Evaluation of SDR Boards-1.1

Evaluation of SDR Boards and Toolchains
Final Report
Authors: Alexandru Csete

Sheila Christiansen
Date: 2020-08-29
Version: 1.1
Revision History
Date Version Sections Changes
2020-03-20 1.0 All First version for LSF review.
2020-08-29 1.1 2.2.4 Updated with new LimeSDR Mini measurements and
2.2.8.3 info.
Table of Contents
1 Introduction............................................................................................................1
1.1 Purpose and Scope.........................................................................................................1
1.2 Selected SDR devices....................................................................................................1
1.3 Evaluation Criteria.........................................................................................................2
1.4 Evaluation Methods.......................................................................................................3
2 Evaluation of Technical Specifications.................................................................4
2.1 Test Setup and Procedures.............................................................................................4
2.1.1 Receiver Noise Figure...........................................................................................4
2.1.2 Receiver Dynamic Range......................................................................................7
2.1.3 Receiver Spectral Purity......................................................................................10
2.1.4 Transmitter Spectral Purity..................................................................................10
2.1.5 Transmitter Output Power....................................................................................11
2.1.6 Transmitter Modulation Error Ratio....................................................................11
2.1.7 Test Software.......................................................................................................13
2.2 Test Results..................................................................................................................16
2.2.1 RTL-SDR Blog V3..............................................................................................16
2.2.2 Airspy Mini..........................................................................................................26
2.2.3 SDRplay RSPduo.................................................................................................32
2.2.4 LimeSDR Mini....................................................................................................40
2.2.5 BladeRF 2.0 Micro..............................................................................................58
2.2.6 Ettus USRP B210.................................................................................................77
2.2.7 PlutoSDR.............................................................................................................96
2.2.8 Device Comparisons..........................................................................................116
2.3 Test Setup Verification...............................................................................................120
2.3.1 Signal Generator Output Power Comparison.....................................................120
2.3.2 Test Software Verification..................................................................................123
3 On-Air Tests........................................................................................................124
3.1 Test Setup...................................................................................................................124
3.2 Test Results................................................................................................................126
4 Software Support................................................................................................128
4.1 SDR-Radio V3...........................................................................................................129
4.2 SDR#.........................................................................................................................130
4.3 Gqrx...........................................................................................................................131
4.4 CubicSDR..................................................................................................................132
4.5 SDR Angel.................................................................................................................133
4.6 GNU Radio................................................................................................................134
4.7 Pothos SDR................................................................................................................135
4.8 SoapySDR..................................................................................................................136
4.9 Gr-soapy....................................................................................................................136
4.10 Gr-osmosdr..............................................................................................................137
5 Conclusion...........................................................................................................138
6 Appendices...............................................................................................................I
iii
List of Figures
Figure 2.1: Test setup for noise figure measurements using the MDS method........................6
Figure 2.2: Test setup for noise figure measurements using the Y-factor method....................6
Figure 2.3: Disturbing signal causing increase in the noise level at 100 kHz away.................8
Figure 2.4: Disturbing signal causing spurs as far as 1 MHz away..........................................8
Figure 2.5: Test setup for dynamic range measurements.........................................................9
Figure 2.6: Test setup for receiver spectral purity measurements..........................................10
Figure 2.7: Test setup for transmitter power and spectral purity measurements....................11
Figure 2.8: Constellation with error [8].................................................................................12
Figure 2.9: Test setup for transmitter MER measurements....................................................12
Figure 2.10: Screen captures of the ROVER DVB-S/S2 analyzer.........................................13
Figure 2.11: Functional diagram for the test software used for receiver tests........................13
Figure 2.12: GNU Radio application used for output power measurements..........................14
Figure 2.13: GNU Radio application used for MER measurements......................................14
Figure 2.14: The SATSAGEN spectrum analyzer software for PlutoSDR............................15
Figure 2.15: The RTL-SDR blog V3 receiver........................................................................16
Figure 2.16: RTL-SDR Blog V3 dongle block diagram.........................................................17
Figure 2.17: RTL-SDR V3 noise figure measurements.........................................................18
Figure 2.18: RTL-SDR V3 blocking dynamic range measurements......................................20
Figure 2.19: RTL-SDR receiver spectrum between 145 – 146 MHz.....................................21
Figure 2.22: RTL-SDR receiver spectrum between 1260 – 1262 MHz.................................23
Figure 2.27: The Airspy Mini receiver...................................................................................26
Figure 2.28: Airspy Mini block diagram................................................................................26
Figure 2.29: Airspy Mini noise figure measurements............................................................28
Figure 2.30: Airspy Mini blocking dynamic range measurements........................................29
Figure 2.31: Airspy Mini receiver spectrum between 145 – 146 MHz..................................30
Figure 2.32: Airspy Mini receiver spectrum between 435 – 438 MHz..................................30
Figure 2.33: Airspy Mini receiver spectrum between 1260 – 1264 MHz..............................30
Figure 2.36: The SDRplay RSPduo receiver..........................................................................32
Figure 2.37: The SDRplay RSPduo block diagram...............................................................33
Figure 2.38: SDRplay RSPduo noise figure measurements...................................................35
iv
Figure 2.39: SDRplay RSPduo blocking dynamic range measurements...............................36
Figure 2.40: SDRplay RSPduo receiver spectrum between 145 – 146 MHz.........................37
Figure 2.41: SDRplay RSPduo receiver spectrum between 435 – 438 MHz.........................38
Figure 2.42: SDRplay RSPduo receiver spectrum between 1260 – 1265 MHz.....................38
Figure 2.43: SDRplay RSPduo receiver spectrum between 1265 – 1270 MHz.....................39
Figure 2.44: LimeSDR Mini transceiver................................................................................40
Figure 2.45: LimeSDR Mini block diagram..........................................................................40
Figure 2.46: LimeSDR Mini noise figure measurements......................................................42
Figure 2.47: LimeSDR Mini noise figure measured with and without the aluminum case.. .43
Figure 2.48: LimeSDR Mini blocking dynamic range measurements...................................45
Figure 2.49: LimeSDR Mini receiver spectrum between 145 – 146 MHz.............................47
Figure 2.50: LimeSDR Mini receiver spectrum between 435 – 438 MHz.............................47
Figure 2.51: LimeSDR Mini receiver spectrum between 1260 – 1270 MHz.........................48
Figure 2.58: LimeSDR Mini transmitter spectrum at 145.6 MHz, gain 0..............................52
Figure 2.59: LimeSDR Mini transmitter spectrum at 145.6 MHz, gain 10............................52
Figure 2.60: LimeSDR Mini transmitter spectrum at 437.6 MHz, gain 0..............................53
Figure 2.61: LimeSDR Mini transmitter spectrum at 437.6 MHz, gain 10............................53
Figure 2.62: LimeSDR Mini transmitter spectrum at 1280 MHz, gain 0...............................54
Figure 2.63: LimeSDR Mini transmitter spectrum at 1280 MHz, gain 10.............................54
Figure 2.64: LimeSDR Mini transmitter spectrum at 2425 MHz, gain 0...............................55
Figure 2.65: LimeSDR Mini transmitter spectrum at 2425 MHz, gain 10.............................55
Figure 2.66: LimeSDR Mini CW transmitter output power as function of gain....................56
Figure 2.67: The BladeRF 2.0 micro transceiver...................................................................58
Figure 2.68: BladeRF 2.0 micro block diagram.....................................................................59
Figure 2.69: BladeRF 2.0 micro noise figure measurements.................................................60
Figure 2.70: BladeRF 2.0 micro blocking dynamic range measurements..............................62
Figure 2.71: BladeRF 2.0 micro receiver spectrum between 145 – 146 MHz.......................63
Figure 2.72: BladeRF 2.0 micro receiver spectrum between 435 – 438 MHz.......................63
Figure 2.73: BladeRF 2.0 micro receiver spectrum between 1260 – 1270 MHz...................64
v
Figure 2.82: BladeRF 2.0 micro transmitter spectrum at 145.5 MHz, gain 0........................69
Figure 2.83: BladeRF 2.0 micro transmitter spectrum at 145.5 MHz, gain 30......................69
Figure 2.84: BladeRF 2.0 micro transmitter spectrum at 437.5 MHz, gain 0........................70
Figure 2.85: BladeRF 2.0 micro transmitter spectrum at 437.5 MHz, gain 30......................70
Figure 2.86: BladeRF 2.0 micro transmitter spectrum at 1280 MHz, gain 0.........................71
Figure 2.87: BladeRF 2.0 micro transmitter spectrum at 1280 MHz, gain 25.......................71
Figure 2.94: BladeRF 2.0 micro CW transmitter output power as function of gain..............75
Figure 2.95: The Ettus USRP B210 transceiver.....................................................................77
Figure 2.96: USRP B210 block diagram................................................................................78
Figure 2.97: USRP B210 noise figure measurements............................................................79
Figure 2.98: USRP B210 blocking dynamic range measurements.........................................81
Figure 2.99: Ettus USRP B210 receiver spectrum between 145 – 146 MHz.........................82
Figure 2.100: Ettus USRP B210 receiver spectrum between 435 – 438 MHz.......................82
Figure 2.101: Ettus USRP B210 receiver spectrum between 1260 – 1270 MHz...................83
Figure 2.106: Ettus USRP B210 spectrum between 2400 – 2425 MHz.................................85
Figure 2.110: Ettus USRP B210 transmitter spectrum at 145.5 MHz, gain 35......................88
Figure 2.114: Ettus USRP B210 transmitter spectrum at 1280 MHz, gain 35.......................90
vi
Figure 2.122: Ettus USRP B210 CW transmitter output power as function of gain..............94
Figure 2.123: The Analog Devices PlutoSDR transceiver.....................................................96
Figure 2.124: PlutoSDR block diagram.................................................................................97
Figure 2.125: PlutoSDR noise figure measurements.............................................................98
Figure 2.126: PlutoSDR blocking dynamic range measurements........................................100
Figure 2.127: PlutoSDR receiver spectrum between 145 – 146 MHz.................................101
Figure 2.128: PlutoSDR receiver spectrum between 435 - 438 MHz..................................101
Figure 2.129: PlutoSDR receiver spectrum between 435 – 438 MHz.................................102
Figure 2.130: PlutoSDR receiver spectrum between 1260 – 1265 MHz.............................102
Figure 2.140: PlutoSDR transmitter spectrum at 145.5 MHz, power -40 dBm...................108
Figure 2.144: PlutoSDR transmitter spectrum at 1280 MHz, power -40 dBm.....................110
Figure 2.152: PlutoSDR transmitter spectrum at 5830 MHz with 20 MHz span.................114
Figure 2.153: PlutoSDR output power as function of Satsagen power setting.....................115
Figure 2.154: Max input power vs. noise figure at 5 kHz separation...................................117
Figure 2.155: Max input power vs. noise figure at 100 kHz separation...............................118
Figure 2.156: Max input power vs. noise figure at 1 MHz separation.................................118
Figure 2.157: Noise figure measurements for cross-checking the test software..................123
Figure 3.1: On-air test setup................................................................................................126
vii
List of Tables
Table 1.1: Selected SDR devices under test.................................................................2
Table 2.1: RTL-SDR Blog V3 technical specifications..............................................17
Table 2.2: RTL-SDR V3 settings used during noise figure measurements................18
Table 2.3: RTL-SDR V3 settings used during BDR measurements...........................19
Table 2.4: Airspy Mini technical specifications.........................................................27
Table 2.5: Airspy Mini settings used during noise figure measurements...................27
Table 2.6: Airspy Mini settings used during BDR measurements..............................28
Table 2.7: SDRplay RSPduo technical specifications................................................34
Table 2.8: SDRplay RSPduo settings used during noise figure measurements.........34
Table 2.9: SDRplay RSPduo settings used for BDR measurements..........................35
Table 2.10: LimeSDR Mini technical specifications..................................................41
Table 2.11: LimeSDR Mini settings used during noise figure measurements...........42
Table 2.12: LimeSDR Mini settings used for BDR measurements............................45
Table 2.13: LimeSDR Mini transmitter MER as function of gain.............................57
Table 2.14: BladeRF 2.0 micro technical specifications............................................60
Table 2.15: BladeRF 2.0 micro settings used during noise figure measurements......60
Table 2.16: BladeRF2.0 micro settings used for BDR measurements.......................61
Table 2.17: BladeRF 2.0 micro transmitter MER as function of gain........................76
Table 2.18: Ettus USRP B210 technical specifications..............................................79
Table 2.19: USRP B210 settings used during noise figure measurements.................79
Table 2.20: USRP B210 settings used for BDR measurements.................................80
Table 2.21: USRP B210 transmitter MER as function of gain...................................95
Table 2.22: PlutoSDR technical specifications..........................................................97
Table 2.23: PlutoSDR settings used during noise figure measurements....................98
Table 2.24: PlutoSDR settings used for BDR measurements.....................................99
Table 2.25: Lowest measured noise figure...............................................................116
Table 2.26: Highest measured output power.............................................................119
Table 2.27: Coaxial cable insertion loss...................................................................121
Table 2.28: Signal generator output power comparison at 145.5 MHz....................121
Table 2.29: Signal generator output power comparison at 437.5 MHz....................121
Table 2.30: Signal generator output power comparison at 1280 MHz.....................122
Table 3.1: SDR device settings used during the on-air tests....................................124
Table 3.2: Number of packets decoded during the on-air test campaign..................127
viii
Table 6.1: Insertion loss for BLK-89-S+ DC block......................................................I
Table 6.2: Insertion loss for BW-S20W2+ attenuator...................................................I
Table 6.3: Insertion loss for 141-24SMNM+ coaxial cable..........................................I
Table 6.4: Insertion loss for FL086-24SM+ coaxial cable..........................................II
Appendices
Appendix I: Insertion Losses..........................................................................................I
ix
List of Abbreviations and Symbols
ADC Analog-to-Digital Converter
API Application Programming Interface
BDR Blocking Dynamic Range
CW Carrier Wave or Continuous Wave
DDC Digital Down Converter
DR Dynamic Range
FFT Fast Fourier Transform
FPGA Field Programmable Gate Array
GPU Graphical Processing Unit
MDS Minimum Detectable Signal
MER Modulation Error Ratio
NF Noise Figure
RF Radio Frequency
RFIC Radio Frequency Integrated Circuit
SDR Software Defined Radio
SNR Signal to Noise Ratio
SFDR Spurious Free Dynamic Range
x
References
[1] SDRplay RSPduo datasheet:
https://www.sdrplay.com/wp-content/uploads/2018/05/RSPduoDatasheetV0.6.pdf
[2] Keysight Technologies, Noise Figure Measurement Accuracy: The Y-Factor
Method, (application note, local copy in Git):
https://www.keysight.com/dk/en/assets/7018-06829/application-notes/5952-
3706.pdf
[3] Wikipedia article on Minimum Detectable Signal:
https://en.wikipedia.org/wiki/Minimum_detectable_signal
[4] Wikipedia article on Johnson-Nyquist noise:
https://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise
[5] Wikipedia article on Boltzmann Constant:
https://en.wikipedia.org/wiki/Boltzmann_constant
[6] Rohde & Schwarz, The Y Factor Technique for Noise Figure Measurements (appli-
cation note, local copy in Git):
https://www.rohde-schwarz.com/pl/applications/the-y-factor-technique-for-noise-fig-
ure-measurements-application-note_56280-15484.html
[7] Walt Kester, Taking the Mystery out of the Infamous Formula, "SNR = 6.02N +
1.76dB," and Why You Should Care, Analog Devices MT-001:
https://www.analog.com/media/en/training-seminars/tutorials/MT-001.pdf
[8] Error vector magnitude on Wikipedia: https://en.wikipedia.org/wiki/
Error_vector_magnitude
[9] ETSI EN 302 307-1 V1.4.1
[10] Softrig Git repository: https://gitlab.com/csete/softrig
[11] GNU Radio: https://gnuradio.org
[12] GrSoapy SDR device wrapper module for GNU Radio:
https://gitlab.com/librespacefoundation/gr-soapy
[13] Git repository for this sub-activity:
https://gitlab.com/librespacefoundation/sdrmakerspace/sdreval
[14] SATSAGEN http://www.albfer.com/en/author/wp_10278758/
[15] RTL-SDR wiki page at Osmocom: https://osmocom.org/projects/rtl-sdr/wiki/Rtl-sdr
[16] Airspy Mini product page: https://airspy.com/airspy-mini/
[17] SDRplay RSPduo Technical Information:
https://www.sdrplay.com/wp-content/uploads/2018/06/RSPDuo-Technical-Informa-
tion-R1P1.pdf
[18] LimeSDR Mini product page: https://www.crowdsupply.com/lime-micro/limesdr-mini
[19] LMS7002M data sheet v3.1:
https://limemicro.com/app/uploads/2017/07/LMS7002M-Data-Sheet-v3.1r00.pdf
[20] Lime Microsystems projects on GitHub: https://github.com/myriadrf
[21] LMS7002M RF and Analog Measurement Results by Lime Microsystems:
https://limemicro.com/app/uploads/2015/08/LMS7002M_Measurements-v1_05.pdf
[22] BladeRF 2.0 micro product page: https://www.nuand.com/bladerf-2-0-micro/
[23] Nuand BladeRF project on GitHub: https://github.com/Nuand/bladeRF
[24] BladeRF 2.0 micro schematics: https://www.nuand.com/bladeRF-micro.pdf
xi
[25] Ettus USRP B210 product page: https://www.ettus.com/all-products/ub210-kit/
[26] Ettus Research on GitHub: https://github.com/EttusResearch
[27] USRP B210 schematics: https://files.ettus.com/schematics/b200/b210.pdf
[28] USRP B2xx Environmental Specifications:
https://kb.ettus.com/B200/B210/B200mini/B205mini#Environmental_Specifications
[29] USRP B200 Performance Data:
https://kb.ettus.com/images/c/cb/B200_RF_Performance.pdf
[30] PlutoSDR Wiki page: https://wiki.analog.com/university/tools/pluto
[31] PlutoSDR System Issues:
https://wiki.analog.com/university/tools/pluto/users/name#system_issues
[32] ADALM-PLUTO Hardware:
https://wiki.analog.com/university/tools/pluto/hacking/hardware
[33] SatNOGS Open Source Ground Station Network: https://satnogs.org/
[34] SatNOGS ground station 1353: https://network.satnogs.org/stations/1353/
[35] SatNOGS ground station 1354: https://network.satnogs.org/stations/1354/
xii
1. Introduction
1 Introduction
1.1 Purpose and Scope

