Review: Airspy vs. SDRplay RSP vs. HackRF

Real World Signal Reception Tests (VHF/UHF)

The methodology of the tests in this review was to tune to a known signal and adjust the gain settings until the best SNR was obtained on that signal. We then compared the SNR achievable by each SDR (sensitivity) and took note of any undesired effects such as interference and intermodulation effects (dynamic range). All tests were performed in SDR#. We understand that SDR# is considered as better for the Airspy, but when comparing results with the RSP on HDSDR and SDR# we saw no changes in signal performance, so decided to use SDR# for side by side comparisons. The only disadvantage we can see is that third party plugins are not available for the RSP.

For a better screenshot comparison we also reduced the visible bandwidth of each SDR down to 6.4 MHz. This is simply a crop and does not change the FFT resolution. The Airspy, RSP and HackRF were running at 10 MHz, 8 MHz and 8 MHz respectively, but the spectrum were zoomed in to 6.4 MHz. FFT resolution for each test was set at 32k.

One important thing to note is that we cannot easily compare SNR values if the bandwidth of each device is different. Reducing the visible bandwidth increases the FFT density which causes the visible SNR to rise. Doubling the bandwidth causes a 3dB drop in visible bandwidth. So with the same signal viewed at 10 MHz and 5 MHz, the SDR with the 5 MHz would show an SNR level 3dB higher than on the 10 MHz SDR. Therefore the difference in SNR in a stable signal when shown at 10 and 8 MHz bandwidths is 3 * (10 - 8) /5 = 1.2 dB. In all these tests that we run at 10 MHz and 8MHz bandwidths, we must reward the Airspy SNR readings by +1.2 dB to get a more accurate comparison. This is a small change and probably within our measurement margin of error, but it still should be noted.

We tested each SDR within 5 minutes of each other to ensure the signal conditions did not change too dramatically, but just in case we repeated the test several times to ensure that the relative SNR's where stable across the test time period. We could not test the units at the same time when connected to an antenna splitter because 1) Each of the SDR's output their own noise which interfered with the other units and 2) our PC was not powerful enough to run all three at once.

In all tests during this review we used a roof mounted Diamond D130 discone antenna. It was placed in a suburban area with good reception from several radio towers that were about 15-30 kms away. Some of the tests we performed include: 

  • No overload test. We optimised the gain settings to provide maximum SNR of a desired signal whilst not increasing too far as to cause images from overloading.
  • Overloading allowed test. In these tests we optimized the desired signal gain for max SNR on our desired signal, and we did not care if overloaded signals showed up as long as they did not affect the desired signal.
  • Test with BCFM filter. Since we had strong broadcsat FM in our area we repeated some of the overloading allowed tests with a BCFM trap in place.
  • Test with LNA. Here we tested with an external LNA. The LNA we used was the LNA4ALL which has an approximate 18-23dB gain at most frequencies.
  • Test with LNA and BCFM filter. Here we tested with an external LNA and a BCFM filter in front of it.
  • Test with LNA and high loss coax. Here we used more lossy coax so that the gain from the LNA was not so harsh.

75 MHz Police

Here we tried to listen to a police frequency at 75 MHz. There was strong broadcast FM interference at 88 - 108 MHz that would cause problems if we increased the gains too much.

Max SNR without Overloading Test

In this test we adjusted the gains to get the highest SNR without having interference that would wipe out the band.

Airspy RSP HackRF

Here the Airspy was able to obtain the best results, receiving with the highest SNR and with a fairly clean spectrum. Here Linearity mode worked the best, with the Sensitivity mode working poorly due to the broadcast FM interference.

The RSP was set the GR 62dB with the internal LNA turned off. It was able to receive well too, but with an SNR about 10dB lower than the Airspy and with more interference shown and some mild imaging across the center. Increasing the gain further caused the RSP to overload and display broadcast FM interference all over the band. The reason the RSP performed more poorly in this test is possibly because this 75 MHz frequency is covered in the same 60 - 120 MHz bandpass filter that covers the broadcast FM band. This means that the broadcast FM band is not blocked out at all while tuned to 75 MHz.

The HackRF performed the worst, overloading the easiest and having the lowest SNR, but the signal was still very easily copyable. Surprisingly even with interfering stations nearby turning on the front end amplifier improved the signal significantly without causing too much interference.

Below we show the type of interference that started to show up when the gains were increased too far. All three units showed similar interference.

Broadcast band interference showing on the RSP when the gain was increased too far.
Broadcast band interference showing on the RSP when the gain was increased too far.

Discussion

The results showed that the Airspy was able to easily receive the police signal the best, with the RSP coming in second. Increasing the gain too far caused overloading issues and the noise floor to rise. We think that the RSP could not perform as well as the Airspy since it relies more on adequate filtering to improve dynamic range than the Airspy does. Since the RSP filter active at 75 MHz is a 60 - 120 MHz bandpass filter, overloading effects from broadcast FM (BCFM) can easily be seen as it is not filtered out.

RSP Bandpass Filter for 60 - 120 MHz
RSP Bandpass Filter for 60 - 120 MHz

Broadcast Band FM (BCFM) Radio

Here we tested each radio on it's ability to receive broadcast band FM. Note that we offset the IF tuned area slightly in order to capture the peak and the noise floor bottom more accurately in SDR#'s SNR measurement tool.

Max SNR Test

Airspy RSP HackRF

The Airspy was able to receive best under Sensitivity mode with little to no interference or imaging, but it was consistently about 3-6 dB less sensitive that the RSP.

The RSP consistently received the best in this test, usually getting SNR values about 3-6 DB's higher than the Airspy. The best reception occurred at a GR 55 with the Mirics LNA turned off.

The HackRF received the poorest, consistently being having about 10-15 dB's or SNR less. With the front end amplifier turned on the band became overloaded.

LNA Test

We also tested BCFM reception with an external LNA connected. It was placed before an extra 3M length of RG174 coax cable.

Airspy RSP HackRF
The Airspy performed similarly to when used without an LNA. Here the RSP showed problems with overloading even when the GR was reduced to the very lowest level and the Mirics LNA was turned off.

The HackRF performed similarly to when used without an LNA.

LNA with higher loss cable test

We also did another test for the Airspy and RSP with the LNA connected to some much higher loss cable. To get higher loss we added in a 12dB attenuator as well as 10m of RG174 which has a loss of about 3dB at 100 MHz. The total loss is about 15dB and the gain from the LNA4ALL is about 22dB, so the total left over gain is 7dB.

Airspy RSP
The Airspy performed well with the LNA and lossy cable as in the last test. Mild interference was seen between strong stations. With less overall gain in front of it the RSP performed much better as was almost identical to the Airspy. There was higher noise levels from overloaded signals between the strong stations though.

Discussion

It was interesting to see in this test that the Airspy could not reach a maximum SNR as high as the RSP could. However, as predicted, the overall performance of the RSP was reduced too much when an external LNA was added, making it unable to receive properly even with the RSP gain set at its minimum value. Reducing the external gain through more loss helped stabilise the RSP results. It is likely that the LNA4ALL itself was overloaded by the BCFM band in these tests.

It was mentioned by Yousseff that a better metric for BCFM comparison is the MPX spectrum of the FM signal, so below we do some MPX comparisons.

BCFM MPX Comparison

Before we published this article we gave the draft to the creators of the Airspy and RSP for review. Yousseff, co-creator of the Airspy suggested that we look at the MPX spectrum of the BCFM signal as his concern was that the stronger SNR shown by the RSP was not accurate as its SNR may be higher due to non-linearities mixing into the actual signal.

To test the MPX spectrum we used HDSDR with the unofficial Airspy ExtIO plugin and the RSP with it's official ExtIO plugin. We piped the audio out through stereo mix and looked at it in SDR#. This allowed us to view the broadcast FM MPX spectrum, which allowed us to see things like the stereo pilot tone and RDS. We then measured the SNR of the stereo pilot tone.

