sdr-power
sdr-power copied to clipboard
polygon
No Airspy support
I've seen the results from the HackRF tests (on the #hackrf IRC channel) and they are very impressive. And I was just thinking that since the code base of the Airspy hardware was originally derived from the HackRF code base it might be a tiny bit easier to get 12 bit support working using Airspy hardware before the BladeRF.
https://github.com/airspy/host
Thank you for your code and feel free to totally ignore my suggestion.
I don't have access to an AirSpy so I won't be able to test.
I took a look at libairspy though and it sure looks very similar to the hackrf. Building a new backend is probably not that hard. If you can provide a shell on a box with an airspy attached I would give this a try.
As a first step I think it's easier to convert the samples to 8 bits instead of changing the internals of the analyzer.
Part one and two are solved. I haven't dug into your source code and you would know best. But in my mind throwing away 4 bits sounds like 16x less awesome than it could be :) But I suppose you don't replace the wheel with a spaceship without going through a few iterations in between.
I started working on this and most of the backend is finished already. I'm having a bit of a headache though because the Airspy only supports two fixed sampling rates (2.5 and 10 MHz) while the current code expects arbitrary sampling rates in a min/max range.
I'm a bit hesitant to touch the original rtl-power calculation code since there might be a lot of changes involved. By fixing the sampling frequency, one would probably lose a degree of freedom somewhere else. For example, the frequency range, bin size or crop might need to be changed to differ from the command line so that the sampling rate always ends up being 10 MHz.
The other option would be to provide downsampling to arbitrary rates in software. That would require some form of fractional resampling. I'm currently thinking along the lines of several filters for different bandwidths and maybe just linear interpolation for the resampling part.
So, I guess what I want to say is, it's definitely more difficult to do this than I originally thought and might take a while. If you or someone else have some ideas on the matter, feel free to comment.
FYI: The new Airspy Mini supports 3 and 6MHz, maybe put this on the back burner I thought this would be easier, but I see your point.
I might have overthought this. Apparently it is already the case that the step-size in the frequency range parameter is just an upper limit. I see if I can extend the tuning planner to support a fixed sampling rate.
I got a mini and plugged that in as well. Probably not much help, but always good to have more hardware than less to test on. I've no antenna connected to it at the moment, only the NOS has an antenna.
Some updates on this. The fixed sampling frequency tuning planner is done and seems to work. Now starting to test with actual hardware.