During recent years, we have seen tremendous development in the availability of low cost,
consumer-grade SDR hardware. Unfortunately, the low cost and the fast pace at which SDR
hardware is developed does not leave much room for delivering complete turnkey solutions to
the users. Hardware is often delivered as “evaluation boards” with little or no RF performance
specifications, and software support is often left to third parties.
The purpose of this activity is to evaluate a selection of currently available SDR hardware
and software from a satellite communications point of view, in order to provide prospective
users with independent data that is often not available from the manufacturers. The evaluation
includes both analytical assessments as well as lab tests supplemented by over the air usage of
the systems.
The work was executed in three parts:
1. Evaluation of the technical specifications, where we tested the SDR devices for key
parameters such as noise figure, dynamic range, and transmit power. This part is cov-
ered in chapter 2 of the report.
2. On-air evaluation, where two SDR devices were compared side-by-side as satellite
telemetry receivers. This part is covered in chapter 3 of the report.
3. SDR software evaluation, where we took a look at how well the SDR devices are sup-
ported by the most common SDR software. This is covered in chapter 4 of the report.
1.2 Selected SDR devices

The SDR devices under test were selected according to the following criteria:
• Relevant for satellite communications, most notably for CubeSat projects
• Affordable and easily available
• Variety in architecture and expected performance
The list of devices that were tested is shown in Table 1.1 below.
1
1. Introduction
Device Type Frequency range Bandwidth Bits

RTL-SDR Blog V3 RX 0.1 – 1700 MHz 3.2 MHz 8
Airspy Mini RX 24 – 1700 MHz 6 MHz 12
SDRplay RSPduo RX 1 kHz – 2 GHz 10 MHz 8-14
LimeSDR Mini RX/TX 10 MHz – 3.5 GHz 30 MHz 12
BladeRF 2.0 Micro RX/TX 70 MHz – 6 GHz 56 MHz 12
Ettus USRP B210 RX/TX 70 MHz – 6 GHz 56 MHz 12
PlutoSDR RX/TX 70 MHz – 6 GHz 20 MHz 12
Table 1.1: Selected SDR devices under test.

Notes:
• In practice, RTL-SDR devices are limited to 2.4 MSPS continuous transfer to the host
computer [15].
• SDRPlay RSPduo has 14-bits resolution up to 6.048 MSPS, 12-bits up to 8.064
MSPS, 10-bits up to 9.216 MSPS, and 8-bits above 9.216 MSPS [1].
1.3 Evaluation Criteria

Given the scope of the activity, the evaluation of the devices was limited to the frequency
bands used for satellite communications:
• VHF: 145-146 MHz
• UHF: 435-438 MHz
• L-band: 1.26-1.30 GHz
• S-band: 2.0-2.1 GHz TX and 2.2-2.3 GHz RX
• S-band: 2.40-2.45 GHz
• C-band: 5.8 GHz
The evaluation criteria according to which the devices are evaluated are grouped in three
groups:
1. Generic parameters (extracted from datasheets):
◦ Frequency range, stability, and phase noise
◦ Maximum bandwidth
◦ Operating temperature range
◦ Cost and availability
◦ Size and weight
2. Receiver-specific parameters (tested):
2
1. Introduction
◦ Noise figure (NF) and minimum detectable signal (MDS)

◦ Receiver dynamic range
◦ Spectral purity
3. Transmitter-specific parameters (tested):
◦ Maximum output power
◦ Spectral purity
◦ Achievable MER
1.4 Evaluation Methods