92.6 MHz Strong Station

This was a strong classical music station. 

Airspy RSP
The Airspy had an SNR of 64.8+1.2=66dB in the frequency spectrum and an MPX SNR of 64.8dB. The RSP had an SNR of 71.5dB in the frequency spectrum and an MPX SNR of 65.7dB. 

Overall the MPX SNR was nearly identical between the two SDR's. We also checked other similarly strong stations and all gave similar results, with the RSP showing higher SNR on the frequency spectrum, but both showing near identical MPX SNR's.

87.6 MHz Weak Station

At 87.6 MHz there was a weak station and we repeated the MPX test. Here we note that when we pushed the spectrum to the side the Airspy performed better and was able to reach a frequency spectrum SNR nearby identical to the RSP.

Airspy RSP
Here the Airspy had an SNR of 29.4+1.2=30.6 dB in the frequency spectrum and an MPX SNR of 45 dB The RSP had an SNR of 30 dB in the frequency spectrum and an MPX SNR of 44.4 dB.

104.2 MHz Weak Station

At 104.2 MHz there was another low power FM station.

Airspy RSP
Here the Airspy has an SNR of 31.2+1.2 = 32.4dB in the frequency spectrum and an MPX SNR of 43.5 dB. The RSP had an SNR of 41.6 dB in the frequency spectrum and an MPX SNR of 44.5 dB.

Discussion

When comparing the MPX spectrum on both strong and weak stations we found that although the RSP showed a higher SNR in the frequency spectrum, the SNR in the MPX spectrum appears to be nearly identical to the Airspy's.

150 MHz

Here at 150 MHz there are some taxi radio signals and also some trunking signals. Since this frequency is very close to a powerful pager at 157 MHz, and also still close to the strong broadcast FM band, it is easy to see effects from overloading.

Max SNR with no Overloading Allowed

In this first test we attempted to maximise the SNR on all three radios, whilst ensuring that the spectrum was kept clean of any overloading effects.

Airspy RSP HackRF

Here the Airspy was able to receive about 10 dB better than the RSP. We note that the Sensitivity and Linearity tuning modes did not work well at all. Here we need to manually tune each gain setting to get the best results.

With the RSP we could not increase the gain further than [GR60 Mirics LNA off] before overloading effects were seen.

The HackRF performed the worst. It was able to receive best with the front end amplifier turned on.

Max SNR with Overloading Allowed

In the next test we tried to maximise SNR by adjusting the gains without regard to the overloading effects - as long as these effects did not completely wipe out the signal. 

Airspy RSP HackRF

In the previous no interference test the Airspy was already near the limit of maximum SNR. Increasing the gain further started to cause the noise floor to rise and cause the SNR to drop. We were only able to increase the gain a little further. By increasing the gain a little more we started to see mild broadcast FM band interference that would only show up when the pager transmitted, as well as some other imaging effects.

The RSP was able to reach a SNR almost as good as the Airspy's after the gain was increased to [GR36 Mirics LNA off], but it came at the expense of seeing some bad broadcast FM band interference show up whenever the pager transmitted.

We were also able to increase the HackRF's SNR by increasing the LNA gain further, but at the expense of some pretty heavy interference of the same type as seen on the RSP.

Max SNR with BCFM Filter

In this next test we placed a broadcast FM block filter in front of each SDR and observed the effect. We attempted to tune for max SNR regardless of any overloading effects, as long as they did not affect the signal of interest.

Airspy RSP HackRF

With the Airspy and broadcast FM block filter we were able to pump up the gains a little further and increase the SNR by about 1 dB over the last test. The mild broadcast band interference that appeared when the pager transmitted no longer showed up and we saw no other effects from the nearby pager.

With the BCFM blocking filter in place the RSP was able to achieve a 3dB+ increase in SNR over the previous test, and without any BCFM overloading effects shown. However, one important problem was that when two pager frequencies transmitted at the same time, then there was very bad interference that affected our signals of interest (see the bottom of the image). We were unable to get rid of this interference for all gain settings, so we just had to ignore it and tune for best SNR. There was also images of the nearby pager that showed at 150, 151, 151.5, 152.5 and 154 MHz.

Oddly, the HackRF SNR levels performed worse with the BCFM filter in place, but the BCFM interference was nullified. We are unsure of the reason why.

LNA Test

Here we connected an LNA in front of 6M of RG174 cable and observed the results.

Airspy RSP HackRF
With an LNA in front broadcast FM interference was increased a lot. The max SNR was now only 59.7+1.2=60.9dB. Here the RSP had similar interference problems to the Airspy and when it was run without the BCFM filter. It reached a max SNR of around 56.4dB.

The HackRF's performance did not change too much with the LNA. There was still lots of interference.

LNA with BCFM Filter

This time we connected a BCFM filter in front of the LNA and retested.

Airspy RSP HackRF
With the BCFM interference removed the Airspy performed well once again, with only mild pager interference showing up. The RSP still had the same problem with the pager interference mixing into the desired signals. The HackRF performed better with the BCFM filter in place, but there was still significant interference whenever the pager transmitted.

LNA with high loss cable

The RSP team suggested that an LNA should only be used to overcome the cable loss. So here we added a 12dB attenuator and 10M of RG174 cabling which gives a 3.5dB loss to get a total loss of about 15.5 dB at 150 MHz. The LNA4ALL has about 23.5dB gain at 150 MHz, giving a total extra gain of 8 dB.

Airspy RSP
Like in the other tests the Airspy saw significant FM interference, though it was less than with higher external gain used. Similarly the RSP still had high FM interference present, much higher than the Airspy. At this frequency cable losses are low and the total extra gain of 8dB was probably still too high for the RSP to work well.

LNA with high loss cable and BCFM Filter

In this test we added a BCFM filter in front of the LNA.

Airspy RSP
With the BCFM filter in place the Airspy performed very well. With the BCFM filter and lower external gain the RSP no longer seemed to show the problem with the pager signal mixing into the desired signal. However, there was still pager images visible in the spectrum.

Discussion

Here at 150 MHz there is strong BCFM signals at 88-108 MHz and a strong pager at 156 MHz. In this environment the Airspy won out as the best receiver. The Airspy was able to get the best SNR for our target signal without any significant interference from overloading. The RSP struggled with overloading interference from BCFM, and even when a BCFM block filter was used it had problems with pager interference destroying our signal of interest. By looking at the bandpass filter used on the RSP when tuned to 150 MHz, we can see that the BCFM band attenuation may not be strong enough to help. Also it does not block out the 156 MHz pager.

With an LNA used all receivers had issues with BCFM interference, and it is likely that the LNA4ALL was overloaded. Adding a BCFM block filter improved things for the Airspy, but although BCFM interference for the RSP and HackRF was reduced, pager interference which was not blocked out now strongly affected the RSP and HackRF.

The 120 - 250 MHz RSP filter active when tuned to 150 MHz.
The 120 - 250 MHz RSP filter active when tuned to 150 MHz.

Marine And Pager Signals

In our next test we moved up to 161 MHz and tried to receive a fairly weak marine weather station (NOAA weather equivalent). This station is very close to the 157 MHz pager, which can easily wipe out the marine signal if the RF gain is set too high.

Max SNR with no Overloading

In this first test we tried to maximise the SNR of the marine signal whilst not allowing overloading effects to occur across the spectrum.

Airspy RSP HackRF

Here the Airspy had the best SNR of around 17.4+1.2 = 18.6 dB, however there was some mild overloading effects present just next to the pager signal which would not go away, as well as a very mild image near 162 MHz.

The RSP was able to reach an SNR of around 15.4 dB without much interference showing up, although the spectrum was not as flat as the Airspy's.

The HackRF had the lowest SNR of 12 dB, and had some bad overloading effects show up near the pager signal. Some FM interference also showed but we did not reduce the gain further as it was only mild.

