Well, my answer to that is I consider the SDR app to be the OS of the SDR, and as such want as little
CPU and Mem footprint as possible so the system is devoted to catering to the needs of apps used in
conjunction with the SDR control app such as Sigmira, Sorcerer, MultiPSK and others. Don't get me
wrong, I like a fancy waterfall and spectrum display too but when you can get similar information
from HDSDR at a fraction of the CPU time used by many other SDR control apps, HDSDR seems a no
brainer. By the way, Sigmira currently doesn't do the direct sample mode needed to place the V3 on
HF without an upconverter.
I'm likewise still testing SDRConsole, it's a interesting and very stable SDR app but so far I can't find
out how to direct the app to use the Q branch of the V3 dongle that enables use on HF.
Secondly, use no more USB sample rate higher than results in least IMD and Images, try .96Msps to
1.44Msps. This will reduce USB bus traffic as well as CPU time and heat production.
Thirdly, the larger the antenna system, the greater the gain - usually. It doesn't take too much RF to
overload the V3 dongle, so a 20ft piece of wire will do better than say a 430ft wire loop atop some
phone poles. Use an attenuator if you have one. Remember the issue with AMBC swamping where
AM stations pop up in various parts of the HF spectrum and use a preselector and/or attenuator if
you can. Shortwave broadcast stations can also create spurs in the V3.
Some radio parts houses carry a variable attenuator meant for cable TV or VCR player use that can
be employed at HF with the use of some F to UHF or whatever connector your antenna system
employs adaptor, this can be installed in the antenna system and adjusted to result in least usable
signal getting to the V3, which assures best dynamic range. Then one would adjust the FFT Spectrum
gain in your SDR control app of choice to best level on a quiet band, say 14MHz. This will ensure you
don't have to mess with adjusting the gain on lower frequency bands just to keep the band noise
baseline above the bottom of the FFT window
Fourthly, try different USB ports on your PC with the dongle so as to find the one with least
interference. Some USB ports are quieter in regard to system noises than others. These will typically
be seen on the waterfall display and are present without an antenna being connected to the dongle.
And lastly, learn to deal with imd and spurs as they will be common when using the RTL SDR V3 on
HF, it's just the nature of the beast. Not the V3's fault alone, pretty much every 8bit ADC is going to
react the same way to a diet of HF signals. However, even with the imd and spurs, I found plenty of
signals in the clear that were worth listening to.
Adjust the RBW to a coarser setting on the spectrum display for the RF Spectrum of the app to
reduce CPU and Video Card usage. Try a setting one step finer than makes stair stepping in the
spectrum display apparent. This should give you just the right amount of detail on signals with the
least in CPU and Video Card time. Do the same for the Audio spectrum display, but this being a much
smaller area on the screen can be set to its highest resolution without much if any impact upon CPU
and Video Card use due to the fraction of spectrum being processed, i.e. 5KHz in the case of a Audio
spectrum display versus perhaps a MHz or more for the RF spectrum display.
On the I/Q or Audio input settings to the app, if present, use sensible rates, don't force rates higher
than your sound card allows in the case of sound card input, and remember that a lot of decoder
apps take the 192KHz, 96KHz, or even 48KHz sound card sample and downconvert it to something on
the order of 11KHz or even less to perform processing. For example, in the case of HDSDR an output
of 48KHz was used so that it matched the rate I selected in the sound card property sheet so no
conversion had to be made between the rates of sound card and SDR api, the input is the USB
sample rate.
Buffer Size can also have a dramatic effect on CPU use, even as little as one step above default
buffering can bring the CPU usage up noticeably, so test this control wisely. If the app runs well at
default buffer size, leave it or even reduce it till performance is degraded, then increase it one step to
restore performance at the least in buffer size.
If an image or spur is on your frequency of interest, try adjusting the USB sample rate one way or
the other to move the spur off your frequency, this often helps.
If the SDR app has Noise Blanker options for Baseband and IF, use them. Adjust them for least
distortion of signals, they will make for better audio out. Imagine HF without the pops and static!
A DSP Noise Reduction option may be present, these can reduce Decoder app effectiveness when
enabled so keep that in mind when using Decoder apps.
RTL SDR V3 Pros; a tiny SDR package that tunes MF to 2000MHz or so and has a TCXO, can't be beat
for the price, and free software to run the thing is all over the net.
RTL SDR V3 Cons; Has no suboctave bandpass input filters thusly resulting in images and imd and
spurs due to the dynamic range of 8bit ADCs, as well as internal mixing of unwanted and wanted
signals from phase noise of the reference oscillator. Many of the apps used to control the RTL SDR
dongles need additional drivers, dlls, and etc to function, some of which is a pain to deal with but can
be overcome for the most part following the many guides and howto’s that can be found online. If
only it had a few more bits in the ADC. I'm wanting to try a 12bit machine now, they're a few times
more spendy than the 8bit SDR rigs but I've heard they do better at HF.
Resources;
Get you a RTL SDR
http://www.rtl-SDR.com/new-rtl-SDR-blog-units-now-available-in-store-hf-via-direct-sampling-
software-switchable-bias-tee-less-noisespurs/
More preselectors
http://www.bobsamerica.com/swl.html
HDSDR
http://www.hdSDR.de/
SDR#
http://airspy.com/download/
SDRuno
http://www.SDRplay.com/windl2.php
SDRConsole
http://SDR-radio.com/
MultiPSK decoder
http://f6cte.free.fr/index_anglais.htm