开云体育

ctrl + shift + ? for shortcuts
© 2025 开云体育
  1. QRPLabs
  2. Topics

QMX+ SSB audio distorted on one band only

 

Hi,
I installed the 1_01_003 beta firmware yesterday and I am amazed at how it has transformed my QMX+ (rev 2 board).
?
I am, however, perplexed at the fact that the SSB audio is distorted (on another receiver) only on 20M. This only occurred after calibration and the audio becomes absolutely clear again after disabling the phase predistortion. All other bands have good audio with and without predistortion, and the RF output is within normal limits on all bands, although 20M is a little lower than others at about 3W. I am using a 12V stabilised PSU and a good dummy load, via a Bird isolator, giving me a solid signal into my receiver.
?
Trying to think of an explanation made me consider that the LPF used for 20M (LPF 3) may not present a good load to the PA, so I looked at the LPF sweep using PuTTY. Here is the sweep for 20M:
The 14MHz line is in a less than ideal part of the curve, whereas the bands which share this LPF are better placed. Is this significant?
?
The phase calibration sweep for 20M also looks odd and is different to the others which are all nice curves (except 160M which sounds fine!):
?
This is the 20M USB IMD sync optimisation curve, the title of which has been overwritten by the asterisks of the curve.
?
The 20M USB curve looks identical:
?
Here is the adjacent band 30M curve for comparison:
?
Everything on the board looks good and the radio otherwise works perfectly. I am perplexed at why only one band is distorted (it really is unreadable) yet 160m, which has similar calibration curves but a great looking LPF, Sounds OK.
?
Any ideas?
?
The simple answer is to switch off predistortion, but that doesn't solve the problem!
?
73
?
Ellis
GM4GZW
?
?
 

Ellis
?
Distortion on one band only could be RF getting into audio. See if same thing happens when transmitting into dummy load. And maybe an external signal report.?
?
Curt
?
 

On Thu, Mar 27, 2025 at 05:19 PM, Ellis Simon GM4GZW wrote:
The phase calibration sweep for 20M also looks odd and is different to the others which are all nice curves
Based on some experimentation with some similar phase calibration results on a couple of bands on two of my QMX, it appears to me that when the phase distortion curve is started, but not completed, then the audio gets distorted.? And the lower the value it stops at, the worse the distortion.? In your example, the curve of values only goes to about a value of 600 on the horizontal DAC axis.? So my theory is that whenever your voice waveform has an amplitude that results in any DAC value of over 600, there is not a proper result stored in the calibration table for those values, and 'bad stuff happens', resulting in distorted audio.
?
So (if my theory is correct, and it seems to have worked on my 2 QMX), if you can get the calibration to put in measured phase values across the whole range of the DAC (about 2600 for 12V builds and 2000 for 9V builds), then the resulting voice quality is good, even if the graph looks uneven or strange.
?
I was able to get my calibration to finish on the 'bad' bands by adjusting (as others have noted) the "Frequency min" and/or "Frequency center" in the band configuration screen.? For some reason these do affect the way the phase calibration runs.? I had to play with different values in one or both of those settings, but without too much trouble got all of the bands to have "complete" phase distortion curves.? The 'curves' were not always pretty, but I don't think that's so much the issue as having data initialized for the full range of DAC values? that will be used (based on the input voltage and drive level, both of which Hans pointed out need to be set before calibrating).
?
After I changed the band config settings for calibration, I restored the band settings to their original values so the other diagnostic sweeps would still function in the same way, and so my default 'center frequency' would be what I want on each band.? [These settings don't affect normal transmit/receive operation, we just needed to 'seed' them differently to get completion of the calibration table.]
?
I hope this is helpful to someone, and maybe someone can corroborate my theory on getting rid of the distorted audio....
?
Stan KC7XE
 

开云体育

Stan,

This seems to have worked for me. Thank you! It took some time playing with the limit/center values to fix 17,15,12, and 10 meters. 160 and 11 are still broken but I don't use either of them.?

