white noise in audio chain reveals modulator issue

ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 10:56 am

Dear colleagues,

After getting complaints about dark audio I decided to feed some white noise directly into VAC1. A great freeware program for audio analysis is Friture.org.

After doing so it was clear to see that the Anan 200D flat spectrum was absolutely not flat, severely dropping above 1 kHz. Then playing with the TX equaliser I could get it a bit better but only by extremely putting all sliders below 1 kHz to minimum and above to maximum. Not reacting as expected.

Anybody an idea what might be causing the high frequency roll off ? I am running Thetis 2.8.11 x64. Something tells me it is Thetis causing this.

All processing was off. Audio enhancing stuff like CFC and levelers also :-)

Will try to reproduce this test on my Red Pitaya

Pedro ON7WP-C5WP-AA9HX
w9mdb
Posts: 446
Joined: Sun Apr 09, 2017 5:53 pm

Re: white noise in audio chain reveals modulator issue

Postby w9mdb » Sat Oct 23, 2021 12:10 pm

Interesting app. Not sure I believe it's FFT spectrum levels though.
I don't see any rolloff on my 7000 DLE MKII except where the bandpass is (as it's supposed to do).

But the Friture spectrum is wrong.
Spectrum Lab correctly shows a flat response from 200 to 3000 here but Friture shows a gradual increase from 200Hz to 3000Hz.

Try Spectrum Lab and see what you get. These pictures are looking at my antenna on an FT8 band from both programs.

Screenshot 2021-10-23 070517.png
Screenshot 2021-10-23 070517.png (15.34 KiB) Viewed 7008 times
Attachments
Screenshot 2021-10-23 070341.png
Screenshot 2021-10-23 070341.png (22.1 KiB) Viewed 7008 times
Mike W9MDB
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 12:45 pm

I am not talking about a few dB.

This is what my TX looks like taken from a -50 dB coupler on the antenna: All settings flat. White noise.
Attachments
TX spectrum.PNG
TX spectrum.PNG (91.98 KiB) Viewed 6996 times
w9mdb
Posts: 446
Joined: Sun Apr 09, 2017 5:53 pm

Re: white noise in audio chain reveals modulator issue

Postby w9mdb » Sat Oct 23, 2021 12:50 pm

What does it look like with no antenna plugged in?
What's your bandwidth?
Mike W9MDB
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 12:53 pm

apparently everybody I speak to noticed this behaviour with their Hermes Lite's etc. They all have a TX spectrum that goes down like hell.
Nobody cares, they just put their TX analysers extremely to compensate.

To my opinion this is a serious programming bug in the modulator software.

It should be flat within 3 dB. My 40 years old Kenwood TS830 is... Progress ??? :-)

I am not a DSP programmer but a CIC filter has this negative slope. Normally they use a PIR filter behind to correct this. I was told both filters were implemented in Hermes lite. Somebody made a mistake somewhere ???

Pedro.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 12:56 pm

You can see the lines of my TXbandwidth filter. 300 Hz to 2900 Hz.

It is a 50 dB coupler. 30 dB directional coupler and 20 dB attenuator. It does not matter if there is an antenna or not. The thing has great directivity....
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sat Oct 23, 2021 3:12 pm

You must be doing something wrong, or perhaps you are suffering from some sort of hardware problem.

While I can't duplicate this on a 200D, using the 8000 here I setup my DAW to generate white noise and to analyze the audio output. I also ran Friture in parallel. Under no conditions is there a problem detected.

Thetis was set up with absolutely no processing on TX or RX. Leveler, phase rotator, expander, CFC, both EQ's and COMP all disabled. Passband was 50 to 5000Hz.

On receive, no EQ's, no NR or NB. Passband was 0 to 5500Hz. I probably should have set AGC to fixed, but I don't think it would have made any difference.

