Frequent frequency shifts and syncing

Discuss KerberosSDR - 4x Coherent RTL-SDR
Post Reply
LuYfbHEe
Posts: 3
Joined: Mon Jul 27, 2020 7:17 pm

Frequent frequency shifts and syncing

Post by LuYfbHEe » Mon Jul 27, 2020 7:26 pm

I have a project that requires frequency changes every measurement. Since this needs a new calibration for each frequency, I made automated calibration boards controlled by the GPIO of the Raspberry Pi - so I can get new delay values for each frequency and then use those values for each frequency. That's fine, but after some time, the code hangs when reading the buffer. Eventually, I determined that the delays are added cumulatively in sync.c, and eventually, they exceed the buffer size. My question is, what is the purpose of this delay accumulation? For my application, would it make more sense to use fresh delay values each time, and not accumulate them?

ckoval7
Posts: 48
Joined: Fri Sep 27, 2019 8:54 pm

Re: Frequent frequency shifts and syncing

Post by ckoval7 » Tue Jul 28, 2020 1:38 pm

How often are you resyncing? I have a similar setup, but I don't change frequencies so often that I run into issues like that.

LuYfbHEe
Posts: 3
Joined: Mon Jul 27, 2020 7:17 pm

Re: Frequent frequency shifts and syncing

Post by LuYfbHEe » Tue Jul 28, 2020 9:21 pm

I change frequencies, and thus set sampling delays, every 20 seconds, and recalibrate every 5 minutes. There are basically three scenarios: 1) accumulated delay is less than negative BUFFER_SIZE (fails) 2) accumulated delay value value exceeds BUFFER SIZE (fails) or 3) neither 1 or 2 (successful). Perhaps I could make the accumulated delay value modulo BUFFER_SIZE to circumvent these failures. Or set the delays within the python code after sampling, rather than through sync.

rtlsdrblog
Site Admin
Posts: 2784
Joined: Mon Nov 19, 2012 11:54 pm

Re: Frequent frequency shifts and syncing

Post by rtlsdrblog » Tue Aug 11, 2020 11:32 am

I'll forward this to the dev, but we are getting very close to releasing our own new code (fully rewritten) that self calibrates on retuning automatically. You do need a retrofit switching board for the V1 Kerberos which we will have available later this year too.

If you have your own switch board it should also work, just modify the code, or tie the switches to the noise source GPIO on the Kerberos main PCB,

Post Reply