search

airspy / airspyone_firmware

AirSpy Firmware
http://airspy.com
GNU General Public License v2.0
153 stars 49 forks source link

Arbitrary sample rate support #15

Closed commthu closed 8 years ago

commthu commented 8 years ago

I need arbitrary sample rate, your sample rate is so limited for our applications

touil commented 4 years ago

OK. Why not go to WebSDR github and ask them to support more sample rates? => Because it doesn't exist. In fact, they don't care about your specific requirement and probably don't even know you exist. But here we are in a programming issue tracking interface. You're not a programmer, you have barely any understanding of DSP and you claim your rights to have things implemented in a public repository that are at best very bad implementation ideas. And we must solve all the underlying problems because some other entity imposed restrictions to their proprietary software so that people like you don't abuse it.

I think your best bet at this point is to find a competent DSP programmer to analyze your problem and write the right 5 lines of code to solve it.

Have fun (parts of the hobby) and good luck (part of life) !

touil commented 4 years ago

Here's your homework done:

airspy_rx -a 4096000 -t 2 -r /dev/stdout -f ${center_frequency} -g ${gain} \
  | csdr convert_s16_f \
  | csdr fir_decimate_cc 2
ON5HB commented 4 years ago

Well as I said before, websdr.org is closed source, the developer doesn't give the source. Trust me I asked him several times to make us run more sample-rates. But he declines.

A whole lot of programmers have been playing with the issue and we also lack documentation. The connection point is RTL-Dongle with a lot of sample-rates but once get past 2048 there is another at 2880 but when used with more then 1 receiver it dies.

So we have to stick to 2048. That leaves 3 to 4 receivers at 2048, unless we manage to let them output even less, then it's possible to use more receivers.

The RTL-dongle also outputs 2048 and below it decimates, also limiting us. I was hoping the Airspy could make our systems more productive because they contain firmware that may alter the outputs.

Also the USB-bus craps out, often before we get started, as such we need limited sample-rates. I do program, but it's limited, but trust me, we tried all possible ways, even via pulseaudio with hacks, they all crap out.

The input we need is CU8......no S16_F. I understand you try to help, we tried all. The only solution is lower sample-rates in the firmware.

Have a look, I use this: https://github.com/ON5HB/airspy_tcp

touil commented 4 years ago

Add a last conversion step to what I proposed. Note that this is a decimation by 2. It is much faster than whatever you tried before.

ON5HB commented 4 years ago

There is no point. I need firmware that accepts more sample-rates. You keep dancing around the issue. That means too muck processing and USB usage. As such the Airspy isn't a good receiver for our systems to use. I will stick to the RSP's for now and wait until the Airspy does take better sample-rates that we can use. Such a shame, as I really do like the design and filters.

Our problem is that the hardware should NOT output stuff we do not need, most do. For the processing we do, we run into trouble as numbers add up with receivers and they go up fast.

touil commented 4 years ago

There is no point.

Fair enough. Good luck!

ON5HB commented 4 years ago

Please let me know if the firmware changes and there are options that I can use.