While I trust the DSP and displays on Thetis implicitly, I wanted to obtain true, full duplex audio output during transmit. Since this is impossible with RX1, I set RX2 to use ADC0 (same as RX1) and disabled RX2 mute during TX. RX2 audio was fed via VAC2 to both Friture and the DAW (Reaper).

The receive signal was obtained using an Xtronic coupler that is located on the output of my amplifier. The amplifier was in standby. The 8000 was set to 25W output, however a quick look at 100W output shows no difference at that power level. I did not want to keep the radio in transmit at 100W for the time it took to obtain the screenshots. I operated into a dummy load.

On both audio spectrum analyzers spectrum weighting factors were disabled. The specific settings for Friture are shown.

As you can see from the screenshots below, the channel response through both TX and RX is essentially ruler flat. I used peak hold in Thetis to more clearly show the response inside of Thetis. Note that the apparently slow roll-off at the low end of both audio spectrum analyzer displays is due to the log scaling of the frequency axis.

Note that an external source of white noise is not necessary. In Setup > Test one can select many different test signals for transmission. The results are the same with the built-in white noise source.

Using the built-in signal generator, one can obtain a much more precise look at the spectral response of the passband using the swept tone mode and peak hold. That is shown in the last screen shot. Unfortunately neither Friture nor the other audio spectrum analyzer I used have a peak hold mode, but the Thetis displays show how perfectly flat it is. If you zoom into single dBs in the Y axis, you can see a very small ripple, but that is inconsequential. Note all of the above was using the low latency filters.

Capture.JPG
Capture.JPG (594.14 KiB) Viewed 6966 times


Capture2.JPG
Capture2.JPG (603.74 KiB) Viewed 6966 times


Capture3.JPG
Capture3.JPG (616.11 KiB) Viewed 6966 times
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 4:42 pm

exactly, this is what I would expect. Nice plots. Flat.

So the only difference in setup is you have an 8000 and me a 200...

even more puzzled now.... Hardware issue in Anan200d built on the Hermes legacy ? Alle Hermes Lite users report the same issue...

Pedro ON7WP
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 4:50 pm

Great hint, did not know the Thetis had built in white noise. Just tested, perfectly flat !!!

So the issue is related to windows audio handling or VAC...

As far as I know all my windows settings if any are flat.

Using windows direct sound.

Any more ideas ?

Pedro ON7WP
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sat Oct 23, 2021 4:56 pm

I use VAC exclusively, so it isn't that per se.

How did you generate the white noise? How did you input that white noise into Thetis?

Screen shots are always helpful.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sat Oct 23, 2021 5:13 pm

Anan 200 gets audio feed via VAC and setting "windows direct sound", coming from an USB microphone or in this case (to trace the issue of this one sounding dark) friture.org.

Friture.org white noise generator has and windows directsound output, so nothing in between...

Laptop built in microphone also has the same issue, no decent audio above 1 kHz. Microphone itself is flat as hell tested with external noise source.

Laptop is a high end Dell XPS13.Realtek audio chipset.

Could try different laptop....
But hey ????
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sat Oct 23, 2021 10:12 pm

I just checked the Friture noise output and it is flat. Friture is a cool tool, I'm liking it :)

So it's not Friture, not VAC, not Thetis, not the firmware or the hardware.

I can't judge your microphones.

You failed to mention what virtual audio software you were using to bridge the gap between Friture and Thetis. That could be important.

Do you have any equalizer software installed on your PC that messes around with things at the OS level, for example Equalizer APO?
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Oct 24, 2021 9:53 pm

Friture can output directly digitally to windows direct sound, so straight into VAC of Thetis. Click on the toolbox in "generator"

Great software not ? I use it to teach my students audio measurements at home during lockdowns.

Not using any equalisers in windows....
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Mon Oct 25, 2021 1:30 am

ON7WP wrote:Friture can output directly digitally to windows direct sound, so straight into VAC of Thetis. Click on the toolbox in "generator"

Yes, of course. And that is how I used the generator the other day to confirm it's output is flat. But I have virtual sound devices installed, specifically Voicemeeter, and that allowed me to route the Fritus generator audio to my DAW for analysis.