One of our objectives during this activity was to provide useful and reproducible results. To
that end, we have defined our test procedures according to the following criteria:
1. Use simple methods that do not require expensive RF facilities or laboratory equip-
ment.
2. Carry out the tests under realistic conditions that are representative for a setup under
which the SDR devices are normally used.
Such methods can not provide results with the same accuracy as measurements carried out in
a professional RF laboratory. However, in order to increase confidence in our results, we have
cross-check the measurements using different test methods and tools. These cross-checks are
described in section 2.3.
Furthermore, since we use the same methods to test all devices, the results can be used for
a relative comparison between the devices, even if the measurements are not accurate on an
absolute scale.
3
2. Evaluation of Technical Specifications
2 Evaluation of Technical Specifications
2.1 Test Setup and Procedures
2.1.1 Receiver Noise Figure
2.1.1.1 Background
The receiver sensitivity determines the smallest signal that a receiver can reliably detect [2].
Therefore, it is also called the minimum detectable signal [3], MDS, and specifies the strength
of the smallest signal at the input of the receiver that causes the output signal power to be M
times the output noise power.
We use M = 1 and measure the MDS as the input signal that leads to a 3 dB increase in out-
put power.
The receiver noise figure and noise factor are measures of degradation in signal-to-noise
ratio in the receiver caused by internal thermal noise in the receiver [4]. The noise figure (NF)
is the noise factor (F) expressed in dB:
NF =10 log 10 ( F )
The noise factor is closely related to the MDS of the receiver [2]:
MDS=k B ×T 0×F×B×M
where kB is the Boltzmann constant in units of J/K [5], T0 is the temperature of the device, B is
the detection bandwidth in Hz, and M is the power ratio used in the MDS measurement.
Using T0 = 290 K and expressing the MDS in dBm:
MDS=−174 dBm+ NF +10 log 10 ( B)+ 10 log 10 ( M )
Using M = 1, we can calculate the noise figure from the MDS:
NF =MDS+174 dBm−10 log 10 (B)
Another common way to measure the noise figure is the Y-factor method. The Y-factor
method uses a calibrated noise source with well known excess noise ratio (ENR). The noise
4
source is connected to the device under test and the noise power output density from the de-
vice is then measured with the noise source ON and OFF. The noise figure of the device is
then given by [6]:
ENR
NF =10 log 10
( )
10
Y
10
10 −1
10
=ENR−10 log 10 ( 10Y / 10−1 )
where ENR is the noise figure of the noise source and Y is the difference between output
power density from the device under test when the noise source is ON and OFF.
When the noise figure of the device becomes larger than the ENR of the noise source, the
difference between noise source ON and OFF becomes difficult to measure. Therefore, we
will be using both the MDS method and the Y-factor method to measure the noise figure.
Moreover, the MDS method is simpler because it does not require a calibrated ENR source
and is therefore interesting to compare it to the more accurate Y-factor method.
2.1.1.2 Test Procedure
The device under test is connected to a calibrated signal generator or ENR source, and the out-
put power from the receiver is measured using the test SDR software described in section
2.1.7.
Settings:
• Sample rate and analog bandwidth are set to what is considered optimal for the device
• The DDC is tuned as to avoid DC and other spurious signals in the pass-band
• All automatic gain control in hardware and software is turned OFF
• The measurement bandwidth is set to 500 Hz
The measurement is carried out according to the following steps listed below.
MDS method:
1. With the signal generator OFF, measure the output power in 500 Hz bandwidth
2. With the signal generator ON, find the input power in dBm that gives a 3 dB increase
in output power; this is our MDS
3. Calculate the noise figure using the formula NF = MDS + 147
4. Repeat steps 1-3 for different gain settings in each frequency band of interest listed in
section 1.3
5
Figure 2.1: Test setup for noise figure measurements using the MDS method.
The MDS value obtained using the method described above has to be corrected for the inser-
tion losses introduced by the DC block and the coaxial cable, as well as the deviation of the
attenuator from the nominal 20 dB attenuation. The correction is performed by subtracting the
insertion losses to the MDS obtained from the signal generator:
MDS actual =MDS−IL DCBLK −IL ATT −IL cable
The insertion losses of the individual parts in the setup are listed in Appendix I.
Y-factor method:
1. With the noise source OFF, measure the output power in 500 Hz bandwidth
2. With the noise source ON, measure the output power in 500 Hz bandwidth
3. Calculate the Y-factor using the formula Y = PON − POFF
4. Calculate the noise figure using the formula NF = ENR − 10 log10(10Y/10 − 1)
5. Calculate the MDS using the formula MDS = NF − 147
6. Repeat steps 1-5 for different gain settings in each frequency band of interest listed in
section 1.3
Figure 2.2: Test setup for noise figure measurements using the Y-factor method.
To avoid unnecessary losses, the SDR device is connected directly to the DC block. There-
fore, only the insertion loss of the DC block has to be taken into account and this has the ef-
fect of increasing the measured noise figure.
6
2.1.2 Receiver Dynamic Range
2.1.2.1 Background
The dynamic range of a receiver tells us the largest signal-to-noise ratio (SNR) that we can
achieve in the receiver. In other words, it expresses the difference between the strongest and
the weakest signal a receiver can receive at the same time and is therefore of great importance
when receiving weak signals from satellites in the presence of strong local interference.
A theoretical estimate for the dynamic range of an SDR receiver can be obtained from the
bit resolution and the quantization noise of the analog to digital converter (ADC) in the re-
ceiver [7]:
SNR [dB]=6.02×Q+1.76
where Q is the bit resolution of the ADC. However, analog circuitry in front of the ADC as
well as digital signal processing after the ADC can enhance or degrade the dynamic range of
the complete system.
There are different metrics that can be used to characterize the dynamic range of a receiver
system. We will limit our measurements to finding the blocking dynamic range, BDR, and the
1 dB compression point P1dB.
Blocking occurs when the receiver performance is degraded by a strong input signal. This
often happens because of:
1. Reciprocal mixing, i.e. when a strong signal outside of the received pass-band mixes
with the phase noise of the receiver's local oscillator or another strong signal, generat-
ing extra spurs or noise in the receiver, or
2. A strong input signal overloads an analog component (amplifier, mixer, ADC) making
it behave in a non-linear fashion
In both cases, the degradation can be measured either as a change in noise level, see Figure
2.3, or as a clearly noticeable "receiver breakdown" in the form of spurs in the receiver spec-
trum, see Figure 2.4.
7
Figure 2.3: Disturbing signal causing increase in the noise level at 100 kHz away.
Figure 2.4: Disturbing signal causing spurs as far as 1 MHz away.
The power level just before the reciprocal mixing or overload, Pin, is the maximum power the
receiver can handle. The blocking dynamic range is then found as the difference between the
power of this input signal, Pin, and the minimum detectable signal, MDS, found during the
noise figure measurements:
8
BDR [dB ]= Pi n−MDS
The 1 dB compression point, P1dB, is the input power level at which an increase in input
power by X dB leads to an increase in output power by X-1 dB, hence the 1 dB compression.
When this occurs, the dynamic range of the receiver is calculated as the difference between
the P1dB and the MDS found during noise figure measurements:
DR [ dB]=P1 dB −MDS
2.1.2.2 Test Procedure
The dynamic range is measured in a similar setup that was used for noise figure measure-
ments, however, we use both a power detector and spectrum plot at the same time.
Figure 2.5: Test setup for dynamic range measurements.
The blocking dynamic range is measured as follows:

1. Tune the receiver to the desired frequency f0 (closed to what was used during the noise
figure)
2. Tune the signal generator (unwanted signal) to a frequency 5 kHz away from f0
4. With the signal generator ON, find the input power Pin that gives a 3 dB increase in
output noise power, or until spurs appear in the vicinity of f0 and the strongest spur is 3
dB above the noise floor, c.f. figures 2.3 and 2.4
5. Calculate the dynamic range using the formula BDR = Pin − MDS
6. Repeat steps 2-5 with a signal 100 kHz and 1 MHz away
Since this setup also uses a DC block and coaxial cable, Pin has to be corrected for the inser-
tion losses of these components:
9
Pactual =P i n−IL DCBLK −IL cable
The receiver gain compression, P1dB, is measured as follows:

1. Tune the receiver and the signal generator to the desired frequency f 0 (closed to what
was used during the noise figure)
3. With the signal generator ON, increase the input power Pin by 2-3 dB at a time until 1
dB compression is observed in the output power
4. Calculate the dynamic range using the formula DR = P1dB − MDS
Note:
Receiver gain compression could not be measured on any of the devices under test because
the ADC got saturated before any gain compression could be observed. Therefore, there are no
P1dB measurements.
2.1.3 Receiver Spectral Purity
The spectral purity of the receiver is tested through visual observation of the spectrum using
an FFT in the SDR software. The tests are carried out in the frequency bands listed in section
1.3:
1. Connect a 50-ohm terminator to the receiver input
2. Adjust the receiver frequency, sample rate, and decimation to cover as much of the
band of interest as possible
3. Set the receiver to maximum gain
4. Take a screenshot
Figure 2.6: Test setup for receiver spectral purity measurements.
2.1.4 Transmitter Spectral Purity
The transmitter spectral purity is tested by transmitting a CW test signal in the frequency
bands of interest listed in section 1.3. The output is checked using a spectrum analyzer.
10
Figure 2.7: Test setup for transmitter power and spectral purity measurements.
2.1.5 Transmitter Output Power
The transmitter output power is measured by transmitting a CW test signal at various gain set-
tings and measuring the output power using a spectrum analyzer. The setup used for this is the
same as the setup used for transmitter spectral purity measurements shown in Figure 2.9.
Since this setup also uses a DC block and a coaxial cable, the measured CW power is cor-
rected for the insertion losses from these components:
Pactual =P FSW + IL DCBLK + IL cable
where PFSW is the power measured by the spectrum analyzer.
2.1.6 Transmitter Modulation Error Ratio
The transmitter modulation error ratio (MER) is a measure used to quantify how much the
transmitted symbols deviate from their ideal value as illustrated on the constellation diagram
in Figure 2.8 and given by the formula:
Psignal
MER(dB)=10×log 10
( )
P error
11
01 11
Perror
Pref
00 10
Figure 2.8: Constellation with error [8].
Transmitter MER is measured by transmitting a DVB-S2 signal and measuring the SNR and
MER using a DVB-S/S2 signal analyzer and using the following settings:
• 6.25 Msymbols / sec
• Modulation: 8-PSK
• FEC rate: 9/10
• FEC frame: Normal
• Roll-off factor: 0.20
• Pilots: ON
Figure 2.9: Test setup for transmitter MER measurements.
The analyzer used in our setup was a ROVER HD Flash DVB-S/S2 signal analyzer. It mea-
sures the signal power, MER, and noise margin in dB, the latter being presumably a margin
above the ES/N0 that is required for error free reception, which is 11 dB for the modulation and
FEC-rate used in our setup, c.f. Table 13 in ETSI-EN-302-307-1 [9].
12
Figure 2.10: Screen captures of the ROVER DVB-S/S2 analyzer.
2.1.7 Test Software
For the receiver tests (noise figure, dynamic range, spectral purity), a dedicated Qt-based ap-
plication [10] with the following components has been devloped:
1. Device interface to configure, control and read samples from the SDR device under
test.
2. Spectrum plot using FFT and Qt graphics.
3. Digital down-converter to select a channel with adjustable bandwidth.
4. RMS power detector with adjustable averaging time.
5. Basic AM, FM, CW, and SSB demodulators.
A functional diagram for the test software used for receiver tests is shown on Figure 2.11
below. In order to ensure sufficient support for the SDR devices, the test software does not use
any SDR device wrappers but interfaces directly to the SDR device driver libraries.
Figure 2.11: Functional diagram for the test software used for receiver tests.
13
For transmitter tests (output power, MER, spectral purity), simple applications based on
GNU Radio 3.8 [11] and the gr-soapy [12] device interface module were created. These appli-
cations are available in the project repository [13] and two examples using the BladeRF 2.0
micro are shown on Figure 2.12 and Figure 2.13 below.
Figure 2.12: GNU Radio application used for output power measurements.
Figure 2.13: GNU Radio application used for MER measurements.
Unfortunately, our GNU Radio transmitter scripts using the gr-soapy SDR API did not
work with the PlutoSDR. The CW output couldn’t generate any CW-like signal and te DVB-
S2 output was oscillating in amplitude. In order to perform at least some CW power measure-
ment, we decided to use a new tool called SATSAGEN [14], which is specifically written for
the PlutoSDR.
14
Figure 2.14: The SATSAGEN spectrum analyzer software for PlutoSDR.
15
2.2 Test Results

