I am using a RXTX6.3 SDR and ran into a problem when optimizing side band suppression with a spectrum analyzer. With only small deviations from 0 for amplitude and phase, multiple peaks appear over a bandwidth of about 30 KHz with the peaks 20 db below the carrier (CW mode) and 1.9 kHz apart. This happens under Linux and XP and does not allow me to optimize sideband suppression.
I usually use my direct sampling hardware, not SoftRock hardware. I did test with my RxTx Ensemble when I wrote this feature, and I did not see this problem. I did find it difficult to cancel the image because the settings varied with frequency, and the Tx audio level was critical. When I have a chance, I will test it again.
I did some more tests with quisk3.6.0 under Linux with the LO at 7.110 MHz and the TX-Frequency 10 kHz below it (CW-mode).
1) When ampl. adj. > 0 and phase adj. always 0: hf spektrum is normal and audio amplitude of Q-Signal decreases.
2) When ampl. adj. < 0 and phase adj. always 0: hf spektrum shows regular multiple peaks separated by 20 kHz,
I- and Q-Signals (audio) show additional higher frequency components.
3) When phase adj, is between -0.45 and .45 and ampl. adj is always 0: hf spektrum is normal and the I,Q signals look
normal as well. No phaseshift is visible on oscilloscope (are the phase adj. values in degree?)
4) When phase ad. is outside this range and ampl. adj is always 0: hf spektrum shows regular multiple peaks
separated by 20 kHz, I- and Q-Signals (audio) show additional highel frequency components as in 2)
This problem occurs under XP and Linux on a desktop and an atom270 lap top as well.
I am working on this now. Please try adding the following to your config file:
mic_out_volume = 0.6
to reduce the output level and provide some headroom. I think the amplitude and
phase adjustments may be causing clipping. I will continue to work on this,
but please try this for now and let me know what happens.
Thanks for testing this. The problem was that the default 1.0 level produces an output near max. But the amplitude and phase adjustments can increase the level and lead to clipping. I will add a feature to limit the level to 0.7. I did not see this since I was using 0.6 as the output level.