Max SNR with overloading

In the next test we tried to maximise the SNR without regard to the effects of overloading - as long as it did not wipe out the marine weather signal.

Airspy RSP HackRF

Here the Airspy was able to reach a maximum gain of 28.5+1.2=29.7 dB's and only pager intereference was seen, none from the BCFM band.

The RSP showed some pretty heavy BCFM and pager interference but luckily the marine signal was not too affected by this as it was not underneath any interference. It was able to reach a maximum SNR of around 32.1 db, and was consistently about 2 dB higher than the Airspy.

The HackRF could only get a max SNR of around 22.7 dB.

Max SNR with overloading and BCFM filter

In this next test we added a broadcast FM band filter in front of the LNA and tuned for best SNR regardless of the overloading effects shown.

Airspy RSP HackRF

Here the Airspy produced results very similar to without the BCFM filter. It appears that the Airspy was not affected by BCFM interference in this test, only by the strong pager.

The SDRplay on the other hand showed a good improvement in terms of a reduction in BCFM interference from overloading. With the BCFM filter in place only pager interference shows.

Similarly the HackRF showed a good improvement with no more BCFM interference shown.

Max SNR with LNA

Here we added a 6M length of RG174 and placed an LNA4ALL at its beginning to try and overcome the loss.

Airspy RSP HackRF

Here the Airspy performed similarly to when used without a LNA.

The RSP used a gain setting of [GR50 Mirics LNA off] and couldn't achieve and SNR as good as without the LNA. Increasing the gain further caused very bad overloading. Overall performance was similar to the Airspy.

The HackRF also showed interference very similar to without the LNA.

Max SNR with LNA and BCFM Filter

Here we placed a BCFM filter in front of the LNA.

Airspy RSP HackRF

The Airspy performed very similarly in terms of SNR as when used without the BCFM filter, but the nature of the pager interference changed. It became more clumped up nearer to the actual pager frequency.

The RSP seemed to work much worse with the BCFM filter in place. The SNR couldn't reach as high as when used without it. The gain used was [GR62 Mirics LNA off] and it was optimized for the best SNR. The BCFM filter also caused an odd rise in the noise floor around the pager.

With the HackRF the BCFM interference was removed, but heavy pager interference remained. We weren't able to pump up the gain as much as when used without the BCFM filter, but we could achieve a higher SNR with the filter in place.

Max SNR with LNA high loss cable and BCFM Filter

Airspy RSP
Here the Airspy reached a maximum SNR of around 23.8+1.2=25 dB. The RSP reached a slightly higher maximum SNR of 27.5 dB, but had a bit more pager interference. The lower external gain seemed to reduce the problems seen in the last LNA test.

Discussion

Again in this test the Airspy showed that it was able to receive well without any effects of overloading showing up. The Airspy was pretty immune to the effects of BCFM overloading at this frequency, but still had trouble with the pager overloading. The RSP had issues with BCFM and Pager overloading.

Adding a BCFM block filter had no affect on the Airspy which was already unaffected by BCFM interference. On the RSP it removed the BCFM interference but caused the pager interference to have more effect. 

Adding an LNA did not help much with such a strong pager nearby causing overloading to occur at much lower gain levels on all three SDRs. Placing a BCFM filter in front of the LNA had almost no affect on the Airspy once again. Interestingly the BCFM + LNA combo with the RSP caused a big increase in overloading effects, the reason why we are unsure. When less external gain was used due to the high loss cable the RSP's performance improved.

At 162 MHz, as with the last test at 150 MHz, the RSP used  its 120-250 MHz bandpass filter which may not have been sufficient to fully block out BCFM and of course it could not block out the pager.

463 MHz

At 463 MHz we have a business band full of various types of trunked radios as well as some telemetry signals.

Max SNR with no Overloading

Here we tried to maximise the SNR of a target signal without any overloading showing up.

Airspy RSP HackRF

Here the Airspy performed the best, reaching an SNR value of 46.3+1.2 = 47.5 dB. To get good results we had to use the Free mode and carefully tune the gains manually. The sensitivity and linearity modes could not reach a SNR value anywhere near as good at 46 dB.

The RSP was able to reach a SNR of 38.5 dB. Increasing the gain further caused BCFM interference to show. The gain that gave the best SNR without overloading was GR66 with the Mirics LNA turned on. Mild BCFM interference was showing at this gain level.

The HackRF could only reach a max SNR of 34.7 dB.

In all three tests increasing the gain further than those used caused a decrease in SNR from overloading, so the max SNR with overloading allowed test was skipped.

Max SNR with BCFM filter

Here we tried to maximise SNR of the target signal with a BCFM block filter in place.

Airspy RSP HackRF

With the Airspy using the BCFM block filter we were now able to get an SNR up 49.4+1.2 = 50.6 dB. We were also able to use the sensitivity gain setting which produced the best SNR result. There was mild interference from the strong pager shown at around 460.5 MHz.

The RSP was now able to reach an SNR similar to that of the Airspy, at 49.7 dB's. The max SNR was obtained at GR51 with the Mirics LNA turned on. However, there appears to be some halo effect on the signals which may be phase noise or non-linear mixing.

The HackRF was now able to reach an SNR of 44 dB's. Some pager interference was seen below 462 MHz whenever it transmitted.

Max SNR with LNA

In this next test we placed an extra 6M of RG174 coax cabling, and an LNA4ALL before that to try and mitigate the losses.

Airspy RSP HackRF

Here the max SNR that could be obtained with the Airspy was only 32.6+1.2 = 33.8 dB, and there was some pretty decent BCFM interference that showed up at all gain levels.

The RSP reached a similar gain of 31.1 dB and also had BCFM interference showing up. Its gain was set to [GR62, Mirics LNA off].

Surprisingly the HackRF worked the best here, reaching a gain of 34.2 dB's with very little BCFM interference shown.

Max SNR with LNA and BCFM Filter

In the next test we added a broadcast FM filter in front of the LNA4ALL.

Airspy RSP HackRF

The Airspy was now able to reach a SNR of 52.4 dB, with no BCFM interference shown. Only mild pager interference was seen at 460.5 MHz.

The RSP had some trouble receiving with the BCFM filter in place. Even at the lowest gain setting severe pager intermodulation occurred whenever the pager transmitted. The gain in the image was set to the lowest possible, [GR78, Mirics LNA off].

The HackRF also could not reach a higher SNR with the BCFM filter in place. Increasing the gain any further caused full band overloading.

Max SNR with LNA and BCFM Filter and High Loss Coax

Here we used longer coax cable and a 12 dB attenuator to simulate a long cable run. At 463 MHz the total loss in this test is about 12 dB from the attenuator plus 6.828 dB loss from 10M of RG174, giving a total loss of 18.828 dBs. At 463 MHz the LNA4all gives about 23.5 dB of gain, so the total external gain is 23.5-18.828 = 4.672 dB's of gain remaining.

Airspy RSP
With the lower external gain the Airspy performed much better, with almost no external interference visible. The RSP still had the same problems with BCFM interference as with the lower loss cable test.

Max SNR with LNA and BCFM Filter and High Loss Coax and BCFM Filter

Airspy RSP
In this test the Airspy had an SNR of about 51.2 + 1.2 = 52.4 dB. No interference or overloading was noticed. The RSP reached an SNR of 50.6 dB. Unlike the test with low loss coax, no interference or overloading was noticed.

Discussion

Once again the Airspy was the best receiver as it was able to reach higher SNR values without overloading effects showing up. When we added a BCFM block filter in front of the receivers the max reachable SNR levels became nearly identical, though the RSP showed a problem that looked something like phase noise to us. The fact that the BCFM block filter improved results on the RSP seems to indicate that its internal 420 - 1000 MHz bandpass filter did not completely block out the interfering BCFM signals.