The following sections present the test results from our lab measurements. For each device
tested, we begin with a brief description and the technical specifications of the device under
test, followed by the receiver and transmitter performances measured during our tests. Finally,
in section 2.2.8 we provide some comparisons between devices.
The data that was used to generate the graphs presented in this report is available in plain
text format in the project GIT repository.
2.2.1 RTL-SDR Blog V3
RTL-SDR dongles are based on the Realtek RTL2832U DVB-T demodulator chip that was
found to be capable of transferring raw I/Q samples to the host computer. This characteristic
was discovered already in 2012 and has made the RTL2832U-based dongles the most popular
SDR receivers to date. Compared to generic RTL-SDR devices, the RTL-SDR Blog V3 don-
gle has been redesigned with SDR users in mind providing significantly better performance
than the generic RTL-SDR dongles.
Figure 2.15: The RTL-SDR blog V3 receiver.
The RTL-SDR Blog V3 dongle uses a Rafael R820T2 24-1700 MHz RF tuner, which de-
livers analog I/Q samples to the RTL2832U demodulator chip. The RTL2832U chip acts as 8-
bit analog to digital converter and USB interface to the host computer and can also function as
a direct sampling shortwave receiver, see figure 2.16 below.
16
Figure 2.16: RTL-SDR Blog V3 dongle block diagram.

Software support for RTL-SDR devices is available through librtlsdr, a light-weight, open-
source user space library written in C.
A summary of the technical specifications for the RTL-SDR Blog V3 dongle is listed in Ta-
ble 2.1 below.
Frequency Range 500 kHz – 1766 MHz

Sample rate 2.4 MSPS
RF bandwidth 350 kHz – 8 MHz
Frontend filters Tracking RF filters in the R820T2 tuner
RX paths 1
RX inputs 1
ADC resolution 8-bit
Claimed noise figure -
Clained dynamic range -
Reference clock 1 PPM TCXO
Other features Bias T
Direct sampling
Expansion ports
Temperature range -
Size 10 cm × 2.5 cm × 1.4 cm
Weight 30 g
Approximate price 22 USD
Product page https://www.rtl-sdr.com/buy-rtl-sdr-dvb-t-dongles/
Table 2.1: RTL-SDR Blog V3 technical specifications.
2.2.1.1 Receiver Noise Figure
The settings used during the RTL-SDR noise figure measurements are listed in Table 2.2 and
the results are shown on Figure 2.17 below.
17
Input rate Decimation Sample rate

2.4 MHz 8 300 kHz
Table 2.2: RTL-SDR V3 settings used during noise figure measurements.
We note that:
• There is very good consistency between the noise figures calculated from the MDS
and the noise figures measured using the Y-factor method.
• The noise figure at high gains is significantly worse on 1280 MHz than on 145 and
437 MHz. This is not unexpected as insertion losses in the signal path usually increase
with frequency. Furthermore, the datasheet of the R820T2 tuner only specifies the per-
formance parameters up to 1002 MHz and anything above this frequency can be con-
sidered out of spec.
Figure 2.17: RTL-SDR V3 noise figure measurements.
2.2.1.2 Receiver Blocking Dynamic Range
The settings used during the RTL-SDR dynamic range measurements are listed in Table 2.3
and the results are shown on Figure 2.18 below.
18
Input rate Decimation Sample rate

2.4 MHz 1 2.4 MHz
Table 2.3: RTL-SDR V3 settings used during BDR measurements.
We note that:
• Ax expected, the highest input power that the device can tolerate without overload, P in,
is decreasing as the gain is increased. The relationship is mostly linear except on 1280
MHz.
• The dynamic range is mostly constant over the entire gain spectrum, slightly dropping
at the highest gains.
• The measured dynamic range is significantly higher than what we would expect from a
device with an 8-bit ADC (48 dB). The only way we can explain this is high oversam-
pling by the ADC and that non-linearity artifact are below the quantization noise of the
ADC.
19
Figure 2.18: RTL-SDR V3 blocking dynamic range measurements.
In order to verify that the measured dynamic range is realistic, we performed a test where
the signal from an antenna and the signal from the signal generator were combined using a 3
dB power splitter. The receiver was tuned to a weak beacon transmitting constant power and
the signal from the signal generator was tuned to 5 kHz away from the beacon and the power
was increased to the measured Pin + 3 dB. We could then verify that the signal from the signal
generator did not decrease the signal to noise ratio of the beacon.
20
2.2.1.3 Receiver Spectral Purity
The receiver spectra of the RTL-SDR V3 receiver are shown on Figures 2.19 to 2.26 below.
As we can see from the spectra, there are very few spurs generated by or picked up by the de-
vice when connected to 50 ohm terminator.
Figure 2.19: RTL-SDR receiver spectrum between 145 – 146 MHz.
21
22
23
24
25
2.2.2 Airspy Mini
The Airspy Mini has similar architecture as the RTL-SDR, except that the RTL2832U DVB-T
demodulator chip is replaced with an LPC4370 micro-controller with integrated 12-bit, 20
MSPS analog to digital converter, providing both higher sample rates and potentially better
dynamic range.
Figure 2.27: The Airspy Mini receiver.
The Airpsy Mini uses IF sampling to capure the bandwidth of interest. Conversion to quad-
rature samples is done by the driver running on the host PC. Thus, the Airspy Mini does not
suffer from the I/Q imbalance, DC offset, and 1/F noise often seen in SDR receivers that use
direct conversion [16].
Figure 2.28: Airspy Mini block diagram.
Software support for Airspy devices is available through libairspy, a light-weight, open-
source user space library written in C.
26
A summary of the technical specifications for the Airspy Mini is listed in Table 2.4 below.
Frequency Range 24 – 1700 MHz

Sample rate 3, 6, 10 MSPS
RF bandwidth 2.4 – 6 MHz
Frontend filters Tracking RF filters
RX paths 1
RX inputs 1
ADC resolution 12-bit (10.4 ENOB)
Claimed noise figure 3.5 dB between 42 – 1002 MHz
Claimed dynamic range 70 dB SNR, 95 dB SFDR for the ADC
Reference clock 0.5 PPM
Other features Bias T
Temperature range -10 to 40 °C
Size 8 cm × 2.7 cm × 1 cm
Weight 20 g
Approximate price 99 USD
Product page https://airspy.com/airspy-mini/
Table 2.4: Airspy Mini technical specifications.
The settings used during the Airspy Mini noise figure measurements are listed in Table 2.5
Input rate Decimation Sample rate Gain mode

6 MHz 16 375 kHz Linearity
Table 2.5: Airspy Mini settings used during noise figure measurements.
We note that:
• There is very good consistency between the noise figure calculated from the MDS
measurements and the noise figure measured using the Y-factor method.
• Like with the RTL-SDR, the noise figure on 1280 MHz is worse that the noise figure
on 145 and 437 MHz.
• We could not reproduce the claimed noise figure of 3.5 dB.
27
Figure 2.29: Airspy Mini noise figure measurements.
The settings used during the Airspy Mini dynamic range measurements are listed in Table 2.6
Input rate Decimation Sample rate Gain mode

6 MHz 2 3 MHz Linearity
Table 2.6: Airspy Mini settings used during BDR measurements.
We note that:
MHz.
• The measured dynamic range is in the neighborhood of the claimed dynamic range of
70 dB SNR and 95 dB SFDR for the ADC.
28
Figure 2.30: Airspy Mini blocking dynamic range measurements.
The receiver spectra of the Airspy Mini receiver are shown on Figures 2.31 to 2.35 below. As
we can see from the spectra, there are very few spurs generated by or picked up by the device
when connected to 50 ohm terminator.
29
Figure 2.32:
2.31: Airspy Mini receiver spectrum between 435
145 – 438
146 MHz.
Figure 2.33: Airspy Mini receiver spectrum between 1260 – 1264 MHz.
30
31
2.2.3 SDRplay RSPduo
The SDRplay RSPduo is a dual-tuner 14-bit receiver based on a tuner and USB chipset from
Mirics, covering the entire spectrum from 1 kHz to 2 GHz with up to 10 MHz maximum
bandwidth [1].
Unlike the other devices with multiple receive paths, the two tuners in the SDRplay RSP
duo function independently of each other and can receive different frequency bands at the
same time. This makes them particularity interesting for applications such as a SatNOGS
ground station, where VHF and UHF could be received simultaneously using just one device.
Other noteworthy features of SDRplay devices include the extensive set of band-pass and
notch filters as well as the extensive documentation of receiver performance measurements
available on the website [17].
Figure 2.36: The SDRplay RSPduo receiver.
Software support for SDRplay devices is available through a user space library called lib-
mirsdrapi-rsp. Unlike all the drivers for other SDR devices tested during this activity, the
SDRplay driver library is closed-source and available as binary download only. Furthermore,
the software API provided by the driver is not as simple and intuitive as the ones provided by
e.g. RTL-SDR or Airspy.
32
Figure 2.37: The SDRplay RSPduo block diagram.

A summary of the technical specifications for the SDRplay RSPduo is listed in Table 2.7
below.
33
Frequency Range 1 kHz – 2 GHz

Sample rate 2 – 10.66 MSPS
RF bandwidth 200 – 8000 kHz
Frontend filters 2 MHz low-pass
2-12 MHz
30-60 MHz
60-120 MHz
120-250 MHz
250-300 MHz
300-380 MHz
380-420 MHz
420-1000 MHz
1000 MHz high-pass
RX paths 2
RX inputs 3
ADC resolution 14-bit up to 6.048 MSPS
12-bit between 6.048-8.064 MSPS
10-bit between 8.064-9.216 MSPS
8-bit above 9.216 MSPS
Claimed noise figure ~ 2.5 dB below 500 MHz
~ 4.5 dB below 1.3 GHz
Reference clock 0.5 PPM TCXO
Other features Independent tuners
Bias T
External clock
Temperature range -
Size 9.5 cm × 8.0 cm × 3 cm
Weight 315 g
Approximate price 240 €
Product page https://www.sdrplay.com/rspduo/
Table 2.7: SDRplay RSPduo technical specifications.
The settings used during the SDRplay RSPduo noise figure measurements are listed in Table
2.8 and the results are shown on Figure 2.38 below.
34
Input rate Decimation Sample rate IF Bandwidth IF gain

4 MHz 8 500 kHz 600 kHz -20 dB
Table 2.8: SDRplay RSPduo settings used during noise figure measurements.
We note that:
• As with the previous devices, the noise figure on 1280 MHz is worse than on 145 and
437 MHz.
• The lowest noise figure we measured was 2.5 dB using the MDS method and 2.1 dB
using the Y-factor method. This indicates that our test setup is capable of measuring
such low noise figures.
• The measured noise figures are overall consistent with the data available on the SDR-
play website, although direct 1:1 comparison is not possible due to the complicated
gain API provided by the driver.
Figure 2.38: SDRplay RSPduo noise figure measurements.
The settings used during the SDRplay RSPduo dynamic range measurements are listed in Ta-
ble 2.9 and the results are shown on Figure 2.39 below.
Input rate Decimation Sample rate IF Bandwidth IF gain