When the generator toolbox button is clicked the only devices shown are either actual hardware devices or virtual devices. You say you have no virtual devices. There is no way audio can be routed from Fritus directly into Thetis via VAC without a virtual device of some sort.

This issue is probably the root of your problem. Screenshots of your Fritus generator setup and Thetis VAC setup would be helpful in confirming this as the problem.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Nov 07, 2021 8:15 pm

I use the windows primary output, picture enclosed.

I get the same "dark" audio from friture as I get from my USN microphone that is perfectly flat on its own.

enclosed also my thetis audio settings, using VAC1 as you can see..


So tell me, what do you suspect now, as the built in Thetis noise generator is perfectly flat and the incoming windows audio isn't.

Pedro ON7WP
Attachments
thetis audio settings.png
thetis audio settings.png (69.62 KiB) Viewed 6589 times
friture output.png
friture output.png (40.94 KiB) Viewed 6589 times
User avatar
oe3ide
Posts: 389
Joined: Tue May 05, 2020 8:36 am
Location: JN78XK
Contact:

Re: white noise in audio chain reveals modulator issue

Postby oe3ide » Sun Nov 07, 2021 9:32 pm

Laptop is a high end Dell XPS13.Realtek audio chipset.


Hi,
I am not sure, if its topic-related, but on my tests to make my station remote operatable I ran in a similar problem.
Laptop was an Lenovo Thinkpad with built in sound card (Realtek). Microphone was a cheap usb headset.
Noisy and "dark" audio....the built in mic was even worse.

I spent a few bucks and tried it with an cheap external usb soundcard (~10 EUR). All was working perfect, no noise and a good audio...

Just a hint

73 Ernst
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sun Nov 07, 2021 10:37 pm

ON7WP wrote:I use the windows primary output, picture enclosed...So tell me, what do you suspect now, as the built in Thetis noise generator is perfectly flat and the incoming windows audio isn't.

By using the audio devices the way you have them assigned, you are allowing the OS to somehow corrupt the audio.

A better (but not simpler) way to do this is to use virtual audio software to make the connection between Friture and Thetis.

I recommend Voicemeeter Potato from VB Audio. It is free. After installing it reboot, then go into the Windows Sound Control Panel and make certain all Voicemeeter related devices are set to 48KHz.

Then in Voicemeeter you must first assign a hardware output device "A1" in Voicemeeter. Your normal PC speaker output is fine. Assign Thetis to the first virtual audio channel, "Voicemeeter VAIO". Assign Friture to the second virtual audio channel, "Voicemeeter Aux". Then make the connections between the two channels in Voicemeeter (there are a lot of Voicemeeter tutorials on Youtube).

Once you have started to use Voicemeeter you may wonder how you ever got along without it.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Tue Nov 09, 2021 8:08 pm

Hello Ernst and WU2O (can I have your name...).

Ernst, external card is a no-go. No room for such when wandering around with my laptop... But U know the external USB devices are brilliant, I use them with my students for audio processing at high adc bit conversion.

WU2O, thanks for the hint, will try Voicemeeter Potato, but still wondering how OS can manipulate audio when no equalizers installed anywhere. Would like to know where this happens. Never had any issues routing audio inside my laptop, just with Thetis.
Maybe we can have a chat one of these days on HF ? Are you around somewhere ?


Pedro Wyns ON7WP-C5WP-AA9HX
https://www.linkedin.com/in/on7wp/
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Fri Dec 31, 2021 10:31 am

Hello guys,

The issue is still there and has been confirmed by other people using Thetis with Hermes boards.
WHY is the audio not flat coming from W10. Just tried a new install W10, no audio tweaking whatsoever. Internal Anan white noise is perfect, the VAC audio is going down significantly with increasing audio frequency.