Interestingly adding an LNA into the system caused large amount of BCFM interference to show up on both the Airspy and RSP causing issues by making the max SNR obtainable much lower. Adding a BCFM filter in front of the LNA solved this overloading problem completely on the Airspy. On the RSP the addition of the BCFM filter solved the BCFM overloading problem, but instead we got stronger pager interference which was not seen on the Airspy. The interference was seen even with the RSP set to the lowest gain levels.

When using higher loss coax after the LNA the Airspy performed well with just the LNA, but the RSP still had BCFM issues. After adding the BCFM filter in before the LNA both the Airspy and RSP performed well.

420 to 1000 MHz Bandpass filter on the RSP.
420 to 1000 MHz Bandpass filter on the RSP.

860 MHz

This band again is a business band which several TETRA wide signals around.

Max SNR without Overloading

Here we tried to receive TETRA signals at 860 MHz. We adjusted the gains until we obtained maximum SNR on a target signal whilst stopping before overloading occurred.

Airspy RSP HackRF
Here the Airspy reached an SNR of about 31.5+1.2 =32.7 dB's and was about 6 dB less sensitive compared to the RSP. No matter what gain settings we applied we could not get it to match the SNR of the RSP.

The RSP had the highest SNR and reached a max SNR of 38.8 dB. The gain used for the max SNR was [GR 52 Mirics LNA on].

The HackRF was able to reach a SNR of 30.3 dB's.

Increasing the gains further simply raised the noise floor and lowered SNR, so we did not perform the max SNR with overloading allowed tests. Although the RSP was getting higher SNR results we are unsure if this is accurate as there is some odd rise in the noise floor on the signals. It looks like phase noise or perhaps non-linear mixing might be pushing the measured SNR up, but we could not confirm as we could not find a TETRA decoder that would show the bit error rates.

Max SNR with BCFM Filter

Next we tried with the BCFM filter in place. 

Airspy RSP HackRF

With the BCFM filter in place we were able to push the gain a little higher before full band overloading occurred, and get a higher SNR on our target signal of 35.1+1.2 = 36.3 dB. Increasing the gain further caused pager interference to show up, and mild pager interference can be seen on this Airspy screenshot already (mild rise on noise floor when the pager transmitted).

The RSP remained at a similar SNR level of 39.6 dB, about 1 dB higher than without the filter but within variance tolerances. However, with the filter in place effects that look like phase noise or non-linear mixing started to show up on most signals.

The HackRF saw poorer performance with the BCFM filter in place, with the target signal max SNR being reduced to around 26.1 dB.

Since the Airspy's SNR improved with this filter we can conclude that the Airspy was being desensitized slightly by BCFM interference.

Max SNR with LNA

Here we added an LNA and 5M of RG174 and recorded the results.

Airspy RSP HackRF

With the LNA in place the Airspy SNR was boosted up to 36.3+1.2 = 37.5 dB, just about matching the SNR of the RSP without the LNA.

The RSP's SNR was reduced to 37.8 dB with a gain setting of GR46 and the internal Mirics LNA turned off.

Interestingly the HackRF was able to reach the highest SNR now with an SNR of 38.7 dB.

Max SNR with LNA and BCFM Filter Before LNA

Here we placed a BCFM filter before the LNA.

Airspy RSP HackRF

For the Airspy placing the BCFM filter before the LNA significantly worsened reception. The max SNR was now only 30.5+1.2 = 31.7 dB and pager interference started showing up in the form of noise floor rises whenever it transmitted.

Similarly, with the RSP reception was worsened with the BCFM filter in place. Now pager interference showed up strongly whenever it transmitted and the max SNR was significantly reduced to 18.7 dB. Gain settings where GR65 internal Mirics LNA off.

The HackRF saw similar problems with the max SNR reducing to 21.5 dB and massive noise floor rises whenever the pager transmitted.

We're not sure why placing the filter before the LNA caused trouble in this test.

Max SNR with LNA and BCFM Filter After LNA

Here we decided to see how reception was affected if we placed the BCFM filter after the external LNA.

Airspy RSP HackRF

The Airspy worked much better with the BCFM filter placed after the LNA. Now the SNR was increased to around 34.6 + 1.2 = 35.8 dB.

Similarly the RSP saw an improvement and it's SNR was increased to 38.9 dB. The gains used were GR36 internal Mirics LNA off.

The HackRF was also improved getting a max SNR of 29.6 dB.

Max SNR with LNA and high loss cable

Here we used longer coax cable and a 12 dB attenuator to simulate a long cable run. At 858 MHz with a 12 dB attenuator and 10 dB of loss from 10m of RG174 cable we have a total loss of 22 dB. The LNA4ALL has a gain of about 20 dB at 858 MHz. Thus there is a total of 22 - 20 = 2dB of gain left over.

Airspy RSP
The Airspy reached an SNR of 23.2+1.2=24.4dB and showed no signs of any sort of interference. The RSP only reached a maximum SNR of 17.8 dB. Unlike in the other tests the RSP started showing signs of BCFM interference with this longer cable for some reason.

Max SNR with LNA and high cable and BCFM filter

In this test we added the BCFM filter before the LNA4ALL.

Airspy RSP
The Airspy now reached a maximum SNR of 28.7+1.2 = 29.9 dB. The spectrum was clean.  The RSP now reached a maximum SNR of 33.8 dB and the spectrum was also clean.

Discussion

At these frequencies the RSP seemed to be more sensitive and appeared to be much less affected by BCFM and pager interference, and so this time the RSP was able to reach higher SNR values. By adding a BCFM filter in front of the receivers the Airspy's max reachable SNR became closer to that of the RSP's, but there was still about a 3dB difference. It seems that at this frequency the Airspy was more affected by the BCFM band than the RSP was.

With an LNA added the maximum SNRs reachable by the Airspy and RSP were both reduced significantly as overloading occurred much easier, but both SDRs were almost identical in performance. When we added a BCFM filter in front of the LNA, reception on both receivers worsened a lot, especially on the RSP. When we placed the BCFM filter after the LNA reception improved, and the RSP was able to reach signal levels about 3 dB higher than the Airspy once again. We're unsure of why this would happen.

With higher loss cable used after the LNA the RSP suffered some BCFM interference and again we are unsure why. The Airspy performed normally, but both suffered heavy loss from the cable. We think that the estimated cable loss was perhaps about 10dB higher than estimated at this frequency. With the BCFM filter placed before the LNA4ALL both the Airspy and RSP performed similarly.

The same built in bandpass filter used in the RSP as with the 460 MHz test is used in this 860 MHz test so it is interesting to see that this frequency is much less affected by BCFM overloading than the 460 MHz band is. 

L-Band Test

Here we tested the three SDR's on their ability to receive L-band satellites such as Inmarsat with an L-band patch antenna. 

Max SNR Test

Airspy RSP HackRF

The Airspy was able to reach a max SNR of about 7.4+1.2 = 8.6 dB.

The RSP was able to reach a max SNR of about 8.1 dB. It's looks a bit lower due to the signal jumping up and down when taking the screenshot, but overall the RSP seemed to be about ~1 dB more sensitive than the Airspy at L-band.

The HackRF also received decently with an SNR at 8 dB, but some interference was seen.

Max SNR with LNA

With an LNA attached just before the receiver, without any extra run of coax cable.

Airspy RSP HackRF

The Airspy was able to reach a higher SNR of 13.9+1.2 = 15.1 dB.

Strangely, with an LNA connected the RSP was not able to receive L-band signals at all. This could be due to possible overload from BCFM or DVB-T signals.

The HackRF had its SNR boosted to 9.7 dB, but there was interference seen from what looks like other L-band signals, a DVB-T signal, and perhaps various trunking signals from 460 MHz.

