Audio nonlinearity

w9mdb
Posts: 446
Joined: Sun Apr 09, 2017 5:53 pm

Audio nonlinearity

Postby w9mdb » Tue Mar 03, 2020 5:07 pm

I installed 2.6.9 1/25/20 and was calibrating my audio from WSJT-X.

Anything above -3.0dB on WSJT-X is being essentially ignored by Thetis and not raising the signal level.

I suspect the last bit in the 16-bit audio value is having a problem during conversion.

I always calibrate WSJT-X at full scale and then -3.0dB should (and previously has) decreased signal (watts) by 50% as it should.

de Mike W9MDB
Mike W9MDB
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Audio nonlinearity

Postby w-u-2-o » Tue Mar 03, 2020 8:16 pm

I can not duplicate your problem. I am running 2.6.9 1/22/20, but I doubt that slight change in version date makes a difference.

I am also using Voicemeeter Potato 3.0.0.8, and WSJT-X 2.1.2.

In DIGU mode, VAC TX and RX gain set to 0dB, VAC set to "Stereo", all relevant sound devices set to 16 bit and 48KHz in the Windows Sound Control Panel, the results are "dB perfect". Set WSJT-X TX output level to 0dB, and the mic meter in Thetis reaches 0dB when asserting TUNE in WSJT-X. Set the WSJT-X control to -3dB, and the mic meter reads -3dB.

You probably just need to go through all of the various audio settings in Thetis, Windows and whatever you use for virtual audio connections and groom them carefully.

Post screen shots of all of your settings if you want (use the Attachments tab below the text entry window here on the forum, then after adding the files use the "Place Inline" feature to put them into your post.
w9mdb
Posts: 446
Joined: Sun Apr 09, 2017 5:53 pm

Re: Audio nonlinearity

Postby w9mdb » Tue Mar 03, 2020 11:34 pm

The MIC level does increase the last 3dB as expected but no more power out of the rig.
I'm only running 7W from the Apache 100 into my amp so it's not like it's any limit.

What's the next stage after the MIC level?

Mike
Mike W9MDB
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Audio nonlinearity

Postby w-u-2-o » Wed Mar 04, 2020 12:19 am

There is no stage after MIC level if you have DIGU selected, as you should for any digi-mode. All audio processing except TX gain is disabled.

I tested your use case here. I set 7W (indicated by the Thetis power meter) drive power with 0dB audio level. With my KPA500 amp, I got 120W indicated on the KPA500 RF ouptut power meter at 0dB audio level. At -3dB I got 60W.

For any given setting of the RF drive control, power output with audio drive should be almost perfectly linear.

The RF drive control itself is badly non-linear, meaning that if you calibrate (Setup > PA Settings) for 100W at RF drive = 100, then at RF drive = 50 you probably won't get 50W. This is a hardware limitation, not a software or firmware problem. The output of the DAC is almost perfectly linear. But the linearity of the RF amplifier stages is very poor, and the software has no way of knowing what it actually is.
W4WMT
Posts: 325
Joined: Sun Apr 09, 2017 10:12 pm

Re: Audio nonlinearity

Postby W4WMT » Wed Mar 04, 2020 10:42 am

I know zip about WSJT, but if it's sending full-scale samples to VAC1 in stereo that could be a problem. The stereo samples get summed arithmetically before being sent along to the transmitter. You could try disabling "Stereo" in the VAC1 setup form, in which case only the left channel is sent to the transmitter.

That doesn't explain why things changed relative to your earlier version of Thetis, I'm quite certain that no code was changed in VAC1.

When in doubt, delete the database.xml file and try again :-)

Which middleware software are you using in between WSJT and VAC1?

73
w9mdb
Posts: 446
Joined: Sun Apr 09, 2017 5:53 pm

Re: Audio nonlinearity

Postby w9mdb » Wed Mar 04, 2020 3:10 pm

I always use just the Left channel on input & output -- summing the two makes no sense for WSJT-X for either input or output.
I just tested with Stereo checked/unchecked and Combine VAC channels checked/unchecked and see no difference.

VB Banana AUX control panel shows two channels active on Input Levels and one channel active on Output Levels. When transmitting from WSTJ-X only channel 1 has any data in both.

Mike
Mike W9MDB
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Audio nonlinearity

Postby w-u-2-o » Wed Mar 04, 2020 5:43 pm

Mike,

The most likely scenario is that something is misconfigured somewhere, because both PowerSDR and Thetis definitely work properly for others in this respect. Again, post some screen shots.

73,

Scott

Return to “Thetis”