SSB generation by shifting audio spectrum

Hi, I am thinking of the filter method of SSB generation. In the simplest form, the 0.3-3khz audio is mixed with the bfo to produce both sidebands in the IF, then an IF filter (crystal or mechanical) is used to remove one of these sidebands. This requires difficult to build crystal filters.

I am thinking of a way to relax the need for so strict filters. If the 0.3-3khz audio was somehow "shifted" in the audio domain (say 17.3-20khz) and then using this shifted audio signal for the mixing to RF, then the two sidebands produced by the bfo (information passband signals within the sidebands) would be several khz apart. Two signals of about 35KHz may be easier to filter even with a good LC filter.

How does this idea sound to you? This isn't the way the weaver method works I think.

Possible, yes.

How do you plan to shift the audio? If you intend an analog single side band I/Q mixer, it's not less effort than implementing it for the final RF frequency. Or do you start digital signal processing now?

Shifting the audio is the problem. You substitute one obstacle for another.

With DSP it is possible but if you want to do it with analog circuits, you are essentially adding a second mixer stage into the design, one to up-convert the voice spectrum prior to the RF mixer. Think of how you swap sidebands too, you need to spectrally invert the audio frequencies which means mixing with a frequency just above or just below the audio range. For example, to retain the 'normal' audio direction, the audio up-mixer LO has to be 17KHz (using your figures) which would place it within the LPF range. To reverse the spectrum it would have to be at 20.3KHz which would imply a very steep roll off in the LPF between 20KHz and 20.3KHz which would be difficult to achieve.

Brian.

betwixt said:

Shifting the audio is the problem. You substitute one obstacle for another.

With DSP it is possible but if you want to do it with analog circuits, you are essentially adding a second mixer stage into the design, one to up-convert the voice spectrum prior to the RF mixer. Think of how you swap sidebands too, you need to spectrally invert the audio frequencies which means mixing with a frequency just above or just below the audio range. For example, to retain the 'normal' audio direction, the audio up-mixer LO has to be 17KHz (using your figures) which would place it within the LPF range. To reverse the spectrum it would have to be at 20.3KHz which would imply a very steep roll off in the LPF between 20KHz and 20.3KHz which would be difficult to achieve.

Brian.

Click to expand...

Yes DSP in a PC using software would be the easiest. One program that comes in mind is spectrum lab. It has a really steep and highly configurable filter plugin. Basically you drag the mouse and you create the audio filter you want. I am not sure it can go close to 20khz for that I guess you would need a really good sound card which supports high sampling rate. Another problem is latency.

I did not understand the thing about inversion.
Let's suppose 0.3-3khz baseband audio bandwidth mixed with a 17khz carrier. This will produce USB and LSB close to 17khz. You filter out the LSB, then you have the USB which is not spectral inverted. Then you apply this USB to an up convert RF mixer and you have two sidebands (say close to 455khz for the example) each one 17.3KHz apart from the BFO. If you tune the BFO lower in frequency then you can pass the USB from the (now relaxed) IF filter.
No spectrum inversion nowhere, this is true for USB.

yes, as a ham, i am sure you know that you can adjust the BFO in the receiver and make the audio sound like it is back at the right frequency.

I'm also a ham but as I understand Neazoi's proposal, the intention is to somehow shift the voice frequencies (without expansion) up to a higher frequency so that one sideband can be eliminated with an LPF. Essentially moving the sidebands apart sufficiently that a simple filter would be adequate to eliminate one of them. I think it would work for one sideband but not for the other. To reverse sideband, the audio spectrum would have to be mirrored, otherwise you would get one sideband or both sidebands simultaneously (DSB) as the cut-off frequency would have to extend beyond the upper sideband range and therefore let the lower sideband through as well.

Brian.

Frequency shift in audio band is almost always implemented using I/Q SSB modulator rather than filters. Using this method, the effort for inversion (lower side band generation) is essentially the same as for upper side band.

The most demanding block is the precise wide band differential 90 degree phase shift, which can be best implemented by all-pass chains. Alternatively pure digital I/Q SSB.