Discussion

The RSP again showed that it has good sensitivity that is on par or slightly better than the Airspy at this frequency. However, when an LNA was added the RSP could no longer receive L-band signals at all! We guess that the RSP may have been overloaded by strong DVB-T signals at 500 MHz. We guess this because it looks like we can see the edge of a DVB-T signal ghosting in the waterfall image.

L-Band Test 2

Because of the poor results the RSP had with the LNA we decided to recheck results at another location with very weak BCFM and other terrestrial signals like DVB-T so that the RSP would not overload.

Max SNR Test

Airspy RSP HackRF

Without the LNA the Airspy has a SNR of about 3.5+1.2 = 4.7 dB.

The RSP had an SNR of 4.6 dB.

The HackRF had an SNR of 2.8 dB.

Max SNR with LNA

Airspy RSP HackRF

With the LNA connected the Airspy once again showed a significantly increased SNR, now showing 7.1+1.2 = 8.3 dB.

This time with very little interference present the RSP worked fine with the LNA and got 6.3 dB.

The HackRFwas able to reach a higher SNR of 6.4 dB.

Discussion

In this location without strong earth based signals the RSP worked fine with the LNA in place and received the expected boost in SNR. From these results it appears that without an external LNA the RSP works a little better, but with an external LNA signal on both SDRs is improved, but better on the Airspy.

HF Tests and conclusions continued on the next page.

Next Page ->

Subscribe
Notify of
guest

78 Comments
Inline Feedbacks
View all comments
Gareth

Great forum that I wished I had discovered earlier, better late than never !

Enjoyable product review but is there an SDR on the market that offers Tx and the wide bandwidth of the HackRF but with the performance quality of the AirSpy, please ?

If I had multi thousand livres available I would love to own a Yaesu or similar piece of kit but as I don’t have that kind of loose change and radio Ham being new to me even though I have had a lifelong interest only now being realised I feel that SDR is the way to go until I am positive that this is for me.

Thank you in anticipation of your reply.

Gareth

NLA

I know this center spike in ZeroIF is by design either in the SDRplay API or the hardware. This is not an issue with my Airspy Mini.

NLA

SDR++ v1.0.4 was released today allowing Low IF mode which got rid of the center spike issue with my RSP units. Working good so far using “LowIF 2048Khz, IFBW 5000KHz”. I wish the spike was not present using ZeroIF mode though.

nickels

hackrf is deaf

Croc

I find sample rate to be a negligible feature which I don’t tend to consider when purchasing software defined radios. The reason behind this simple; firstly, my computers aren’t the fastest in the world and thus can not process the amount of data that larger sample rates incur. Secondly, I like to easily visualize strong signals across the visible spectrum. I find it’s easier to home in on signals with a 2 mega-samples per second rate. The only exception to the rule is while homing in on WFM signals, the signal width counteracts larger sized sample rate.

With regards to RSP1A versus HackRF One; It’s like comparing a Ferrari to a Porsche. They are two very good SDRs and they perform very well. On one hand, the HackRF One supports the Portapack H1 add-on, which now can be loaded with the Mayhem firmware. I find this configuration to be extremely portable, depending on antenna choice.

To put it in more simple terms… I think the SDRPlay performs well as my base station setup while my HackRF One performs well in mobile situations. To anyone considering purchasing either of these, please consider the cost. I think the SDRPlay has far better value for money.

Dan

Excellent review, thank you. This is probably the best comparison between SDRplay and AirSpy on the web. It would nice if you could revisit this article and update it with newest SDR: Airspy HF+ Discovery and SDRplay RSPdx. It would be nice to compare their performance now.

Paul Kobetz

So, I have an RTL-SDR and really want the best SWL SDR…I am planning it life to be spent as a backup rcvr to my amazing KX3 Elecraft and as a loaner for those wanting to “listen in” to ham radio. Suggestions please.
Paulk

Ken

I have a 100% Grade-AAA MINT Drake R8-B with VHF card and Drake MS8 external speaker that have been sitting in a closet for 15 years…I would sell for a respectable sum that appreciates their condition (perfect).

Chris

I have a problem with the dongle shutting off by just barely bumping the computer or when I’m in the car u barley touch it and it goes off this is the most frustrated thing otherwise everything ok. Keep in it running is most important to me.

Wilhelm

My choice here is SDRplay for one reason: usability.
Currently I use an RTL-SDR dongle with an upconverter that has a bypass feature. Going from HF to VHF/UHF I just have to flip a switch on the converter, but I can’t remember how many times I stared at a screen with no signal until I realized I forgot to flip that darned switch. Or forgot to enabled/disable the offset in SDR#, dialing in the wrong frequency. With SDRplay that won’t happen.
Regarding Airspy: do they really expect me to (un)screw that converter every time I change HF/VHF? No way.

BRENT R JONES

I am new to the SDR era. Started listening on a Zenith TransOceanic that my mother had bought in the late 40s. Eventually, as a high school student, I got a 12 tube superhet receiver. Later a Sony 2001. Still later a Techsun PL-380. Now I got a RTL-SDR (RTL2832U). Considering an SDRplay 2, mainly because I like LW and SW. Thanks for the review. I would love to see the comparison include SDRplay2.

Luca

Kudo to the reviewer. I find your work very useful indeed!

Two questions that I would appreciate if you could reply:

– Did you test Airspy, or Airspy R2 (the manufacturer claims the R2 version to have better USB noise rejection).

– At the beginning of the article, you wrote: “Initially when writing this review we had deep problems with the imaging of strong signals on the RSP. However, a recent Dec 22 update to the drivers has fixed this imaging problem tremendously.”. Do the test reports in this article refer to the SDRplay with the updated drivers, or without?

Many thanks
Luca

David

I feel really dumb, but….

How do I get the HF functions working in gqrx under Linux? Presumably I need some new drivers. From where? And does gqrx support this?

Jay

May I ask why every single hack rf test graphic shown the noise floor is consitently higer than each of the other two with the airspy always having the lowest noise floor? Are you guys affiliated with any of these SDRs?

Bertie

There is an 8-bit ADC in the hackrf (and debug mode output from rtl-sdr dongles) which will provide a dynamic range of 49.92dB, but there is a 12-bit ADC in the airspy and sdrplay which provide a dynamic range of 74dB. So the sdrplay and airspy should have a lower noise floor than the hackrf. Having a higher dynamic range is useful if there are really strong signals right beside the weak signal that you are interested in. If there is not then you can just adjust the gains to bring the signal you are interested in within the window of your dynamic range. Or if needed you could decimate and have ~3dB of additional of dynamic range every time you half the bandwidth – basic DSP 101.

GasmAN

I find the article very well done and in many instances, a good reference.
The thing to remember is value. For the money I believe the SDRplay is the better investment.
I have added an LPF board from an TS-180S and a RTL/SDR SUB-LNA from JaniLab in Hungary.
This greatly improves sensitivity and image rejection. I have just started with the new software, but because I share my SDRplay, I continue to use SImon Brown’s software on a regular basis.
Above all…consider Omni-Rig and enjoy parallel tracking for smooth contesting on less expensive radios. I use the FT-DX-1200 with my SDRplay.

73,
Terry W3GAS

Bill

So as a noob with a $10 generic RTL dongle and a free copy of Cubic, who is ready to take this SDR stuff to the next level, what is the takeaway? I’m ready to buy, but I’d prefer to buy once. Airspy seems to be the winner but… My primary interest is in SWL – news from the other side of the world. I also live next to a very active USAF base and would like to listen in on their comms, and I’m on the coast so we’ve got a lot of Coast Guard and marine traffic as well. I landed here in the process of doing my homework before I buy, and until reading this I was leaning towards a NooElec dongle with the Ham it Up upconverter and a LNA4ALL as my rig. PS – I’m settled on a Par endfedz longwire to replace my homebrewed POS unless someone has a better suggestion.