Any ideas.

Try it yourself with friture.org as audio source.

Pedro C5WP - ON7WP - AA9HX
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: white noise in audio chain reveals modulator issue

Postby DL2XY » Fri Dec 31, 2021 11:46 am

10kc gaussian noise over Muchenko VAC, full duplex monitor:


noise.png
noise.png (425.59 KiB) Viewed 5950 times


Rig was 200d (orion), also tested on 100d (hermis) - same result.


Just tried a new install W10, no audio tweaking whatsoever.


Maybe that's the problem. Window is doing everthing to transfer sound streams from the source to the sink, no matter what format manipulations are neccesary to do this.

Some hints to prevent quality degrations:

  • Dont use Windows default (standart) Interfaces, these are running trough windows internal mixers and converters.
  • Configure all used Interfaces to exclusive access.
  • Disable all sound extensions and effects on these interfaces.
  • Configure all interfaces to the exactly same format (Samplerate, nbr. Channels, bit resolution).
  • Configure the virtual audio interface to the same format and use the setting "Cable Format".

vac.png
vac.png (11.85 KiB) Viewed 5950 times


audio.png
audio.png (14.04 KiB) Viewed 5950 times


Walter
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Fri Dec 31, 2021 3:09 pm

ON7WP wrote:Hello guys,

The issue is still there and has been confirmed by other people using Thetis with Hermes boards.

And who would those people be? Why are they not posting in this topic, here? If you get them to actively participate in the discussion, perhaps the mystery might be solved.

WHY is the audio not flat coming from W10.

Walter and I have gone out of our way to try to replicate your results. Neither of us can. Nobody else on this forum, the largest discussion group for Apache hardware and openHPSDR code by far (1312 members vs. 283 in the groups.io forum as of this time/date), has reported a problem like this, or sought to take part in the discussion. That they are not taking part is not a sign of uncaring or rudeness, it's a sign that nobody has anything to offer on the subject, i.e. people are not experiencing it and can't begin to guess why you are.

Try it yourself with friture.org as audio source.

Already done, both with Voicemeeter and, as you do, with Windows as the connection mechanism. In neither case could I duplicate your results, as already discussed above.

You did try the internal signal generator on Thetis and got good results. Have you tried installing either Muzy VAC (like Walter uses) or Voicemeeter (like I use) and connected Friture to Thetis like that? Try that and see if it solves the problem for you. It might provide additional clues to solve this mystery.

Definitely do not play signals into a microphone to make tests like this. Any such results are hopelessly colored by speaker and mic frequency response characteristics.

Let it be clear: there is, so far, zero evidence of an Apache hardware problem, an openHPSDR software/firmware problem, or a problem with Friture, Muzy VAC or Voicemeeter. I don't know if Walter has tried your approach to connect Friture, but I have and can find no problem with Windows either.

73,

Scott
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 9:44 am

Scott, Walter,

thanks for the hints. It is /was never my intention to appear rude but having a totally shitty modulation with a 3000 euro radio renders me desperate...

A group of people gathering on 80m I often speak with using Hermes Lite radios report the same issue. They are not on the internet. They all have there equalizers raped to enhance high frequencies dramatically to result in a flat curve. They are happy with that. I am not.

I don't have this issue when running SDR console on my LimeSDR.

Walter one more question. WHERE do I find the VAC control panel ??? I cannot find it in the Thetis manual.

Pedro
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 9:47 am

so the only difference is you two use voice meeter or the Muzy VAC, and I am not. So it must be there. (I guess walter showed the control panel of Muzy VAC ?)

will try voicemeeter but I do not like all these additional programs if something is built-in. Voicemeeter is overkill for this application to my idea... But hell yeah I will try.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 1:12 pm

Hi scott,

tried this voicemeeter thingy but I cannot manage to route the audio from friture to Thetis. Weird program, when I switch its output the virtual input disappears... No idea how and what to select, ASIO, VIAO, WDM, MME... Lost in all these standards...

