Tx Audio quality? And viewed on Thetis Oscilloscope

[JohnFT61]

My microphone input seems to be working OK now, at least there is no delay after PTT. I may still have microphone issues (plugged into the front or back). I will know more after tomorrow, when I receive a cord to connect my high-quality Test Tone Generator from the MicroHam Smorf RF Vector Signal Meter. It has a programmable Test Tone Generator that can go down to 1uV! I will connect it to the front microphone input for the same tests.

I have attached Thetis screenshots from various frequency inputs via speaker to microphone element. The element may be the issue, I have so far tried this element and an Elecraft MH3 hand microphone. Both yield strange waveforms when viewed on the Thetis Oscilloscope! I have attached plots. When talking to people, they say I am very loud, but something sounds not just right!

I will send an update after I run the same tests with the MicroHam unit tomorrow, after I make the cable for proper connection.


73
John
N2KBE

[w-u-2-o]

Strongly recommend you rewire the microphone with a TRS connector instead of the TRRS connector you are currently using. There is no guarantee the TRRS is making proper connection to the TRS jack.

[JohnFT61]

One test I was using a TRS connector and just the microphone element, so it was correctly wired and making contact with the Mic Connector. It was from an old Logitech Microphone for PC (very old). Enabled using MOX.

As for the Elecraft MH3 Mic, the G2 transmits when I push the PTT, and there is Mic audio, so I am sure it is making the right contacts. The Mic audio is always hot.

I would bet that there is a good chance the issue is caused by the Mic elements, that they are not being biased correctly. I have another microphone I will try on Friday, after I receive some hardware. It requires 8V to one of the pins on the RJ45 female connector.



73
John
N2KBE

[JohnFT61]

OK, I just tried another microphone. Same results. It uses a 8V source, in whitch I used a battery. It didn't need Mic Bias. I noticed when I lower the Mic Gain or turn off 20dB Boost, it just appears to lower the amplitude, and the distorted signal is still present, but changes amplitude.

I also noticed that with the microphone removed from the front panel, and hitting the Tune button, there is also a distorted signal on the Oscilloscope! It only appears when there is some RF Drive (1 or more). Interesting?



73
John
N2KBE

[W7GES]

As long as I'm in my shack working on getting my G2 integrated into my SO2R station I gave this a try.

This screenshot is from a cheap computer headset microphone plugged into the front panel jack. I didn't mess around with the level to get the correct amplitude but just wanted to show the scope signal looks pretty normal.

I'm transmitting a couple of watts.

G2 with front panel...

George / W7GES

2023_0823_Screenshot.png (699.19 KiB) Viewed 11055 times

[W7GES]

This time I used the XLR / Balanced mic jack on the rear panel and boosted the mic gain...

BTW, this is Whiskey Seven Golf Echo Sierra spoken over and over. I've seen the waveform enough times to say that it looks normal to me...

George / W7GES

2023_0823_xlr_scope.png (702.81 KiB) Viewed 11051 times

[JohnFT61]

Hello George,

Can you set your Scope Mode to 25 or 50uS (Setup/General/Display)? That is not detailed enough.

Also see what it looks like in Tune?


73
John
N2KBE

[W7GES]

Hello George,

Can you set your Scope Mode to 25 or 50uS (Setup/General/Display)? That is not detailed enough.

Also see what it looks like in Tune?


73
John
N2KBE

Hi John, I’ll try that tomorrow. I just shut down the shack and disconnected the antennas as I received a “lightning nearby” warning. Also, my wife should be home from work any time now so I will be getting the “time to go out to dinner” warning as well!

[JohnFT61]

OK,

Thank you!



73
John
N2KBE

[W7GES]

Top screenshot is using my Sennheiser headset mic bottom screen shot is Tune enabled. Scope time is set to 50uS.

73,

George / W7GES

[JohnFT61]

Hello George,

Thank you for the testing with your G2!

I don't believe it should look that bad, especially in Tune Mode, it should be close to a pure Sine wave, depending on how clean it is. It is only a single carrier, or at least it should be! Take a look at my screen shots, one is in Tune Mode, and the other is an oscilloscope display showing a signal being received. You can see how the received signal is made up of many Sine waves at different frequencies (very clean). There are no triangular or square waves (distortion/not pure). Doug did the same test on a G2 he has, and got the same results.

I have a cord to connect my Test Tone generator in my MicroHam RF Vector Signal Meter to my G2's microphone input. It has very low THD, and will go down to 1uV. I will run more tests.

[KC2QMA]

I wonder if the scope distortions are artifacts of the the GUI display or actual distorted output?
I think the next step is to hook up some test gear and see if the actual RF output has distortion/MID.