TIA – Bill

Celso Ferrarini

I’m on the same opinion. As for starters, I picked the NESDR Smart dongle and a balun 9:1 from Nooelec that seems to be an affordable choice, but for the upconverter I was also interested in the Ham It Up, but I does not seem to go well on the comparation tests. I wonder if the spyverter is a good choice to match my NESDR SMART. Altough the T bias line feeding will not work. And the Tuner will not have the same bits resolution of the more expensive choices as AirSpy. It seems awkward to spyverter not even have a bypass swith as ham it up does. Does anyone have tested this combination NESDR + Spyverter?

Костик

Please check SDRPlay receiver, on a plot of 400-420 and 220-300. In the region of 400-420 I can’t hear almost anything at all. In the area of 258 MHz I don’t hear the SATCOM. Although another receiver on the same antenna hears. If anyone has the opportunity, please check the reception quality of 400-420 and 220-300 MHz. Above 300 MHz is no problem, everything works fine. Also works well aviation, 150-170, 420-470 and so on.

Mario

Found this review to be very helpful; many thanks! Just an FYI have used the HackRF One on the AM broadcast band (540 – 1600 KHz) and it receives the full spectrum very well (using a 43 foot vertical outdoor antenna). Can use SDR# and SDR Console, both are compatible with the HackRF One. Very easy to set up, similar to SDR dongle; run the Zadig driver, download SDR# or SDR Console and you are up and running. As with all receivers, a good antenna is the sine qua non for optimal reception, outdoor is better, higher is better, and bigger is better hi hi. Also, a pre-selector like the MFJ-956 or similar will help along with judicious use of the RF gain.

Kudos to GreatScottGadgets (HackRF One), SDR# folks, SDR-Radio (SDR Console), and Carl (RTL-SDR.com) for the great products, software, and technical information to keep the hobby evolving.

Agree totally with br0band, the length and content of your review is perfect. Thanks very much!

Klaus

I like most parts of your review. The only thing i dont like: you are compairing a 20 MSPS sampler (HackRF) with two ~8 MSPS samplers using SDR#. This is TOTALY WRONG. From a digital signal processing point of view you only can compare samplers if their sampling rates a equal to each other. For your review this means, that you have to upsample or downsample in many cases befor doing the measurement.

In other words: comparing 20 MSPS and 8 MSPS samplers to each other is only possible by
a) downsampling from 20 MSPS to 8 MSPS and compare to 8 MSPS samplers (which can not be done in SDR# or
b) upsampling from 8 MSPS to 20 MSPS and compare to 20 MSPS (which can not be done in SDR#)

Jim Viau

That is One Real Good Analysis! I now know pretty much everything that there is to know about these 3 units. Constructive comment: I really wished you had included the R820T2 RTL2832U SMA Dongle as a forth compared item. I know very well that the specifications are not of the same level. Obviously a $25 unit cannot compete with a $150 to $300 SDR but it would have been fun to see how RTL2832 make it in this competition. Good job! Thanks!

Joe

Many thanks – I was missing a comparison like this for a long time … well done and it reflects a the a real life usage model which so many others miss out on – great job!

DaveB

Finally (well probably not) what appears to be an objective in-depth review based on the real world with its hugely complex mix of signals. However for me what is missing is the image rejection performance. In the case of the SDRPlay – the zero IF image rejection and the image rejection in low IF mode. In the case of the AirSpy the low IF image rejection.

I can offer the following figures for SDRPlay using API 1.8.1 which has in built I/Q balancing. Zero IF image rejection around 60 dB and low IF image rejection with 2.048 Mbps sample rate wasabout 40 dB – the image appearing 4.096 MHz below. This on the FM broadcast band – FM DXing is my main interest and images falling on a weak DX signal is the last thing that we want.

I don’t own an Airspy and would like a solid ‘real world’ number for the low IF image rejection at 10mbps sample rate – it would need to be closer to 70 dB to make a purchase worthwhile – as my SDRPlay performs adequately and I need a significant improvement to warrant opening my wallet.

Out of interest I measured the low IF image rejection on an R820T2 dongle at 2.048 Mpbs sample rate at around 45 dB – better than the SDRPlay in low IF mode. Again in the FM broadcast band.

These are ballpark figures merely arrived at by reading off the peak signal strength for the main signal and comparing it top the peak signal strength of the image as shown on the SDR#.

DaveB

Rick

Wish I could post a picture here. A picture tells a thousand words – they say 🙂

Just done my own S/n comparisons between my Airspy and SDRPlay on 163Mhz.
Same weak RF generate signal from a calibrated HP service monitor, same patch lead and same software (SDR#) with gain controls adjusted for best S/n on both front ends.

Bottom line:
My SDRplay is significantly better than my Airspy.
Either my Airspy is slightly deaf, or something is a bit ‘screwy’ with the above review.
Would love for others to do similar comparisons.

Anonymous

Can you do some tests on a range of frequencies? I’m curious to see the results.

Adam

I think that the lab testing is the only test that can give us the real figures and performance. Sensitivity is only one test, so there are other tests that should be performed, selectivity, linearity, SFDR, two tones, multi tones, LO PN etc. Upon the mentioned measurements one can get a clear picture of how any receiver will perform in a desired environment.

The way the tests were performed in the article may be handy but far away from the precise because the readers are looking for the fraction of the dB. Such accuracy is difficult to get even in the lab sometimes not to mention problems using the external signals through the antenna where influence of the external factor may be significant and selective. Let say weak signal test where the QSB is present and even multiple reflections on the higher frequencies.

I see the main problem in testing the devices that are designed for different audience with different ideas and approach. Of course, it will be nice to have one unit combining the HackRF frequency range, Airspy linearity, RSP sensitivity, HackRF TX possibility, Airspy GUI and RSP filtering. But we do not have that on the low end market.

Tim

May be you fried your Airspy with ESD.

Truth

An RTL-SDR with a LNA will outperform the RSP in the lab any day. Does it make the combination a good receiver? Definitely not!
Most noobs (and also some old farts) seem to confuse s-meter and signal quality. Put your super RSP in the field with real signals and it will show its weaknesses like in this review.
I can understand the RSP crowd feel like raped after reading this article… HAHAHA

Anonymous

That’s ok Yousef you can stick to your crap Airspy that needs an up converter, ill stick to my RSP All-In-One box that needs nothing else and has people that actually help others.

Truth

Sorry to disappoint you dude. I’m not Youseff. But I genuinely think you have problem with that chap.
For your RSP: Eat shit 1 million flies can’t be wrong!

Anonymous

So you have a Lab and tested your theory out then.

nf2g

I would like to suggest separate reviews for receive-only hardware (Airspy, RSP, RTL dongles – including E4000, etc.) and SDR transceivers (HackRF, Flex series, etc.).

There should also be head-to-head comparisons of software, with each package reviewed while operating each hardware platform it claims to support. Then we will finally see comparisons of apples to apples.

Klaus

YES PLEASE, please include always the Dongle (RTL-SDR at least) in those tests.
Everybody (I think) has one at home, so it will be great to see the benefits of the others, 8 or 10 times more expensive SDRs.

