TX interpolation >= 16x problem

[srcejon]

As was discussed on #1892, the M17 modulator did not appear to work with low baseband sample rate (<100k).

I suspect the problem is actually with interpolation when set to 16x or greater.

Here's a picture of TX and RX spectrum of the packet mod, with Int = 8 - which looks OK (and pretty similar to Int 1, 2 and 4)

And here it is with Int = 16

Which looks like it has a problem.

[f4exb]

"MyB210" suggests you are using a USRP here. I have tried with a HackRF Tx + RTL-SDR Rx combo and I do not notice the issue:

https://github.com/f4exb/sdrangel/assets/6192319/6a074c5d-d7ab-48f6-bb5c-7510129091fe

Also what modulator do you use? I have used the packet modulator in a loop transmission to obtain similar pulses.

[srcejon]

Yep, a USRP with with Packet Mod.

In usrpoutputthread.h

   Interpolators<qint16, SDR_TX_SAMP_SZ, 12> m_interpolators;

In hackrfoutputthread:

  Interpolators<qint8, SDR_TX_SAMP_SZ, 8> m_interpolators;

So different data-types / scaling may explain it.

[srcejon]

BladeRF uses the same interpolators, but I don't have one, to check if it has the problem.

Also, might need something with higher res ADC than RTL SDR to see the problem, as SNR needs to be more than 40dB.

[f4exb]

Yes I just tried with the BladeRF 2 at an interpolation of 16 and using the Aaronia RTSA as the receiver which cannot suspected of bad dynamic range. Even with a dynamic of 60dB on the spectrum no artifacts are visible.

[srcejon]

Ok, thanks. Will try to narrow it down some more here.

[github-actions[bot]]

This issue is going to be closed due to inactivity

[sp5wwp]

Are there any updates to this?

[srcejon]

Working on something else at the moment, I'm afraid.

[sp5wwp]

Is this bug still present?

[srcejon]

I suspect so, unfortunately.