Tony
AD0VC

From: [email protected] <[email protected]> on behalf of Stan Dye via groups.io <standye@...>
Sent: Thursday, March 27, 2025 8:32 PM
To: [email protected] <[email protected]>
Subject: Re: [QRPLabs] QMX+ SSB audio distorted on one band only
?
On Thu, Mar 27, 2025 at 05:19 PM, Ellis Simon GM4GZW wrote:
The phase calibration sweep for 20M also looks odd and is different to the others which are all nice curves
Based on some experimentation with some similar phase calibration results on a couple of bands on two of my QMX, it appears to me that when the phase distortion curve is started, but not completed, then the audio gets distorted.? And the lower the value it stops at, the worse the distortion.? In your example, the curve of values only goes to about a value of 600 on the horizontal DAC axis.? So my theory is that whenever your voice waveform has an amplitude that results in any DAC value of over 600, there is not a proper result stored in the calibration table for those values, and 'bad stuff happens', resulting in distorted audio.
?
So (if my theory is correct, and it seems to have worked on my 2 QMX), if you can get the calibration to put in measured phase values across the whole range of the DAC (about 2600 for 12V builds and 2000 for 9V builds), then the resulting voice quality is good, even if the graph looks uneven or strange.
?
I was able to get my calibration to finish on the 'bad' bands by adjusting (as others have noted) the "Frequency min" and/or "Frequency center" in the band configuration screen.? For some reason these do affect the way the phase calibration runs.? I had to play with different values in one or both of those settings, but without too much trouble got all of the bands to have "complete" phase distortion curves.? The 'curves' were not always pretty, but I don't think that's so much the issue as having data initialized for the full range of DAC values? that will be used (based on the input voltage and drive level, both of which Hans pointed out need to be set before calibrating).
?
After I changed the band config settings for calibration, I restored the band settings to their original values so the other diagnostic sweeps would still function in the same way, and so my default 'center frequency' would be what I want on each band.? [These settings don't affect normal transmit/receive operation, we just needed to 'seed' them differently to get completion of the calibration table.]
?
I hope this is helpful to someone, and maybe someone can corroborate my theory on getting rid of the distorted audio....
?
Stan KC7XE
 

Thanks Curt.?

it was all on dummy load. I’ve not transmitted SSB on an antenna yet!
?
Ellis
 

Stan,
Many thanks for this. It makes complete sense in my head. Can you give me a clue as to which direction you changed the band sweep values?
?
Ellis
 

Update.
?
After a lot of fiddling with the 20m band edges I ended up with this:
?
?
Which leads to this:
?
Which seems to have cured the distortion.
?
I have no understanding of what the frequency numbers represent as changing them did not seem to change any of the diagnostic filter sweep widths. I found very little improvement with changing the minimum frequency or the centre frequency. The best change involved increasing the maximum frequency and placing the centre frequency near the top of the range, as you see from my 20M column.
?
I hope that this may help someone else. Thanks for your help with this.
?
73
?
Ellis
GM4GZW
 

I'm wondering if there is a correlation between the band edges and centers settings improving the curves and the actual LPF peaks.?
--
73, Dan - W2DLC
 

I had a quick look at the 20M LPF curve and it looked the same. I didn't examine it carefully though.
?
Ellis
 

I would expect they would look the same, I'm just wondering if the predistortion curve is related to the peak amplitudes permitted by the LPFs. Maybe moving the band edges and centers to center on the peak of the LPF curves is the reason for the changes in the predistortion curves.
--
73, Dan - W2DLC
 

I could only get calibration on 80m to get beyond ~20% by using all the configuration parameters from the 40m band configuration for the 80m band, both frequencies and LPF.? Still required a lot of tweeking!
 

Hi Dan,
?
Should we be adjust band parameter just so that we can get a complete configuration run?
Because we cant get a complete configuration run, does that not signify that something is not quite right? i.e. LPF or BPF needs adjusting?
?
We're tweaking band configuration just so that we can get the 'right' result.. Are we going the right thing?
?
Mel. M0KMD.
 

On Fri, Mar 28, 2025 at 02:26 AM, Ellis Simon GM4GZW wrote:
Can you give me a clue as to which direction you changed the band sweep values?
I think it will be different for each QMX.? It may be associated with LPF parasitic issues or whatever.? I have no real clue why/how these settings affect the calibration.?
?
But please everyone, read this:
After you successfully do the calibration,? please return the band edges to their original settings, or at least put the center frequency within the ham band allocation and the band edges equal to or greater than the ham band edges.? Otherwise you will later find that something is not working as expected.?
?
Stan KC7XE?
?
?
?
 

Mel,
I don't know, but I'm wondering if there is a relationship.
--
73, Dan - W2DLC

1 - 14 of 14