Marty Wittrock

I have to add something to the fray – – While I love my HackRF Blue (which is no different from HackRF) the most glaring issue – to me – is the base support for the product. Try collaborating with anyone on their Facebook support page or other forums – – the KEY information for HackRF is nonexistent unless you’re in the handful of developers that have all the register addresses to perform the functionality of the HackRF. The source that’s out there is only a fraction of what you need to make it work. I’ve been to GitHub several times looking for ANYTHING that would give me some clue how to write transceiver code for the HackRF and you only find fragments of what you need and the rest of it is in someone’s head that you have no access to. I even wrote Michael Ossmann himself to ask the questions and never got a response. If the only thing you can use to operate the HackRF as a transceiver is GNU Radio Companion (which, luckily, works pretty well in its environment and compiled in Python) you would be screwed – – there are no other apps that I know of other than GRC that the HackRF can use to perform transceiver functions. You won’t find it in GQRX. You won’t find it in SDR#, and you sure as Hell won’t find it in HDSDR, too. NO APPS OTHER THAN GRC FOR TRANSMIT. So that pretty much locks you down to GRC and Linux under Ubuntu 15.0 and up. And there is no grassroots support for HackRF that I’ve noticed, too. I added transmit and receive functionality files (in GRC) to the HackRF Blue Facebook page and even gave instruction that I WANTED to see people IMPROVE ON WHAT I UPLOADED and. . . . . . . .NOTHING. NOBODY IMPROVED OR ADDED TO WHAT I PUT UP THERE. It just seems that they pulled down the files and played with them – that’s it. So I’m of the opinion that unless you’re in the ‘inner circle’ of HackRF development, you’re pretty much screwed. You WON’T find anyone doing cool things with HackRF at all. It’s a vast wasteland.

THAT’S WHY I WENT TO RED PITAYA AND I’M NEVER LEAVING. At least THEY are developing apps – – and REALLY GOOD ONES AT THAT – – to support the Red Pitaya for a FULL UP HF TRANSCEIVER. My plan is to produce an upconverter for Red Pitaya that will allow it to transmit U/VHF, too, longer term.

So, LONG LIVE RED PITAYA..!

Marty Wittrock

I’m part of the HackRF Blue Facebook group and recently understood that there are some HackRF Blue boards that will be built and available by the April timeframe, too – so they’re still building them. Most of the builds happen in Asia and I’m certain that Gareth Hayes has been busy setting all that up (Gareth appears to be the primary force behind HackRF Blue and I have to admire that he’s making it cheaper so more people can get in and develop). But the whole HackRF movement is VERY MUCH under wraps. Unless you’re within that University ‘inner circle’ you’re not going to get much traction with HackRF. That makes it REALLY frustrating when it’s such a fine transceiver and the only tools you have are GNU Radio Companion on the transmit side. Compiling the GRC flow diagrams to Python code will get you part of the answer, but then you have to ‘roll your own’ interface and hope that it plays nice with the HackRF. If they’d just put the register functionality out there to program the HackRF, you’d see A LOT of people use it and it’d probably eclipse RTLs and even Red Pitaya…But they didn’t. They kept it locked up and those that really know the HackRF aren’t talking. So that’s why I went to Red Pitaya – at least that’s the CLOSEST thing there is to the HERMES/TAPR transceiver there is – it uses the same PowerSDR software to make it run and it plays A LOT like my former Apache Labs ANAN-10E ran (yeah, I sold mine recently because of that). Anyway, Red Pitaya is going to eclipse HackRF if they keep on going the way they’re going with development. There’s some SHARP people working it and all the source is out there to play with. It’s too bad…I really love that HackRF, but I won’t be controlled by trolls that won’t share their source on what’s supposed to be a ‘open source radio’…End of soapbox rant…73

peter

Sorry,
Not Yousseff here.
But how could he resist having an attitude, if this review is presented by dozens of SDR# pictures as reference software? :-))

Adam

OMG………. 🙂
As the author of the post excuse himself in the diclaimer note, I will try to keep my comment short. Othervise this will be a long comment, even longer than the author post 🙂
I can recognise a lot of efforts to perform such a detailed test and to write the article but unfortunatelly this test can not bring the real light on the performance of the tested equiment simply because of the method used to perform the comparison. I will not judge about the tested receivers performance, I will try to point out some wrong statements written about the equipment I made (LNA4ALL).

To quote exactly:
Regarding the LNA tests the Airspy team write:
[In the LNA tests] the LNA [LNA4ALL] might not have enough dynamic range in a high performance setup since it overloads before the Airspy.

This is wrong statement. If we take the R820T datashet figures as the best case scenario the same tuner have the IIP3 at the max gain of (minus) -7.5dBm. The LNA4ALL have the measured (real figures) OIP3 at the max gain (plus) 34dBm. Taking the max gain of 24dB the LNA4ALL IIP3 is then (plus) +10dBm. Of course, the higher number (positive) is better. For the guys that are not familiar with the linearity and IP3 theory, this is the difference like driving the Yugo 55 and Ferrari. It is obvious that the receiver using the R820T tuner will be overloaded before the LNA. Well, the HackRF guys are using the LNA4ALL as the transmitt power amplifier delivering 180mW of output power. This should be clear enough proof of P1dB and what the LNA4ALL can deliver.

The second wrong statement:
The Airspy team suggested that the LNA should improve SNR on the Airspy, but that the LNA we used was not suitable for our environment due to it overloading on BCFM signals. They suggested that we should have used a LNA with a much higher dynamic range such as the PGA103+.

If they ever build and measure the LNA with the PGA103+ they will never write such a statement. But even if they consult the datasheet they would see the the declared higher dynamic range can be achieved just on the higher frequencies, above the frequency range of the R820T tuner (much higher). This myth came from the “copycat” approach where the “designers” are just copying the layout and the declared figures without masuring anything seriously. Most of this LNAs are not even working properly because the builders were so lazy to read the complete documet set like Technical note AN-60-064 regarding the unstability problems. This LNA is not unconditionaly stable below 150MHz thus the Minicircuits release the patch (stabilizing network) at the MMIC input to meet the Rollet factor K over 1. How many PGA103* LNAs have you seen using this patch? Some of the fair designer admits that they have problems with unstability and they did try to solve that problems with the bias coil, but most of the designers does not recognise that problem at all. The MMICs are just smoking after you turn on the desktop light connected to the switched 12V power supply 🙂

On the other side the SDRPlay guys recognised the problem of uing excessive gain (LNA4ALL) in front of the SDRPlay properly. The SDRPlay have already the LNA inside, so extra LNA with such a high gain is not required. The proper test should be comparing the Airspy and SDRPlay without internal LNA but with the external LNA4ALL. I will stop here. If not, I risk to write a really long comment.

Anonymous

Your input is much appreciated Adam, we need more technical feedback in these types of situations.

Uncle Undecided

I do appreciate a long comment of this quality anytime, Adam! Please feel free to blow up the page! 🙂

Uncle Undecided

“However, the MGA-68563 on the RSP is “always on” and cannot be turned off.”

Are you sure? Isn’t it that the MGA68536 is just not having a true bypass, but is turned off by removing power via the software?

B. von Klaus

First, I’m really impressed by the excellent simplification job done by Team RTL-SDR.com. Anyone with a basic setup can reproduce these tests and form his own opinion. The entire article is fluent and a is real reading pleasure.

I tend to ignore random pseudo-technical comments in the Internet, but this one really pushed it too far. Arguing about PGA-103+ not being better than the PSA4-5043+ because it requires an extra coil to just makes any actual RF engineer laugh. The LNA tests clearly show the LNA4ALL started introducing nonlinearities. Since the RF gain was set low on both Airspy and SDR-Play, we cannot speak of saturation in the tested SDR’s. The poor LNA4ALL didn’t improve the NF because it was smacked by the all the VHF/UHF blockers. The only test that have improved things a bit had a FM BC filter *in front* of LNA4ALL. A more robust LNA with better linearity should have smoked the LNA4ALL in all these tests and should have given a better perception of the differences. But I understand it’s supposed to be cheap, so “high performance” was not the target in these tests. Just please… don’t spread BS. You are not doing the concerned population any favor.