4 MHz 2 2 MHz 200 kHz -40 dB
Table 2.9: SDRplay RSPduo settings used for BDR measurements.
35
Figure 2.39: SDRplay RSPduo blocking dynamic range measurements.
We note that:
MHz.
36
• The receiver has a rather poor dynamic range when the disturbing signal is only 5 kHz
away but improves significantly at larger separations, thanks to the narrow IF filter.
• Changing the IF gain can have large influence on the dynamic range and, as we have
seen, not so much on the noise figure. That is why we chose -40 dB instead of the -20
dB used during the noise figure measurements.
The receiver spectra of the SDRplay RSPduo receiver are shown on Figures 2.40 to 2.43 be-
low. As we can see from the spectra, there are very few spurs generated by or picked up by the
device when connected to 50 ohm terminator.
Figure 2.40: SDRplay RSPduo receiver spectrum between 145 – 146 MHz.
37
38
39
2.2.4 LimeSDR Mini
The LimeSDR Mini is a small 10 MHz – 3.5 GHz full-duplex transceiver based on the LM-
S7002M integrated transceiver from Lime Microsystems [18]. Its small size and low cost
makes it a popular choice among hobbyists looking for entry-level SDR transceiver hardware.
Although the LMS7002M is capable of delivering bandwidths up to 96 MHz [19], the compo-
nents used in the LimeSDR Mini limit its capabilities to 30.72 MHz.
Both the LimeSDR Mini hardware and accompanying software come as open-source with
design files and source code available on GitHub [20]. All LimeSDR devices are controlled
through a LimeSuite library, which provides a simple API to user applications.
Figure 2.44: LimeSDR Mini transceiver.
Figure 2.45: LimeSDR Mini block diagram.
40
A summary of the technical specifications for the LimeSDR Mini is listed in Table 2.10 be-
low.
Frequency Range 10 – 3800 MHz
RF bandwidth 30.72 MHz
Frontend filters Variable bandwidth low-pass filter before ADC
RX paths RX1_W for 10 MHz – 2 GHz
RX1_H for 2 – 3.5 GHz
RX inputs 1
TX paths TX1_1 for 2 – 3.5 GHz
TX1_2 for 10 MHz – 2 GHz
TX outputs 1
ADC resolution 12
DAC resolution 12
Claimed transmit power -
Reference clock VCTCXO (±1 ppm initial, ±4 ppm stable)
Other features On-board EEPROM and flash for firmware and FPGA image
GPIO through header
External clock through uFL connectors
Open source design
Temperature range -
Size 9.5 cm × 3.6 cm × 1.3 cm (with enclosure)
Weight 42 g
Approximate price 159 USD (board only)
299 USD (with aluminum enclosure)
Product page https://limemicro.com/products/boards/limesdr-mini/
Table 2.10: LimeSDR Mini technical specifications.
The settings used during the LimeSDR Mini noise figure measurements are listed in Table
41
Input rate Decimation Sample rate RF Path Bandwidth LPF

4 MHz 8 500 kHz Auto Auto ON
Table 2.11: LimeSDR Mini settings used during noise figure measurements.
Figure 2.46: LimeSDR Mini noise figure measurements.

We note that:
• The LimeSDR Mini was in general suffering from excessive noise, which is clearly re-
flected by the poor noise figure we measured even at the highest gain.
• Above 2 GHz, the noise was so high and sensitive to touching the enclosure that we
didn’t find it meaningful to include the results.
• We had two units at our disposal and both of them behaved the same way.
• We have verified that Analog LFP ON/OFF does not impact the noise figure (within 1
dB).
The noise figure of the LMS7002M RFIC is around 2 dB [21] and it is unlikely that the
difference is caused by insertion losses between the antenna connector and the RFIC. Because
the noise level was sensitive to touching the case, we suspected that the case might be the
cause of the poor noise figure. We took the board out of its aluminum case and repeated the
noise figure measurements using the Y-factor method. Figure 2.47 shows these measurements
on the LimeSDR Mini board only vs. the board inside the aluminum case.
42
Figure 2.47: LimeSDR Mini noise figure measured with and without the aluminum case.
43
The measurements show clearly that the receiver noise figure is severely degraded by the
aluminum case. Lime Microsystems investigated the issue and concluded that the issue could
be caused by coupling via the case to the receive port SMA, where there was a scratch in the
case paint. This can be resolved by isolating the SMA connector ground by adding a small
piece of heat shrink around it.
The subsequent tests use the measurements on the board inside the case as reference since
this is how our units were shipped from the factory.
44
The settings used during the LimeSDR Mini dynamic range measurements are listed in Table
Input rate Decimation Sample rate RF path Bandwidth LPF

2 MHz 1 3 MHz Auto 1401 kHz ON
Table 2.12: LimeSDR Mini settings used for BDR measurements.
Figure 2.48: LimeSDR Mini blocking dynamic range measurements.
45
We note that:
is decreasing as the gain is increased.
• At 5 kHz and 100 kHz separation, the dynamic range is mostly constant over the entire
gain spectrum, but varies more than 20 dB at 1 MHz separation.
The receiver spectra of the LimeSDR Mini receiver are shown on Figures 2.49 to 2.57 below.
As we can see from the spectra, there are relatively few spurs generated by or picked up by the
device when connected to 50 ohm terminator. However, the noise floor becomes significantly
higher as we go above 2 GHz, which was also noted during the noise figure measurements.
46
Figure 2.49: LimeSDR Mini receiver spectrum between 145 – 146 MHz.
47
48
49
50
2.2.4.4 Transmitter Spectral Purity
The transmitter spectra for the LimeSDR Mini transmitting a CW signal are shown on Figures
2.58 to 2.65 below. For each frequency, two spectra are shown, one at a gain 0 and one at gain
10. The tests were carried out with a modified gr-soapy interface, where the initial call to the
IQ balancing function was disabled in order to prevent a large IQ imbalance caused by the de-
fault settings in gr-soapy.
51
Figure 2.58: LimeSDR Mini transmitter spectrum at 145.6 MHz, gain 0.
52
53
Figure 2.62: LimeSDR Mini transmitter spectrum at 1280 MHz, gain 0.
54
55
2.2.4.5 Transmitter Output Power
The measured transmitter output for the LimeSDR Mini is shown on Figure 2.66 below. As
we can see there is a linear relationship between the transmitter gain setting and the output
power, except at the lower and upper limits where the curve begins to flatten out.
Figure 2.66: LimeSDR Mini CW transmitter output power as function of gain.
2.2.4.6 Transmitter MER
The measured output power and modulation quality for a DVB-S2 signal transmitted with the
Mini are shown in Table 2.13 below. As we noted in test procedure description in section
2.1.6, we believe that the NsMAR parameter is a margin above the required E S/N0, which is
11 dB for the modulation and FEC settings used in our tests. Also note that the saturation in
NsMAR and MER is due to the limitation of the signal analyzer and does not necessarily rep-
resent a limitation in the SDR device.
56
TX Gain Power (dBm) NsMAR (dB) MER (dB)

-10 -60.9 7.4 19.8
-5 -57.3 12.7 25.0
0 -52.1 12.7 25.0
10 -42.7 11.5 25.0
20 -35.3 12.7 25.0
30 -22.9 12.7 25.0
40 -15.3 12.7 25.0
50 -3.0 12.7 25.0
Table 2.13: LimeSDR Mini transmitter MER as function of gain.
57
2.2.5 BladeRF 2.0 Micro
The BladeRF 2.0 micro is a 2×2 MIMO transceiver based on the AD9361 integrated RFIC
from Analog Devices. It covers 70 MHz – 6 GHz with up to 61 MHz sample rate and 56 MHz
analog bandwidth [22].
The wide bandwidth and frequency range make the BladeRF 2.0 micro an interesting
choice for satellite communications, both as a receiver and as a transmitter.
BladeRF products are supported through an open-source driver library, which provides a
simple and intuitive programming interface to the hardware. Firmware and FPGA sources are
avaialble as well [23].
On the hardware side, the schematics are available in PDF format on the website [24].
Figure 2.67: The BladeRF 2.0 micro transceiver.
58
Figure 2.68: BladeRF 2.0 micro block diagram.
A summary of the technical specifications for the BladeRF 2.0 micro is listed in Table 2.14
below.
59
Frequency Range 70 MHz – 6 GHz

Frontend filters None
RX paths 2
RX inputs 2
TX paths 2
TX outputs 2
ADC resolution 12
DAC resolution 12
Claimed dynamic range -
Claimed transmit power 8 dBm (CW)
Reference clock Factory calibrated VCTCXO
Other features Automatic IQ and DC offset correction
2×2 MIMO
Bias-T on both TX and RX ports
External clock through uFL connectors on the board
Temperature range 0-70 °C
Size 10.2 cm × 6.3 cm × 1.8 cm (board only)
11.1 cm × 7.3 cm × 2.4 cm (in acrylic case)
Weight 90 g (board only)
125 g (in acrylic case covered with copper tape)
Approximate price 480 USD (xA4)
720 USD (xA9)
Product page https://www.nuand.com/bladerf-2-0-micro/
Table 2.14: BladeRF 2.0 micro technical specifications.
The settings used during the BladeRF 2.0 micro noise figure measurements are listed in Table
Input rate Decimation Sample rate Bandwidth

2 MHz 4 500 kHz Auto
Table 2.15: BladeRF 2.0 micro settings used during noise figure measurements.
60
Figure 2.69: BladeRF 2.0 micro noise figure measurements.
We note that:
• There is in general good agreement between the noise figure calculated from the MDS
and the noise figure measured using the Y-factor method, except on 1.28 GHz.
• We consider the MDS measurements at 1.28 GHz to be unreliable because the device
was picking up a large amount of noise from somewhere while connected to the signal
generator.
• The noise level at 5.8 GHz was very sensitive to how the USB connector was posi-
tioned while mated.
• Although no RF official performance data is available, the measurements look as we
would expect it. Relatively good noise figure on VHF and UHF, and slowly increasing
noise figure with increasing frequency.
The settings used during the BladeRF 2.0 micor dynamic range measurements are listed in
Table 2.16 and the results are shown on Figure 2.70 below.
61

