[Elecraft] K2 RF Speech Processor?

[Geoffrey Mackenzie-Kennedy]

Karl Larsen <k5di at zianet.com> wrote:

>    Hi Geoff, cut your weeds and we can both think hard on what "Speech 
> Compression" really is. I got to thinking and about all it can be is 
> clipping of the high voice peaks, and increasing audio drive. I will look 
> at the ARRL Hand Book and see what they have.

----------------------------------------------------------------

Hi Karl,

My apology for my delay in getting back to you earlier, as a Scottish poet 
wrote -'The best laid plans of mice  .........

The end result of compression or clipping is essentially the same, which is 
that the peak to average power ratio of the output signal is less than the 
peak to average power ratio of the input signal, assuming that 'compression' 
*is* taking place likewise clipping (slicing). This obviously means that the 
average power of the signal is in both cases increased.

Compressors, frequently appearing under different names, achieve the result 
more 'gently' than clippers, and there is some delay before they react in 
part due to the time constants within the gain control loop. Whether or not 
this delay results in any performance problem is something that the designer 
will have checked, and in the K2 there does not appear to be a problem.

You will have seen that in the SSM2165 chip the sample of the signal used to 
control gain is taken from a point ahead of the gain controlled amplifier, 
the VCA. This method of 'forward control' makes it much easier to have the 
controlled amplifier's gain set suitably for a signal as it arrives. The 
picture is of course more complex in real life because many signals of 
various amplitudes and frequency occupy the same time frame.

Those systems using 'backward' control, such as a typical ALC where the 
control sample is taken from the output, by their very nature will allow 
some output, perhaps distorted, before the gain of the controlled amplifier 
is reduced - closing the barn's door after the horse has bolted.  Thus in my 
view, ALC is useful if used carefully and as the last line of defence.

Clippers, assuming proper design, react very quickly slicing the signal at 
the required amplitude. If used directly at audio the result can be a mess 
of harmonics and intermod products. For this reason an audio signal is 
upconverted to become a SSB RF signal, symmetrically clipped, filtered and 
downconverted back to audio. With a reasonable RF SSB filter before the 
clipper to obtain unwanted sideband suppression and another after the 
clipper to attenuate harmonics and to limit bandwidth , some intermodulation 
products within this second filter's passband will remain. In practice their 
effect is acceptable provided that a low pass audio filter with a cutoff 
frequency around 3 kHz is added to the output. One function of this audio 
filter is to attenuate any intermod products resulting from the RF to audio 
downconversion which are outside of the required audio passband.

Of course the clipping and filtering can be applied to a transmitter's IF 
SSB signal instead.

I admit to oversimplifying the picture.

73,
Geoff
GM4ESD