UniTrunker Scope Levels

[MrBandit]

Hi,

I'm running dual TCXO NooElec dongles with UniTrunker to decode P25. I'm getting acceptable decoding, but it could be improved. The Scope view of the Voice and Signal VCOs appear to show slight clipping. Is it possible to lower the output of the dongles?

Thanks,
WGR

 

[bassjunkie]

Play around with the gain settings in the info tab - it can help to turn off all the auto gain settings and manually tune your gain until you get good levels.

 

[SCPD]

Play around with the gain settings in the info tab - it can help to turn off all the auto gain settings and manually tune your gain until you get good levels.

The waveform you see in the 'scope tab is the result of FM demodulation. Changing receiver gain does not alter the waveform's audio level.

Is it possible to lower the output of the dongles?

I'll look at adjusting the scale of the discriminator output.

 

[thegriff]

I'll look at adjusting the scale of the discriminator output.

Thank you in advance! I've been running a custom leveling program between the output of UT and the input of DSD+ to reduce the "volume" of the signal going to DSD+. Having this functionality built in would be great.

 

[SCPD]

Hi Griff;

Just to be clear - you should not need to adjust the audio level coming out UniTrunker. Assuming VAC or VB Cable does not touch the audio but merely pass it through as-is, it should be at an optimum level. The scaling I refer to above pertains specifically to FM demodulation in software.

 

[thegriff]

For whatever reason I do have to touch the audio levels, if I don't DSD+ won't decode and when listening to the signal through headphones I can tell it's distorted. When I cut the level between UT and DSD+ I get decoding and the distortion goes away.

The scope in UT shows constant clipping on my local system but if I set gain too low it won't decode due to the high noise floor of the dongles. Maybe I need to save up for an Airspy and loose the voice dongle I'm currently using. ;-)

While on the topic of scope, could you post some captures of what a non-clipped signal should look like?

 

[SCPD]

The top two images you posted were signal role VCOs. The bottom image was likely a voice role VCO at the end of call. Had you snapped an image while a call was active, you'd see the waveform passed to DSD/DSD+ is different.

The clipping you see in the top two is effectively band-width limiting the waveform. It makes the control channel decoder happy but DSD+ does not seem to like it - which is why a different waveform is passed by the voice role VCO.

I can post some images when I'm at my desk.

 

[thegriff]

The screenshots were posted by another member but I'm attaching one of my voice receiver I just captured. To my untrained eye it looks like the waveform is getting clipped but I could be wrong.

Just to be clear none of what I'm saying/asking is a criticism of UT. I'm just trying to better understand how to get the system decoding as best it can and what I've noticed has made a difference in my setup.

 

[SCPD]

Thanks Griff. You saved me a short trip.

You can see the waveform is quite different from the earlier two. There is a small amount of clipping at the far edges. I don't *think* this bothers DSD+ but Mr. DSDPlus would have to speak up on this.

The 'scope images are the waveform before it has passed through VAC or VB Cable. If the waveform looks different at the DSD+ end - something else is responsible.

Generally speaking, for good CQPSK, the main concern is the distance between the tall and shallow peaks - whether positive or negative. Feed the audio into UniScope where you should see a clean eye pattern - three "eyes" lined up vertically. Fuzz along the edges is fine as long as the centers are clear. Any decoding app should get very good results as long as the "eyes" are clear.

 

[MrBandit]

Thank you for looking into the scaling of the FM modulation!

 

[KA1RBI]

Thanks Griff. You saved me a short trip.

You can see the waveform is quite different from the earlier two. There is a small amount of clipping at the far edges. I don't *think* this bothers DSD+ but Mr. DSDPlus would have to speak up on this.

The 'scope images are the waveform before it has passed through VAC or VB Cable. If the waveform looks different at the DSD+ end - something else is responsible.

Generally speaking, for good CQPSK, the main concern is the distance between the tall and shallow peaks - whether positive or negative. Feed the audio into UniScope where you should see a clean eye pattern - three "eyes" lined up vertically. Fuzz along the edges is fine as long as the centers are clear. Any decoding app should get very good results as long as the "eyes" are clear.

Hey Rick

These waveforms look a lot to me like garden variety LSM(style) CQPSK, (i.e., not C4FM). It's puzzling to me why Unitrunker and DSD[+] and all the rest are still stuck in this last-century way of doing business - now that the signal is available in the preferred (pre-demodulated, complex QPSK) form from the SDR, why not take advantage of it? Why is everyone still slinging around FM-demodulated "signals"?

AFAIK, Denny's sdrtrunker work is the only project with proper CQPSK demodulation (outside of OP25, which pioneered it)...

73

Max

 

[SCPD]

Hi Max;

Why is everyone still slinging around FM-demodulated "signals"?

It's on my ever growing list of things to do. It will take me some time to read through the papers (what you've published and others) - to properly understand and implement.

In the mean time, I'm able to get by with adjusting gain levels and moving the antenna to a spot with better reception.

Right now I've got two crash scenarios (kinda obscure ones but still) I need to fix.