3 MHz 1 3 MHz Auto
Table 2.16: BladeRF2.0 micro settings used for BDR measurements.
We note that:
is decreasing as the gain is increased. The relationship is mostly linear.
on some frequencies at the highest gains.
• There is a large (30 dB) variation in the dynamic range as we go from 145 MHz to 5.8
GHz.
62
Figure 2.70: BladeRF 2.0 micro blocking dynamic range measurements.
The receiver spectra of the BladeRF 2.0 micro receiver are shown on Figures 2.71 to 2.81 be-
low. As we can see from the spectra, there are some spurs generated by or picked up by the de-
vice when connected to 50 ohm terminator. We can also note that the device is sensitive to ex-
ternal noise on 145 MHz and 2.4 GHz. This is not surprising as the device comes with an
acrylic enclosure.
63
Figure 2.71: BladeRF 2.0 micro receiver spectrum between 145 – 146 MHz.
64
65
66
67
68
The transmitter spectra for the BladeRF 2.0 micro transmitting a CW signal are shown on Fig-
ures 2.82 to 2.93 below. For each frequency, two spectra are shown, one at a gain setting
where the spectrum is clean only showing the CW signal, and one at a higher gain setting
where spurious artifacts appear on the spectrum.
69
Figure 2.82: BladeRF 2.0 micro transmitter spectrum at 145.5 MHz, gain 0.
70
71
Figure 2.86: BladeRF 2.0 micro transmitter spectrum at 1280 MHz, gain 0.
72
73
74
75
The measured transmitter output for the BladeRF 2.0 micro is shown on Figure 2.94 below.
As we can see there is a linear relationship between the transmitter gain setting and the output
power.
Figure 2.94: BladeRF 2.0 micro CW transmitter output power as function of gain.
BladeRF 2.0 micro are shown in Table 2.17 below. As we noted in test procedure description
in section 2.1.6, we believe that the NsMAR parameter is a margin above the required E S/N0,
which is 11 dB for the modulation and FEC settings used in our tests. Also note that the satu-
ration in NsMAR and MER is due to the limitation of the signal analyzer and does not neces-
sarily represent a limitation in the SDR device.
0 -66.1 2.5 14.8
10 -56.6 11.5 24.1
20 -50.3 12.7 25.0
30 -37.4 12.7 25.0
76

40 -27.0 12.7 25.0
50 -17.9 12.7 25.0
Table 2.17: BladeRF 2.0 micro transmitter MER as function of gain.
77
2.2.6 Ettus USRP B210
The The USRP B210 from Ettus Research is a 2×2 MIMO transceiver based on the AD9361
integrated RFIC from Analog Devices. It covers 70 MHz – 6 GHz with up to 61 MHz sample
rate and 56 MHz analog bandwidth [25].
The wide bandwidth and frequency range make the USRP B210 an interesting choice for
satellite communications, both as a receiver and as a transmitter.
USRP products are supported through an open-source Universal Hardware Driver (UHD)
library, which provides a relatively simple and intuitive programming interface to the hard-
ware. Firmware and FPGA sources are available as well [26].
On the hardware side, the schematics are available in PDF format on the website [27].
Figure 2.95: The Ettus USRP B210 transceiver.
78
Figure 2.96: USRP B210 block diagram.
A summary of the technical specifications for the Ettus USRP B210 is listed in Table 2.18
below.
79
Frequency Range 70 MHz – 6 GHz

RX paths 2
RX inputs 2
TX paths 2
TX outputs 2
ADC resolution 12
DAC resolution 12
Claimed noise figure 8 dB
Claimed transmit power 10 dBm
Reference clock 2 ppm
Other features Automatic IQ and DC offset correction
2×2 MIMO
External clock
Optional GPSDO
Temperature range 25 °C [28]
Size 15.5 cm × 9.7 cm × 1.5 cm
Weight 350 g (board only)
Approximate price 1200 € (board only)
Product page https://www.ettus.com/all-products/ub210-kit/
Table 2.18: Ettus USRP B210 technical specifications.
The settings used during the USRP B210 noise figure measurements are listed in Table 2.19

Table 2.19: USRP B210 settings used during noise figure measurements.
80
Figure 2.97: USRP B210 noise figure measurements.
We note that:
• On 5.83 GHz, the USRP was picking up leaked signal from the signal generator mak-
ing MDS measurements on this frequency practically impossible at high gains. There-
fore, the MDS data for 5.83 GHz is not reliable.
• With the exception of 5.83 GHz, our noise figure measurements are in good agreement
with the B200 performance data available from Ettus Research [29].
The settings used during the USRP B210 dynamic range measurements are listed in Table

3 MHz 1 3 MHz Auto
Table 2.20: USRP B210 settings used for BDR measurements.
We note that:
81
GHz.
Figure 2.98: USRP B210 blocking dynamic range measurements.
82
The receiver spectra of the USRP B210 receiver are shown on Figures 2.99 to 2.109 below. As
when connected to 50 ohm terminator.
Figure 2.99: Ettus USRP B210 receiver spectrum between 145 – 146 MHz.
83
84
85
Figure 2.106: Ettus USRP B210 spectrum between 2400 – 2425 MHz.
86
87
The transmitter spectra for the USRP B210 transmitting a CW signal are shown on Figures
2.110 to 2.121 below. For each frequency, two spectra are shown, one at a gain setting where
the spectrum is clean only showing the CW signal, and one at a higher gain setting where spu-
rious artifacts appear on the spectrum.
88
Figure 2.110: Ettus USRP B210 transmitter spectrum at 145.5 MHz, gain 35.
89
90
Figure 2.114: Ettus USRP B210 transmitter spectrum at 1280 MHz, gain 35.
91
92
93
94
The measured transmitter output for the USRP B210 is shown on Figure 2.122 below. As we
can see there is a linear relationship between the transmitter gain setting and the output power.
Figure 2.122: Ettus USRP B210 CW transmitter output power as function of gain.
USRP B210 are shown in Table 2.21 below. As we noted in test procedure description in sec-
tion 2.1.6, we believe that the NsMAR parameter is a margin above the required E S/N0, which
is 11 dB for the modulation and FEC settings used in our tests. Also note that the saturation in
NsMAR and MER is due to the limitation of the signal analyzer and does not necessarily rep-
resent a limitation in the SDR device.

0 -73.1 2.9 15.4
10 -62.6 12.7 25.0
20 -53.2 12.7 25.0
30 -42.3 12.7 25.0
95

40 -33.3 12.7 25.0
50 -22.4 12.7 25.0
60 -14.5 12.7 25.0
Table 2.21: USRP B210 transmitter MER as function of gain.
96
2.2.7 PlutoSDR
The PlutoSDR (officially called ADALM-PLUTO Active Learning Module) is a low-cost

software defined radio transceiver from Analog Devices Inc., created to be used as a tool for
teaching RF and radio communications at all levels [30]. It is based on the Analog Devices
AD9363 integrated RFIC and can be used as a full-duplex transceiver between 325 – 3800
MHz.
By default, the PlutoSDR is set up to stream raw I/Q samples to/from a host computer
through a USB2 connection. However, the devices contains a ZYNQ processor running em-
bedded Linux, allowing advanced users and developer to run SDR applications on the device.
The PlutoSDR comes as an open-source hardware and software with schematics and PCB lay-
out files available through the Analog Devices website [32].
Figure 2.123: The Analog Devices PlutoSDR transceiver.
PlutoSDR devices are controlled via the Linux Industrial I/O Subsytem (IIO). Compared to
simple SDR-specific APIs, IIO provides a somehow obscure and unintuitive interface for con-
trolling SDRs and streaming samples. Clear and concise programming documentation is hard
to find. Mastering Google and reading the source code of existing applications are must-have
skills if one wants to write an application that interfaces to the PlutoSDR.
97
Figure 2.124: PlutoSDR block diagram.
A summary of the technical specifications for the PlutoSDR is listed in Table 2.22 below.
Frequency Range 325 – 3800 MHz (70 MHz – 6 GHz through bootloader config)
Sample rate 7.5 – 12 MSPS (depending on USB2 host [31])
RF bandwidth 20 MHz
RX paths 1
RX inputs 1
TX paths 1
TX outputs 1
ADC resolution 12
DAC resolution 12
Claimed transmit power -
Reference clock 25 ppm
Other features ZYNQ processor running embedded linux
Onboard flash
Open source design
Temperature range 10 – 40 °C
Size
Weight 127 g (in original case covered with copper tape)
Approximate price 140 €
Product page https://www.analog.com/en/design-center/evaluation-hardware-and-
software/evaluation-boards-kits/adalm-pluto.html
Table 2.22: PlutoSDR technical specifications.
98
The settings used during the PlutoSDR noise figure measurements are listed in Table 2.23 and
the results are shown on Figure 2.125 below.

Table 2.23: PlutoSDR settings used during noise figure measurements.
Figure 2.125: PlutoSDR noise figure measurements.
We note that:
• Although no performance data is available for the PlutoSDR, we notice similar noise
figure and trends as we measured for the USRP B210.
• We can not explain the large difference between the noise figured calculated from the
MDS and the noise figure measured using the Y-factor method on 5.8 GHz. In this
case, we trust the Y-factor measurement since they are more consistent with the trend
and expectation of higher NF at higher frequencies.
99
The settings used during the PlutoSDR dynamic range measurements are listed in Table 2.24

3 MHz 1 3 MHz Auto
Table 2.24: PlutoSDR settings used for BDR measurements.
We note that:
GHz.
100
Figure 2.126: PlutoSDR blocking dynamic range measurements.
The receiver spectra of the PlutoSDR receiver are shown on Figures 2.127 to 2.139 below. As
when connected to 50 ohm terminator. The only band where the PlutoSDR seems sensitive to
exaternal noise is on 145 MHz and 2.4 GHz (WiFi).
101
Figure 2.127: PlutoSDR receiver spectrum between 145 – 146 MHz.
Figure 2.128: PlutoSDR receiver spectrum between 435 - 438 MHz.
102
Figure
Figure2.130:
2.129:PlutoSDR
PlutoSDRreceiver
receiverspectrum
spectrumbetween
between1260
435 – 1265
438 MHz.
MHz.
103
104
105
106
107
The transmitter spectra for the PlutoSDR transmitting a CW signal are shown on Figures Fig-
ure 2.140 to Figure 2.152 below. For each frequency, two spectra are shown, one at a gain set-
ting where the spectrum is clean only showing the CW signal, and one at a higher gain setting
where spurious artifacts appear on the spectrum.
108
Figure 2.140: PlutoSDR transmitter spectrum at 145.5 MHz, power -40 dBm.
109
110
Figure 2.144: PlutoSDR transmitter spectrum at 1280 MHz, power -40 dBm.
111
112
113
114
Although the spectra on Figures 2.140 to 2.151 look nice and clean, we noticed several
spurs occurring when at a wider span. Figure 2.152 shows a wider spectrum using 20 MHz
span around the CW carrier. We believe that the first spur on the left side of the CW carrier is
the DC component, the second being an image, and the third being a mix product between the
carrier and its image. If this is so, these artifacts can be corrected through software.
Figure 2.152: PlutoSDR transmitter spectrum at 5830 MHz with 20 MHz span.
115
As mentioned in section 2.1.5, we were unable to get a useful CW signal out of the PlutoSDR
using GNU Radio and the gr-soapy interface. Instead, we used a new PlutoSDR software tool
called SATSAGEN, which turns the PlutoSDR into a tracking spectrum analyzer and a signal
generator. Unlike our GNU Radio scripts, the user is required to enter a desired power output
in dBm, so in this case we do not know the exact relationship between PlutoSDR gain setting
and output power.
The measured transmitter output for the PlutoSDR is shown on Figure 2.153 below. As we
can see there is a linear relationship between the SATSAGEN power setting setting and the
output power.
Figure 2.153: PlutoSDR output power as function of Satsagen power setting.
Since our GNU Radio and gr-soapy based DVB-S2 transmitter did not work with the Plu-
toSDR, we could not perform the transmitter MER measurements for this device.
116
2.2.8 Device Comparisons
2.2.8.1 Lowest Noise Figure
Table 2.25 below shows the lowest noise figure measured for each SDR device in each fre-
quency band. Although the lowest noise figure of an SDR device is often not relevant in prac-
tical applications because
comparing it to the noise figure of the frontend component used in the device might give an
indication of how well the RF front end is matched at the frequencies of interest:
• The RTL-SDR and Airspy Mini use the R820T2 tuner, which has a noise figure of 3.5
dB below 1002 MHz.
• The SDRplay RSPduo has filters and LNA in front of the tuner. Notch filters are
placed in front of the LNA while band pass filters are placed between the LNA and the
tuner. Both the FM and the DAB notch filters have higher insertion loss at 145 and
437 MHz than what we measured, so these filters are presumably off by default and
the measured noise figure is that of the LNA plus switching circuitry in front of the
LNA.
• The LimeSDR Mini is based on the LMS7002M RFIC which itself has 2 dB noise fig-
ure. The values we measured are far from it.
• The USRP B210, BladeRF 2.0 micro, and the PlutoSDR are all based on the AD936x
RFIC, which has 2 dB noise figure below 1 GHz and 3.8 dB at 5.5 GHz.
Lowest Noise Figure (dB)

