Hello

When working with ssb modulator and especially with the compressor, the signal spectrum expands. This is not only visible in the panorama, but also transmitted. If I reduce the microphone gain or compression, the expansion of the spectrum becomes less noticeable, but such a signal is already quite quiet and not “dense” (almost without compression). How can the ssb modulator and levels be properly tuned so that the spectrum does not expand and at the same time the signal remains loud and "dense" (with good compression)?

It doesn't transmitt via hackrf as i know :(

Hi,

on the second screenshot the AGC volume target is 100% of full range (1.0). This means the compressor will try to maintain the level around this value and for sure it will exceed 100% (as shown by the VU meter) and it will overload subsequent stages. This should be reduced until the red bar of the VU meter never goes above the 100% mark. 

This does not rule out possible issues in the half-band upsampling chain up to the device sample rate. There seems also to be a too big difference between a CW modulation and audio input (~15dB) in the channel power for 100% peak modulation.

Brgds, Edouard.

After a second look the issue is with the compressor. It does not work because I don't know how to code it.
Brgds, Edouard.

Edit: a hard limiter keeping the resulting sample magnitude below 1.0 appears to do the job of suppressing overload allowing to push compression at the expense of increased distortion.
Edit 2: Issue https://github.com/f4exb/sdrangel/issues/336 opened

Thanks, Edouard 

AGC volume target is 100% of full range (1.0) because smaller levels have no effect like in real transceivers (1.0 looks a bit like). But the expansion of the spectrum is present at lower levels, just less often. It is difficult to do without a compressor when the signal level is higher than noise by 10-15 dB, almost all use it in ssb.

Hi Alexander,

v4.7.1 was just released with this fix. The full range is the actual full range mathematically (magnitude 1.0) and thus you cannot go above it. This value should yield the maximum available power at the HackRF output (considering the amplifier level). This seems fine on a single tone but should be verified with a 2 tone test.

Brgds,
Edouard.

Alexander.

I have not been able to download a Windows version of SDRANGEL, I have
no hands on experience with it. But I thought it is an interesting
option based on what I read. So far, I only have RTLSDR's by now, and
a YU1LM receiver that uses the sound card.

Don't know what kind of processing you are doing, but altering
waveforms distorts respect to the original and creates harmonics,
which can be thought as non essential emissions, i.e, can be filtered
out without consequences to intelligibility.

Two things can be done:

- filtering the output of the processor to limit what is transmitted
beyond 3 kHz BW, that will limit the transmit mask bandwidth.

- use preemphasis from low frequencies up to about a kHz, this is
pretty subjective, but in an external hardware RF clipper I found that
reducing the level of low frequencies reduces the processing level and
associate distortion.

Hope this helps,

Jose, CO2JA