It may be that GUI display has trigger issues with some types of input but I'm just guessing

[JohnFT61]

As far as I can tell, it seems like the issue exists with the Transmit Oscilloscope. The Mic input needs to see a higher level than my Test Tone Generator can output. When using PC speakers with a microphone, and the TrueRTA program, it appears that the output of the radio is OK using an oscilloscope across a 50 ohm load. I believe the Tx oscilloscope is the issue. The Spectrum and Oscilloscope display work great during receive. Why can't the Spectrum and Oscilloscope work as good during Transmit?


73
John
N2KBE

[w-u-2-o]

I just spent a couple of hours looking at this issue. As far as I can tell, scope mode during TX is badly broken and I wouldn't trust it.

I used both VAC and an external software signal generator as well as the sig. gen. built into Thetis (Setup > Test). Results were identical with both.

I could only get reliable results by setting the scope trace width to 10μS and a 1KHz test tone. Using those settings the resulting display showed 10 cycles of sine wave so 10μS seems akin to 10ms. However, thinking that I might be under or oversampling, changing the trace width to 10,000 showed a great many cycles on the display. Changing the test tone to anything other than 1KHz gave a wide variety of bad results. You would think that making logical changes like changing to 2μS and 200Hz would show 10 cycles but it doesn't work at all.

Some results were as expected (but only at 10μS and 1KHz). For instance, using a complex TX profile and USB/LSB did show some distortion of the waveform but one would expect that. When changing to DIGL or DIGU the waveform was a very clean sine wave. PureSignal also had no effect and the WDSP Guide implies the scope is tapped off before predistortion so that makes sense. And amplitude levels seemed to track along with ALC.

[laurencebarker]

It might be worth bearing in mind that Protocol 2 TX works with a 192KHz sample rate, so the sample period is approximately 5us. If you looked at highly "zoomed in" oscilloscope settings, you will see the individual samples and that can cause confusion.

I always recommend colleagues to select "random" frequency values for signal generators rather than round numbers. That way, you don't get the same set of samples on the next cycle of the waveform. So select 956Hz rather than 1000Hz, for example.

[w-u-2-o]

It might be worth bearing in mind that Protocol 2 TX works with a 192KHz sample rate, so the sample period is approximately 5us. If you looked at highly "zoomed in" oscilloscope settings, you will see the individual samples and that can cause confusion.

I always recommend colleagues to select "random" frequency values for signal generators rather than round numbers. That way, you don't get the same set of samples on the next cycle of the waveform. So select 956Hz rather than 1000Hz, for example.

None of that seems to apply to this problem, Laurence.

For instance, one would expect that with a 1KHz tone and a Scope Mode Time (the tooltip explains this as "Amount of time to display across the width of the scope display window" of 10,000µS one would see exactly 10 cycles displayed. Alas, this is what you get:

Capture1.JPG (76.94 KiB) Viewed 10745 times

You don't see 10 cycles until you reduce the Scope Mode Time to (ostensibly) 10µS (ignore the scope refresh artifact captured by the screen shot software circled in blue):

Capture2.JPG (119.24 KiB) Viewed 10745 times

When you change the test tone to 956Hz you get this aberrant response:

Capture3.JPG (114.23 KiB) Viewed 10745 times

[laurencebarker]

Are you sure it doesn't apply? I suspect the 956Hz plot has just revealed that the overall trace is made up from several others; having a round 1000Hz (so the waveform always repeats exactly) hid the joins.

[w-u-2-o]

I have no idea what it's doing. But what it is doing is not correct. If this is intended to be a time domain display of the audio (as opposed to RF or IF), and given the timebase settings available (1 microsecond or larger) it appears that it is, then given a sinusoidal tone on the audio input it should display that waveform correctly in accordance with the timebase value specified.

It would also be great if there was an adjustable trigger level value as well such that a stable display could be obtained, rather than the free running timebase that exists now.

[JohnFT61]

It would nice for the Transmit Oscilloscope to work correctly, and also for a trigger level to be added to the oscilloscope!



John
N2KBE

[ramdor]

https://youtu.be/wRxkV25XBuY?si=rD614HYMH5ceU3xx

will be in 2.10.3.4 at some point

[Dan]

The procedure I use:
Verify wiring is correct using a TRS connector.
Record the audio in another receiver, then listen to it.
Adjust it until it sounds like me in person
Use the recording to verify the scope looks good.
Yes I think listening is a better test than looking at the scope. Often times I will hear issues a scope does not reveal.
If you have the ability to set up a second or two of latency that can be used in place of recording and playback.

[Trucker]

Amazing addition Ritchie!

[ramdor]

Amazing addition Ritchie!

now in the 2.10.3.4-pre-release-5 version James. Cheers.