Joe

> B. von Klaus
Dear “Mr actual RF engineer” please build a few PGA-103+ based amps and operate them over a wide frequency range on a long coax as antenna preamps for a while and you will stop laughing. The extra circuitry mentioned in the much later released application note, which is a necessity in order to get this thing half-way stable is not exactly improving the overall performance.
Adam describes the situation very well and believe me it is not a coincidence that a lot of preamps, e.g. for active ADS-B antennas that need a very low noise figure while having a great dynamic range magically are all being distributed with the same MMIC as the LNA4ALL and not the PGA-103+ which just looks much better on the paper. You can be sure that Adam would use the chip if it would be stable, since it is cheaper than the PSA4-5043+ and as mentioned offers less noise while having more dynamic range.
So maybe before you start laughing and making the LNA4ALL look bad, breath a bit a of real world experience, because these MMICs are very tricky to handle … I have been killing a few though I ran them with a current and voltage limiter set below its max current, which is something I have never experienced with any other chip!
I also run a few PGA-103s successfully for a few years but please don’t ask me why this process seems almost like a lottery game. Anyway I would never be selling these to end users and totally understand why so many designs end up using the PSA4-5043+ … it’s definitely not because they are too dumb to create a proper wideband capable RF environment to operate them in!

Adam

Hello Joe,

I see we both learn the hard way how PGA-103 perform. When the PGA-103 was released I was pleased with the datasheet figures but a bit suspicious about device performing up to 6GHz with the parasitic using the SOT-89 package. But hey, I said let’s not judge the device before we test it. Build the amplifier, everything looking fine until we switch on the desktop lite (running on the switching power supply). Replaced the PGA-103, and again popping as the popcorn after desktop light switched on. Consulting the S2p files and analyzing the device we notice the problem bellow the 150Mhz but nothing that we have not seen on the other mmic devices so far. Contacted Minicircuits to see what I am doing wrong and got reply that they are aware of the problem and they email us patch (stabilization input circle) at that time not published yet in the technical note. Patch applied, seen some improvement but still not unconditionally stable LNA. Of course, the LNA NF and other parameters were degraded approaching the PSA4-5043 figures. I did try to solve the problem, used the bias-T advised from the Minicircuits, layout as proposed but the results were always the same. The current and voltage protection on the lab power supply are just to slow for that transition. I did monitor on the spectrum analyzer the wideband output and using the phase shifter at the input creating all kind of combinations and at certain point the device pop up…. the oscillations were present on the HF region even with the stabilizing network. They were lower then without network but still present.
I decide not to use this device as the users may create all kind of input conditions using wires, long antennas, open, short etc. and this device can not handle all that on such a wide range of frequencies.

Minicircuits have quite strong PGA-103 advertising campaign what can be seen reading the various uW magazines. The technical note regarding the stabilization patch was upgraded several times. Driven by the fact that most of the buyers are probably not reading that papers they now include the patch in the datasheet too, at the very end of PDF file.

Recently we got some equipment fail to work. Inspected and found only PGA-103 burned in the simple LO amplifier chain. All other mmics (other types were OK). Further inspection bring me to conclusion that poor passive mixer isolation between the ports feed the HF IF drive to the PGA-103 attached to the same mixer from the LO side and the PGA-103 was damaged.

The problems are present and this post if far away from BS. They can help others not to run in the same problems.

Z33T

I agree with Adam. 73!

10-10

Thanks for the detailed rewiew. Airspy is the clear winner but i cannot afford to buy a 700$ laptop to run it. SDRplay is a good budget SDR but after this review i am a little disappointed on how it compares to Airspy on VHF-UHF; i expected much more from a 8 filters preselector. The perfect budget SDR has not been released yet. Only those interested mainly to receive HF can be happy to see the best price matching with the best performance.

RFnoob

A $700 computer to run the Airspy really? I didn’t know you needed one of those. I am going to have to tell my $60 Intel Pentium about that and ask him to please stop faking things.

Harry Plotter

Too much on one page indeed. I don”t get why the HackRF is always getting into comparisons like this as it’s not even pronouncedly marketed as SDR for listening purposes but as a device for RF measurements

Uncle Undecided

BTW I think putting this on a single page is a bad idea, my browser is totally acting up, the computer freezes and a page reload causes several minutes of elements jumping across the screen while causing more hanging of the whole computer because the browser is pegging the CPU during load. Writing this comment took ages.

Mike

If your computer cant handle this webpage then how is going to handle any sdr?

Bill

LOL… just spit beer on my SDR.

Uncle Undecided

Thanks!

Uncle Undecided

The only thing I don’t like about this review is that it came 3 days AFTER I bought an SDRplay, finding out about the various modes of overloading mayhem all on my own. I’m not even living in a city. A lot of fiddling in the menus is needed to beat it into shape on each band. Besides all theory I found that the LNA is increasing SNR significantly, at least in the 88-140Mhz range (didn’t make test anywhere else yet). Would I have bought one of the other two? No, I think the SDR I really want to buy hasn’t been developed yet.

Mark Heleniak

Excellent write up. Kudos amigo! I’d like to add some contextal opinion to your conclusions:
HackRF is the clear ‘value choice’ for SDR transceivers and is the entry-level gateway to the ‘big world of RF’. Think about it from the viewpoint of someone with limited means just learning about the ‘invisible world’ .. HackRF + tablet (~$600) allows hacking into a $120K Tesla. Designer Michael Ossmann uses this logic when speaking to security & academic crowds saying if we do not put this technology into the hands of the good guys, the bad guys will find it anyway and then we are in deeper trouble. It is indeed a brave new world!

Airspy is the ‘value choice’ of the VHF/UHF SDR bunch. It costs more b/c it uses a tighter crystal oscillator, provides a expansion headers and bias T and uses a higher-requirement USB buss, and does so because this is what is required for serious UHF work. Plus Airspy authors its own software and 3rd-party plug-ins give Airspy the “complete package of impressive UHF performance”. Add to that, only Airspy allows for diversity reception which means synchronization of two receivers which means a higher tier of non-dropout reception.

SDRPlay is the ‘value choice’ for the serious all-band, all-mode hobbyist on a budget. “$150 + a 3rd-hand cpu will get you the world”. It’s the one I bought to go w/ my 10 year old laptop :).

Dr Chris Kear

I know it’s a long time since you wrote this, but could you point me in the right direction for information regarding implementing the diversity features of the airspy products? Thanks. You could email me at tasmedic at yandex dot com. Many thanks.

rfboy

the airspy still have usb noise issue. strong usb noise @ 480MHz,240MHz,120MHz.

Kyle

Great writeup, will take me a bit of time to tuly digest it all. Thank you for immediately pointing out the significant shortcomings of the HackRF. It may be in the same price range (highest priced of the three), but IMHO the performance of HackRF One is unnaceptably poor. I’m hoping that sales suffer significantly and force the creators to fix the glaring issues in that device. It seems that the kickstarter origin may have caused the creators to rush the design and just release a “functioning” model.

Bob

I think you need to look at the history of the HackRF One. There was the 500 government paid for Jawbreaker before it, and before that jellybean, licorice, lollipop, lemondrop and bubblegum revisions of the hardware. I do not think that the Kickstarter caused any rush to market.

Frank K2NCC

I knew the Airspy was better than the SDRP and HRF1, but nice to see more evidence of it.

I got rid of my Spyverter because of the need to use a switch with the Airspy. Sold my HackRF1 with just a couple hours on it, really not a good receiver.

Now I have the SDRP, the Airspy (no Spyverter), RTLSDR dongles of course, and a “real” SDR that hasn’t been reviewed here yet. 🙂

Many many thanks for the article and all your work RTLSDRblog!

vy 73 f

HFMilman

The only downside to buying the Airspy is the bucket load of attitude you get free from the developer.

Angel

True True True

peter

Apparantly he is entitled to some attitude with such a good product.
What have you developed for us lately?

Anonymous

Hello Youssef!

David Brown

It is an awesome product. It is also entirely possible to produce an awesome product and not have the attitude. One does not require the other, and says more about the person wielding the attitude than the product.

br0adband

Thanks for this rather extensive review/comparison, it provided a ton of useful info without going bonkers on the extremely technical side of things (not that that type of review info isn’t useful, mind you) and will most definitely end up helping a lot of people make their purchasing choices as to what SDR hardware might suit them best.

Have fun, always…