A bit of advice and feedback needed, especially from owners of rtl-sdr.com blog v3, Airspy Mini and SDRPlay RSP1A devices.
I’ve dabbled for a couple of years with an rtl-sdr.com blog v3 adapter with great success and more recently decided that I wanted a bit more bandwidth and something with more dynamic range that is better able to cope with in strong in band interfering signals.. (blocking performance etc)
In the £100 price range, I’m willing to pay the two contenders seemed to be the SDRPlay RSP1A and the Airspy Mini, with 14 and 12 bit ADC’s respectively. I did all the research I could, read all the reviews I could find, checked software compatibility etc, and weighed up the numerous pros and cons of both devices and finally settled on the RSP1A.
Unfortunately after using it for a couple of days I feel like I’ve made the wrong decision and I’m actually pretty disappointed in its performance, even compared to my rtl-sdr v3! So much so that I’m very likely going to send it back in the next few days while I still can. (Distance selling laws etc)
The biggest issues I have with it are all with the Zero IF modes – the massive DC spike, (on the order of 25dB in software like SDR Console) which can’t be corrected on much of the software I’d like to use and and numerous internal spurries (birdies) that the device seems to generate and pick up in its IF (post tuner) stages. Pictures tell a thousand words so I’ve captured some videos demonstrating what I’m seeing and uploaded them to Youtube, so I hope people can take a few minutes of their time to watch these to get a better idea of the issues I’m Seeing and maybe try to reproduce it with the same settings. In these examples I’m using SDR++ and SDR Console but I see similar problems in other software as well. In all these videos there is no antenna connected, so I’m only examining internal spurries.
In this first video I’m using SDR++ as it gives more detailed control over the tuner settings than SDR Console, sample rate is set to 8Mhz, IF filter to 6Mhz LNA and IF are set fairly low but not right down. As well as the DC spike there are two unstable, mirrored spurries which wander around like drunks, later in the video I adjust the IF gain which causes the frequency of these spurries to change and move all over the place.
While turning the IF gain up high hides/swamps these particular spurries, the device as a whole performs poorly with the IF gain turned up high, in fact in both SDR++ and SDR Console I find I have to keep the IF gain towards the bottom end for best signal performance, in which case the spurries are clearly evident when tuning to a quiet frequency range.
In the second video I’m using SDR Console and seeing the same issue at a different frequency, I step through some different LNA and IF gain settings to show how it affects the spurries, I then switch to my RTL-SDR v3 adapter also with no antenna connected and step through different gain settings to show how clean the noise floor is even up to relatively high gain levels – there is an uncorrected IQ imbalance causing a small spike (IQ correction in SDR Console doesn’t seem to work for me for RTL-SDR, seems like a bug as it works in other programs on the same adapter) but there are no spurries like the RSP1A. Near the end I switch to 2Mhz Low IF mode and most of the problems go away.
In the third video I demonstrate the massive main clock spur harmonic around 120Mhz and sidebands surrounding it, which are both really quite bad and right in the middle of the aircraft band. (The center spur is 50dB above the noise floor!) There is an image of this spur at 120.8Mhz as well which is about 30dB above the noise floor.
I then switch the LNA and IF gain to various values and while I can suppress the sidebands of the clock harmonic the 120 and 120.8Mhz spurs are still very strong. I then switch to my rtl-sdr v3 adapter and there is a small spur at 120Mhz (also a main clock harmonic I) but it is only 20dB above the noise and there is nothing at 120.8Mhz, no sidebands or other crap showing up on the RSP1A. So apart from that one clock spur the nearby aircraft band’s noise floor is clean.
The amount of crap in this region in Zero IF mode on the RSP1A is pretty bad IMHO. Near the end of this video I switch to 2Mhz Low IF mode and while it clean up the rest of the spectrums including the DC spike, it doesn’t get rid of the clock harmonic or it’s sidebands at 120Mhz, which are right on top of a commonly used aircraft frequency and still 50dB above the noise floor.
In my final video I am set to 2.7Mhz first in 6Mhz then 3Mhz Zero IF mode to show that the spurries are present at low frequencies as well. In 2Mhz low IF mode things are cleaned up considerably but at the expense of losing most of the available bandwidth of the device. So am I doing something wrong or is this device just not very good ? After the anticipation of waiting for it to arrive I’m pretty gutted really.
I don’t seem to be the only person experiencing these problems as I was able to find the following discussions when I looked hard enough:
This one refers to “wobbling spurs” which drift in frequency and the owner went as far as using cold spray to identify which chips may have been generating it. My own theory on the wobbling spurs/spurries is that there may be a free running DC/DC converter in the device which is leaking some of its switching frequency into the IF stages – when you first “start” the device in SDR Console for example the spurs quickly down in frequency then start wobbling, and the fact their frequencies race that when you adjust the gain of the LNA and IF amplifiers suggests to me that a change in the current drawn by these amplifiers is affecting the free running frequency of the DC/DC converter that is leaking into the receiver path.