Device 145 MHz 437 MHz 1.28 GHz 2.25 GHz 2.425 GHz 5.83 GHz
RTL-SDR 5.8 5.4 11.9 - - -
Airspy Mini 7.1 7.4 10.7 - - -
SDRplay RSPduo 2.1 2.5 5.4 - - -
LimeSDR Mini 11.1 13.8 15.6 - - -
BladeRF 2.0 micro 5.0 4.9 7.4 9.0 9.5 15.8
USRP B210 5.5 4.4 5.7 4.8 6.3 9.6
PlutoSDR 6.2 6.1 4.1 4.7 5.0 9.3
Table 2.25: Lowest measured noise figure.
117
2.2.8.2 Dynamic Range
A way to compare the dynamic range of the tested devices without having a figure of merit
such as the third order intercept point, it to plot the maximum input power the receiver can
tolerat vs. the noise figure. This gives a good indication of which device has the highest dy-
namic range at any given noise figure. Figures 2.154 to 2.156 below show this data for all de-
vices at 437 MHz.
Figure 2.154: Max input power vs. noise figure at 5 kHz separation.
118
Figure 2.155: Max input power vs. noise figure at 100 kHz separation.
Figure 2.156: Max input power vs. noise figure at 1 MHz separation.
119
2.2.8.3 Transmitter Power
Table 2.26 below shows the highest measured CW power transmitted by each SDR device in
each frequency band. Note that this is the highest power at maximum gain setting, not taking
into account any spurious emissions the devices might be emitting at this setting.
Output Power (dBm)

Device 145 MHz 437 MHz 1.28 GHz 2.25 GHz 2.425 GHz 5.83 GHz
LimeSDR Mini 16.7 15.9 13.2 7.9 - -
BladeRF 2.0 micro 7.7 6.9 5.0 1.6 0.1 -7.2
USRP B210 17.5 17.6 18.3 13.1 11.7 12.5
PlutoSDR 3.7 8.4 7.7 6.7 6.3 1.8
Table 2.26: Highest measured output power.
120
2.3 Test Setup Verification

Since our test setups and methods were relatively simple, we carried out several cross-checks
to ensure that our measurements are reliable:
1. We used two different methods to measure the noise figure: MDS and Y-factor.
2. We used two different signal generators and verified that they provide the same output
power.
3. We cross-checked the noise figure measurements using another SDR software (SDR
Radio V3).
2.3.1 Signal Generator Output Power Comparison
We used two different signal generators during the MDS measurement:

1. Marconi Instruments 2031 that can support measurements up to 2.7 GHz
2. Rohde & Schwartz SMF 100A generator that covers all frequency bands of interest
The Marconi 2031 is of older date with last calibration date in 2008 whereas the SMF 100A is
being calibrated at the prescribed intervals at a certified calibration facility. Although the Mar-
coni 2031 was only used for a short time period during the initial noise figure measurements,
we decided to compare the two generators against each other in order to ensure that we could
use them interchangeably, should that be necessary.
The comparison was carried out by connecting the output of the signal generator to a Ro-
hde & Schwartz FSP3 spectrum analyzer using the following settings:
• Span: 10 kHz
• Resolution bandwidth: 1 kHz
• Video bandwidth: 10 kHz (auto)
The generators were connected to the spectrum analyzer using a 141-24SMNM+ coaxial cable
from Mini Circuits with the insertion losses listed in Table 2.27 below.
121
Frequency Insertion Loss

145.5 MHz 0.09 dB
437.5 MHz 0.12 dB
1280 MHz 0.23 dB
2250 MHz 0.33 dB
2450 MHz 0.35 dB
Table 2.27: Coaxial cable insertion loss.

These losses are taken into account in the measurements shown in the tables below. The mea-
surements done at -100 dBm signal power are more uncertain than the others since this power
level is rather close to the noise floor of the spectrum analyzer.
Generator setting Measured Power (dBm) Difference

(dBm) R&S SMF 100A Marconi 2031 (dBm)
-100 -100.2 -100.1 0.10
-80 -80.29 -80.39 0.10
-60 -60.27 -60.41 0.14
-40 -40.27 -40.38 0.11
-20 -20.28 -20.39 0.11
Table 2.28: Signal generator output power comparison at 145.5 MHz.

-100 -100.3 -100.1 0.20
-80 -80.52 -80.62 0.10
-60 -60.55 -60.67 0.12
-40 -40.56 -40.68 0.12
-20 -20.56 -20.70 0.14
Table 2.29: Signal generator output power comparison at 437.5 MHz.
122

-100 -100.5 -101.2 0.70
-80 -80.83 -80.93 0.10
-60 -60.83 -61.04 0.21
-40 -40.82 -40.97 0.15
-20 -20.82 -20.97 0.15
Table 2.30: Signal generator output power comparison at 1280 MHz.

-100 -100.8 -101.3 0.50
-80 -81.03 -81.05 0.02
-60 -61.03 -61.11 0.08
-40 -41.07 -41.10 0.03
-20 -21.05 -21.08 0.03

-100 -100.8 -101.3 0.50
-80 -81.13 -81.15 0.02
-60 -61.12 -61.23 0.11
-40 -41.12 -41.21 0.09
-20 -21.13 -21.18 0.05
123
2.3.2 Test Software Verification
In order to verify that our test software used for noise figure and dynamic range measurements
works correctly, we cross-checked the noise figure measurements using SDR Radio V3. The
cross-checks were done with the Airspy Mini using the Y-factor method on 145.5 MHz and
the results are shown on Figure 2.157 below. As we can see from the graph, the two measure-
ment sets are virtually identical.
Figure 2.157: Noise figure measurements for cross-checking the test software.
124
3. On-Air Tests
3 On-Air Tests
Based on the technical specifications and the results obtained from the lab measurements, two
SDR devices were selected for further side-by-side performance testing over the air:
1. SDRplay RSPduo, a low cost dual-tuner receiver, which performed quite well during
the lab tests. The two independent tuners in the device makes it an appealing option
for receiving satellite telemetry simultaneously on two bands.
2. Ettus USRP B210, the most expensive device on our list with dual transceiver capabil-
ities.
3.1 Test Setup

The two devices were installed in a SatNOGS setup [33] at the AMSAT OZ premises. The de-
vices were configured to run as separate SatNOGS ground stations, namely station 1353
(USRP) [34] and station 1354 (SDRplay) [35], sharing the same antenna and receiving the
same satellite passes. This way the performance of the two SDR devices can be compared
based on the number of packets received and decoded during the test campaign.
Figure 3.1 shows a diagram of the on-air test setup. After installation, the stations were run-
ning for 24 hours during which the device settings were adjusted for best performance. Once
the optimal settings were found, the stations were left running unattended for a few days, just
like they would run in a regular SatNOGS ground station setup. The final settings that were
used during the the on-air tests are listed in Table 3.1 below.
Station 1353 Station 1354

USRP B210 SDRplay RSPduo
SATNOGS_RX_SAMP_RATE 2 MHz 2 MHz
SATNOGS_RX_BANDWIDTH 600 kHz 600 kHz
SATNOGS_RF_GAIN 50 34
SATNOGS_ANTENNA RX2 Tuner 1 50 ohm
Table 3.1: SDR device settings used during the on-air tests.
The ground stations were running the latest GNU Radio 3.8 based SatNOGS client soft-
ware available at the time of testing:
125
3. On-Air Tests
• satnogs-client git revision d9ff0e8509df12889d25bf4c1d8403c798d3ce04

• gr-satnogs git revision 0ba9e0e601f1fe561ba4d74f349bb45530fa58c7
• gr-soapy git revision a7399c5e7b586c90f971b47280ac92eb64349bc8
Figure 3.1: On-air test setup.
3.2 Test Results

The final on-air tests were run over a period of 48 hours beginning at 2020-03-22 00:00 UTC
until 2020-03-23 23:59 UTC. The results are listed in Table 3.2 below showing the satellite
name, the nominal downlink frequency, modulation, and the number of packets received by
the two ground stations during the 48 hour test period.
126
3. On-Air Tests
Satellite Frequency Modulation Packets Packets

MHz (USRP B210) (SDRplay RSPduo)
OPS-SAT 437.200 GMSK 9600 1483 1537
UNISAT-6 437.421 FSK 9600 2180 2271
TIGRISAT 435.001 FSK 9600 1010 999
CubeBel-1 436.990 FSK 9600 209 218
MCUBED-2 437.480 FSK 9600 134 139
BUGSAT-1 437.445 FSK 9600 187 191
FOX-1B 145.960 DUV 200 246 220
FOX-1D 145.880 DUV 200 192 181
Total 5641 5756
Table 3.2: Number of packets decoded during the on-air test campaign.
As can be seen from the results, the two receivers performed more or less equally, with the
SDRplay RSPduo having received about 2% more packets the the USRP B210. However, this
difference is rather low and would likely even out if the test campaign would be running over
a longer time period. It is still very impressive that in a SatNOGS setup, a low cost SDRplay
performs equally well as the higher cost USRP B210. On the other hand, we have to remem-
ber that the USRP is a wide-band dual transceiver with frequency coverage up to 6 GHz,
whereas the SDRplay cover only up to 2 GHz and has to some extent been optimized for re-
ception in the amateur radio bands.
127
4. Software Support
4 Software Support
In this chapter we give a brief overview of the most popular SDR software available and to
which extent they suport the SDR devices evaluated during this activity. We have divided the
SDR software into three categories:
1. End user applications – relevant to people who just want to use the SDR hardware with
basic functionality avaialble in common SDR software. In this category we have included:
• SDR-Radio V3
• SDR#
• Gqrx
• CubicSDR
• SDR Angel
2. SDR development frameworks – relevant to people who want to write SDR applications
in a convenient, high-level environment that does not require programming skills. In this cate-
gory we have included:
• GNU Radio
• Pothos SDR
3. SDR API wrappers – relevant to people who want to write SDR applications from the
ground up but without worrying about interfacing to different SDR device drivers. SDR API
wrappers provide a uniform programming interface to most commonly available SDR hard-
ware. In this category we have included:
• SoapySDR
• gr-soapy
• gr-osmosdr
128
4. Software Support
4.1 SDR-Radio V3
Software type: Receive, transmit, satellite tracking

Supported OS: Windows
License: Closed source commercial (free for amateur radio use)
Hardware support: RTL-SDR, Airspy, SDRplay, LimeSDR, USRP, BladeRF, PlutoSDR
Website: https://www.sdr-radio.com/
Description: SDR-Radio V3 is one of the most popular SDR applications for the Win-
dows operating system. It supports most of the SDR hardware available
on the market and is one of the few end user SDR applications with built-
in transmit support. In addition to providing all of the basic radio function-
ality needed, it also has built-in satellite tracking and Doppler tuning func-
tionality.
129
4. Software Support
4.2 SDR#
Software type: Receive