What output I set my Friture to, where/how do I take this Friture white noise into voicemeeter and how do I route voicemeeter output to thetis ? This voicemeeter is very confusing with two locked hardware inputs and one virtual input that changes if you change the output...

Normally I route a USB microphone via windows direct sound to Thetis. But then I have this integration effect on my spectrum, where it all started...

I loved these patch panels for audio in our outside broadcast vans. IP screwed up a lot... :-) I must be getting old....

Pedro
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 1:32 pm

Managed to route my audio test generator to thetis via voicemeeter. Now the frequency behaviour is flat but I did discover another bug. The displayed TX spectrum is totally wrong, it even moves... as you can see from the screenshot.

Did not do anything and then things turned OK.

But to summarize. I cannot route the white noise of Friture via voicemeeter but I could do with single tones from audio test and now things are flat. So what is going wromg with the windows directsound setting ???
thetis TX spectrum.jpg
thetis TX spectrum.jpg (232.88 KiB) Viewed 5819 times
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 1:46 pm

slowly getting there. Voice meeter uses ASIO. The audio integration issue is in VAC. ASIO is a totally different way of submitting audio to the radio.
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sun Jan 02, 2022 2:22 pm

Voicemeeter will use nearly any driver available: MME, WDM, Directsound, WASAPI and, of course, ASIO. You can choose whichever you like. It will also happily bridge between them.

I hope you are using your 200D for these experiments. There is no way for us to vouch for what the Hermes Lite firmware may or may not be doing.

You have to be doing something wrong in your settings. Please do the following:

1. Perform all experiments with the 200D.
2. Make sure DUP mode is ON.
3. Use DIGU, this will ensure there is no audio processing you have inadvertently left enabled in Thetis.
4. Post a screen shot of your Voicemeeter configuration (the main mixing panel UI, all of it).
5. Post a screen shot of your Thetis VAC setup.
6. Post a screen shot of the Windows Sound Control Panel Advanced Settings tab for each Voicemeeter device you are utilizing, both the record device and the playback device. The intent here is to make certain all those devices are set to 48KHz sampling rate. This should be 4 devices total.
7. Post a screen shot of your Friture setup.
ON7WP
Posts: 25
Joined: Sun Apr 18, 2021 7:42 am

Re: white noise in audio chain reveals modulator issue

Postby ON7WP » Sun Jan 02, 2022 10:22 pm

Hi Scott,

I am doing all tests on my Anan200D. I just mentioned that my friends with Hermes lite have the same dark audio phenomenon with Thetis in the windows directsound settings...

I managed to get friture white noise running into the Anan via voicemeeter. All fine. USB micro too, all fine in ASIO setting. No high frequency detoriation. Flat as the chest of Sinead O Connor :-)

The problem only happens when using the very same USB micro (or friture) as primary windows audio source and then VAC1 activated with this source in Windows direct sound. Dark audio... All equalizers off. THAT is the problem. No audio tweaking in windows, no plugins or whatever crap. But shitty audio. Must be on the VAC end of windows but no idea how to access this. Don't have the Russian VAC thingy.

Starting to learn and appreciate the voicemeter app but it does not like on the fly reconfiguring... Reboots necessary to get audio stable again.

Pedro
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: white noise in audio chain reveals modulator issue

Postby w-u-2-o » Sun Jan 02, 2022 11:22 pm

If you are familiar with broadcast vans then using an audio mixer must also familiar to you. The various flavors of Voicemeeter should not be a huge learning curve, then. Glad you are starting to like it!

Voicemeeter should be perfectly stable across changes. That it is not for you still suggests something wrong with your setup. I therefore still encourage you to post the screenshots requested above for review.

Window knows nothing about VAC. That's not how it works. The VAC interface in Thetis merely allows Thetis to choose audio devices, real or virtual, as the source or destination for Thetis audio.

Return to “Thetis”