Another thread here shows screenshots very similar to mine, however SDRPlay support have tried to fob this off by saying that the noise humps are a result of sigma delta encoding without enough noise on the input, however in my opinion that doesn’t explain why they wobble around in frequency so much and the frequency was affected so much by the gain settings, which shouldn’t have any effect. (But would have an effect on the power drawn by the chip potentially affecting the source of the spurious signal)
Does it perform better than my rtl-sdr v3 in some ways ? Yes it does. Its large performance, eg weak reception signal with a nearby strong signal is considerably better, as you would hope from the 14 bit ADC vs 8 bit. This is particularly notable on one FM Broadcast station which is always distorted sounding with a sharp notch visible in the middle no matter what bit rate or filter settings I use on the rtl-sdr v3 – on the RSP1A this FM station looks normal on the spectrum graph and sounds nice and clear pretty much regardless of gain or bandwidth settings. In fact broadcast FM reception is great and definitely better quality than the rtl-sdr v3.
Is it more sensitive for ADS-B ? Yes, but only very marginally. I left both running together with identical dipoles spaced about a meter apart, using dump1090 for the RTL-SDR and SDRPLay_dump1090 for the RSP1A, both feeding into Flight Radar so I could easily see the count of how many packets each received and the RSP1A was receiving about 5-10% more packets, although the range it was receiving them from wasn’t significantly different, and sometimes the rtl-sdr v3 was receiving planes from further away. So it was slightly better here on average but the rtl-sdr v3 held its own very well and neither had a determining advantage.
Is it better on HF than the Direct Sampling mode of the rtl-sdr v3 ? Well, sort of. It certainly has a lot more gain on HF since the rtl-sdr v3 devices direct sampling mode feeds directly into the ADC without the benefit of an adjustable LNA, so the RSP1A had more gain and a way to adjust it. And if you’re willing to sacrifice most of your bandwidth and use a low sample rate in low IF mode adjacent channel selection is better.
However in the Zero IF mode it was pretty much useless because there were so many internally generated spurries that it was hard to know what was real and what wasn’t without disconnecting the antenna to check all the time. While the rtl-sdr v3 had lower gain, the SNR was still surprisingly good and was able to receive everything the RSP1A could, and it was largely free of spurries apart from the mirroring of spectrum around 14Mhz which is a limitation of how the direct sampling mode works. (28.8Mhz sample rate without a 14Mhz low pass filter)
To get good performance from the RSP1A on the HF bands I found it has to be used in Low IF mode restricting bandwidth to 2Mhz or less, so similar to what I can already achieve with the RTL-SDR and the benefit of a birds eye view of a large spectrum width simply doesn’t materialise when Zero IF mode introduces so many spurries and a large DC spike.
Having the switchable MW/FM/DAB filters is nice in theory but I never once found a situation where they helped – I obviously don’t live close enough to any large transmitters (the nearest is 15 miles away I think) to need the notch Filters, so the 11 preselection filters are already sufficient for out of band signals, it’s in band interference between strong and weak signals I’m more concerned about.
So I’m pretty gutted to be honest, as the improvement over the rtl-sdr v3 wasn’t as big as I hoped and in many ways it seems considerably worse – it’s implementation of a Zero IF architecture to be flawed, and there is something in there using a free running oscillator which is inadequately filtered and/or shielded, plus the very large spur from the main clock finding its way into the signal chain.
It’s such a surprise to find the low signal performance of the £30 rtl-sdr.com v3 to be so competitive with a device three times its price, and for its noise floor to be so clean and largely free of the crap that I’ m seeing on the RSP1A. Apart from a slight IQ imbalance in some software like SDR Console and a minor clock spur the rtl device is very clean and it remains an absolute bargain in the price range.
So what next? I’d like feedback/suggestions but I think I’m going to have to send it back in a few days and give an Airspy Mini a try to see if it fares any better. I don’t really need HF as I’m not big into HF and I have some useful HF performance for casual listening from the rtl-sdr v3 anyway. (And it performs surprisingly well now I’ve directly compared it against a device that is specifically designed to cover that range)
I guess what I really want is something similar to an rtl-sdr but with wider bandwidth, a better ADC with better instantaneous dynamic range, and the Airspy Mini with 12 bit ADC is quite possibly that, and possibly what I should have chosen to begin with. Plus it supports SDR# which I think is still my favorite SDR software for casual tuning around – something that I can’t use (apart from very old versions or hacks) with the RSP1A. (I haven’t even mentioned SDRUno, which IMHO is god awful)
Are there any Airspy mini owners who can report on their performance in similar scenarios to the ones I presented in my Youtube videos (minus the HF testing) and can comment on how clean the noise floor is on these and how resistant they are to adjacent channel interference when used in the full 6Mhz mode ? Or who can do a direct comparison with the rtl-sdr blog v3 device I am using as my benchmark ?
High gain and 11 preselection filters means nothing to me if the noise floor even with no antenna is connected is not clean and I can’t tell what’s real and what’s fake – I’d rather have a device with a little lower gain that is clean and free of spurious signals.
Thanks for your time everyone.