Supported OS: Windows
License: Closed source
Hardware support: RTL-SDR, Airspy, others might be supported via plug-ins
Website: https://airspy.com/download/
Description: SDR# is another very popular SDR application for Windows. One of its
core features is plug-in support allowing third parties to extend the func-
tionality of the application without having access to the source code. Offi-
cially SDR# only supports RTL-SDR and Airtspy from the list of tested de-
vices, support for other devices might be available throgh plug-ins.
SDR# does not have built-in satellite tracking functionality, however, plug-
ins exist to interface to third party satellite applications such as Gpredict.
130
4. Software Support
4.3 Gqrx

Supported OS: Linux, Mac OS X, Windows
License: Open source GPL v2
Hardware support: RTL-SDR, Airspy, LimeSDR, USRP, BladeRF, PlutoSDR
Website: https://gqrx.dk/
Description: Gqrx is a simple SDR application written in C++ using the Qt and GNU
Radio toolkits. Although it is primarily targeted to the Linux operating sys-
tem, binary packages are also available for Mac OS X and Windows.
Gqrx can interface to third party satellite tracking applications, such as
Gpredict, through its network based remote control control interface. Gqrx
uses the gr-osmosdr SDR driver wrapper for accessing SDR devices and
will therefore work with all devices supported by gr-osmosdr and
SoapySDR.
131
4. Software Support
4.4 CubicSDR

Supported OS: Linux, Windows, Mac OS X
Website: https://cubicsdr.com/
Description: CubicSDR is another popular, cross-platform SDR application with basic
spectrum view and demodulation functionality. CubicSDR uses the
SoapySDR SDR driver wrapper for accessing SDR devices and will there-
fore work with all devices supported by SoapySDR. We are not aware of
any built-ion functionality or external interface to provide satellite tracking
and Doppler tuning functionality.
132
4. Software Support
4.5 SDR Angel
Software type: Receive, transmit

Supported OS: Linux, Windows
Website: https://github.com/f4exb/sdrangel
Description: SDR Angel is another popular SDR application available for both Linux
and Windows. In addition to the basic spectrum view and demodulator
functionalities, it also provides more advanced SDR functions such as sig-
nal diagnostics, digital voice, video demodulation, and transmitting. SDR
Angel does not use any SDR device wrapper libraries, nontheless it has
built-in support for most of the commonly available SDR devices, including
the ones tested in this activity.
133
4. Software Support
4.6 GNU Radio
Software type: SDR development framework

Supported OS: Linux, Windows, Mac OS X
Website: https://www.gnuradio.org/
Description: GNU Radio is a very popular SDR development framework providing a
convenient graphical environment for creating SDR applications. It in-
cludes blocks for creating signal processing pipelines with or without a
graphical user interface, as well as composite blocks for more advanced
functionality such as digital television.
SDR hardware support in GNU Radio is available through both SDR
driver wrappers like gr-osmosdr and gr-soapy, as well as vendor specific
out of tree modules like gr-uhd, gr-limesdr, and gr-iio.
134
4. Software Support
4.7 Pothos SDR
Software type: SDR development framework

Supported OS: Linux, Windows
License: Open source, Boost Software License 1.0
Website: https://github.com/pothosware/PothosCore/wiki
Description: Pothos SDR os a graphical SDR development framework similar to GNU
Radio. In addition to providing the framework for creating SDR applica-
tions in a convenient graphical environment, Pothos supports execution of
applications in a distributed environment, including networked hosts and
hardware accelerators such as FPGAs and GPUs. The primary SDR
hardware support in Pothos is provided by SoapySDR, therefore it sup-
ports most of the commonly available SDR devices.
135
4. Software Support
4.8 SoapySDR
Software type: SDR device API

License: Open source, Boost Software License 1.0
Website: https://github.com/pothosware/SoapySDR/wiki
Description: SoapySDR provides a vendor-neutral interface to SDR devices available
through C, C++ and Python API. SDR device support in SoapySDR is pro-
vided through plugins making it easy to add new hardware support to any
application that uses SoapySDR as its device interface.
4.9 Gr-soapy

Website: https://gitlab.com/librespacefoundation/gr-soapy
Description: Gr-soapy is a GNU Radio out-of-tree module, making SDR hardware sup-
ported by SoapySDR available to GNU Radio applications. The initial ver-
sion of gr-soapy was developed as part of the SDR Makerspace project
and has since been maintained by Libre Space Foundation as part of the
SatNOGS client software ecosystem.
136
4. Software Support
4.10 Gr-osmosdr

Hardware support: RTL-SDR, Airspy, LimeSDR, USRP, BladeRF, PlutoSDR
Website: https://osmocom.org/projects/gr-osmosdr/wiki
Description: Gr-osmosdr is the Osmocom GNU Radio block prividing a uniform inter-
face to GNU Radio applications an thus has the same role as SoapySDR
+ gr-soapy. The most significant difference compared to SoapySDR + gr-
soapy is that SDR device support in gr-osmosdr is added at compile time
rather than runtime loadable plugins. This makes the framework somehow
more difficult to distribute to end users in that distribution packages must
be accompanied by the driver libraries in order to satisfy the dependen-
cies that have been added during compile time. Because of this and the
fact the SDRplay driver are closed source, most distributions of gr-os-
mosdr do not include SDRplay support.
137
5. Conclusion
5 Conclusion
We have evaluated seven SDR devices, the RTL-SDR V3 dongle, the Airspy Mini, the SDR-
play RSPduo, the LimeSDR Mini, the BladeRF 2.0 micro, the USRP B210, and the Plu-
toSDR. These devices were chosen based on the scope of the SDR Makerspace project,
namely satellite communications. During our evaluation, we have measured the receiver noise
figure, the receiver dynamic range, the transmitter output power, and achievable transmitter
modulation error ratio. The measurements were carried out under realistic conditions that are
representative for an RF environment in which these devices are typically used.
Our tests have provided us with key performance data that is not readily available for most
SDR devices under test. In fact, out of the seven tested devices only Ettus Research and SDR-
play have published detailed performance data for their devices. For these devices our mea-
surements are in good agreement with the published data.
We have defined some relatively simple test setups and test procedures for the receiver
tests that involve only a signal generator, then cross checked the measurements using a cali-
brated noise source. The procedures turned out to be sufficiently accurate for the purpose of
comparing the performance of wide band SDR devices. During the test campaign we have
also developed open-source test software to support the test procedures.
Two of the SDR devices under test, SDRplay RSPduo and USRP B210, have also been
tested on-air in a SatNOGS satellite telemetry receiver setup. Both devices performed very
well receiving virtually the same amount of packets.
Finally, we have taken a brief look at some of the most popular SDR software for Linux,
Windows, and Mac OS X in order to evaluate how well the tested SDR devices are supported
by existing software. All devices are well supported by most software with the exception of
SDRplay, which due its closed-source driver only has limited support in pre-packaged Linux
software.
138
6 Appendices
Appendix I: Insertion Losses
Frequency (MHz) Loss (dB)

145 0.03
435 0.06
1280 0.10
2250 0.14
2425 0.15
5830 0.46
Table 6.1: Insertion loss for BLK-89-S+ DC block.

145 19.65
435 19.67
1280 19.71
2250 19.74
2425 19.74
5830 19.78
Table 6.2: Insertion loss for BW-S20W2+ attenuator.

145 0.09
435 0.12
1280 0.23
2250 0.33
2425 0.35
5830 0.57
Table 6.3: Insertion loss for 141-24SMNM+ coaxial cable.
I
145 0.12
435 0.21
1280 0.43
2250 0.59
2425 0.62
5830 1.00
Table 6.4: Insertion loss for FL086-24SM+ coaxial cable.
II

Evaluation of SDR Boards-1.1

Uploaded by

Copyright:

Available Formats

Evaluation of SDR Boards-1.1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Evaluation of SDR Boards-1.1

Uploaded by

Copyright:

Available Formats

Evaluation of SDR Boards and Toolchains

Authors: Alexandru Csete

1.1 Purpose and Scope

1.2 Selected SDR devices

Device Type Frequency range Bandwidth Bits

Table 1.1: Selected SDR devices under test.

1.3 Evaluation Criteria

◦ Noise figure (NF) and minimum detectable signal (MDS)

1.4 Evaluation Methods

2 Evaluation of Technical Specifications

2.1 Test Setup and Procedures

2.1.1 Receiver Noise Figure

MDS=−174 dBm+ NF +10 log 10 ( B)+ 10 log 10 ( M )

Using M = 1, we can calculate the noise figure from the MDS:

NF =MDS+174 dBm−10 log 10 (B)

2.1.1.2 Test Procedure

MDS actual =MDS−IL DCBLK −IL ATT −IL cable

2.1.2 Receiver Dynamic Range

Figure 2.4: Disturbing signal causing spurs as far as 1 MHz away.

BDR [dB ]= Pi n−MDS

2.1.2.2 Test Procedure

Figure 2.5: Test setup for dynamic range measurements.

The blocking dynamic range is measured as follows:

Pactual =P i n−IL DCBLK −IL cable

The receiver gain compression, P1dB, is measured as follows:

2.1.3 Receiver Spectral Purity

Figure 2.6: Test setup for receiver spectral purity measurements.

2.1.4 Transmitter Spectral Purity

2.1.5 Transmitter Output Power

Pactual =P FSW + IL DCBLK + IL cable

where PFSW is the power measured by the spectrum analyzer.

2.1.6 Transmitter Modulation Error Ratio

Figure 2.8: Constellation with error [8].

Figure 2.9: Test setup for transmitter MER measurements.

Figure 2.10: Screen captures of the ROVER DVB-S/S2 analyzer.

2.1.7 Test Software

Figure 2.13: GNU Radio application used for MER measurements.

Figure 2.14: The SATSAGEN spectrum analyzer software for PlutoSDR.

2.2 Test Results

2.2.1 RTL-SDR Blog V3

Figure 2.15: The RTL-SDR blog V3 receiver.

Figure 2.16: RTL-SDR Blog V3 dongle block diagram.

Frequency Range 500 kHz – 1766 MHz

Table 2.1: RTL-SDR Blog V3 technical specifications.

2.2.1.1 Receiver Noise Figure

Input rate Decimation Sample rate

Table 2.2: RTL-SDR V3 settings used during noise figure measurements.

Figure 2.17: RTL-SDR V3 noise figure measurements.

2.2.1.2 Receiver Blocking Dynamic Range

Input rate Decimation Sample rate

Table 2.3: RTL-SDR V3 settings used during BDR measurements.

Figure 2.18: RTL-SDR V3 blocking dynamic range measurements.

2.2.1.3 Receiver Spectral Purity

Figure 2.19: RTL-SDR receiver spectrum between 145 – 146 MHz.

Figure 2.20: RTL-SDR receiver spectrum between 434 – 436 MHz.

Figure 2.21: RTL-SDR receiver spectrum between 436 – 438 MHz.

Figure 2.22: RTL-SDR receiver spectrum between 1260 – 1262 MHz.

Figure 2.23: RTL-SDR receiver spectrum between 1262 – 1264 MHz.

Figure 2.24: RTL-SDR receiver spectrum between 1264 – 1265 MHz.

Figure 2.25: RTL-SDR receiver spectrum between 1266 – 1268 MHz.