It’s caused by the algebraic sum of the SSB audio envelope from filtered PWM and SSB PLL SI5351A, square wave through capacitor coupling. For each clock cycle, at logic level 0v, the gates see only SSB envelope voltage (threshold gate voltage is maintained at menu 8:2 PA bias) and at logic level 5v , the BS170s are fully ON). This combination creates a 50% of restored SSB envelope interleaving with 50% CW, that reflects your experiments with single and dual tone. At the beginning, PE1NNZ had used a cascaded modulation in his QCX-SSB prototype (drain modulation). The BS170 gates are clocking with PLL SSB carrier while drains Vds SSB audio modulation is performed by a different transistor. On my RU-vid channel I have posted 2 videos showing performance of gate and drain modulation. Spectrum of gate modulation has a clearly visible carrier frequency and poor linearity.
Thanks for the 'peer review' John. I was under the impression that my problem may be unique, now I'm not so sure. :-) I'll link this vid in my reply to the lengthy comment I just had from Manual over on my vid.
I just read DL2MAN's comment on your latest video. I will re-perform this IMD test later today, changing various settings in the (tr)uSDX as I do so, including rx bandwidth settings. FWIW, this was a pre-built unit bought from China. I hope viewers will note that I am happy enough with my (tr)uSDX that I recently bought another one, this one for the higher bands and in kit form. I have updated my posting of IMD performance on my site here: ae5x.blogspot.com/2022/11/video-transmitter-imd-of-trusdx.html with images of spectral plots at both minimum and maximum BW settings on the (tr)uSDX.
@@AE5X I think I'm zeroing in on the source of the trouble. Have you done a firmware update on yours since it arrived? If so, what are your pa bias min and max settings? EDIT: Wait, nope, that's not it.... I just fiddled around with the settings and it didn't change the clipping. Of note though, my radio slipped off the shelf and dropped about 8 inches to the desk with a thump. Now the distortion is not nearly as bad. Still there, but much improved. I guess I'm back to examining the hardware for yet another poor connection somewhere.
@@AE5X You'll want to see this from Manuel. He's going to work the problem and we'll figure out the cause. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-RsZIJtpEM64.html
@@AE5X I updated the boot loader and firmware on one of my radios, and just the firmware on the other. They both have identical results. It appears that the mic gain is reduced with the firmware re-load. The splatter was reduced, as the TX output was lower. So I changed the TX drive from 1 to 4 to get the power up to about 3+ W. The 3rd order IMD wasn't changed with any of the changes I could make, supply voltage, TX Drive, etc.. I was able to make some contacts, so it does work on SSB but the IMD problem needs a solution. I think it is a big part of the problem. I added an RF "shield" between the two boards, but it did not help. Is there something driven into a non-linear region in the audio, causing the IMD problem or is it in the firmware/processor? Other than this IMD problem, I like the radio and appreciate the creative design.
It's a shame that the transmitted audio from these little radios sounds so incredibly nasty on SSB. I only use mine for CW and was aware of the grim SSB performance when I bought it. It's fine as a multi-band CW rig but I wouldn't subject anyone to that awful SSB.
Then the thing to do is to clean it up. I figure we're hams, not CBers. Even with just a tech license we're licensed to do whatever needs doing to our gear. That's the fun part of the hobby to me. Talking is of little interest to me, but getting a rig working so I can talk is very satisfying.
@@johnwest7993 it's not something that can be "cleaned up" as it's a direct result of the methods being used to generate the SSB signals. It's about as good as it can get. It's amazing they can generate anything usable at all given the limited horsepower.
I don't currently have an easy way to inject other frequencies into the rig. I suppose I could use Audacity to generate them and may do so in the future. For now, I want to try to do everything I can to mitigate the IMD performance as shown - I will be working on that this week.
I don't have a unit, but just had a quick look at the schematic. THE ref A point which feeds a bias to the mic is only lightly filtered at audio with a 100nF. I wonder if a series R of a 100ohms or so and some uF will provide some extra filtering to reduce any tendency. You could scope the A reg line to see what happens to it when it goes ghastly..