[Elecraft] K3 DSP AGC Question

[Don Wilhelm]

Dave,

I believe you are asking for a more "rounded knee" on the AGC attack 
curve.  I don't know if that is feasible without a lot of work on Lyle's 
part - he would have to answer.  Right now, (as you noticed) the slope 
of the AGC line between the threshold point and the point where the 
hardware AGC kicks in is almost linear.  I am not certain how much more 
computation would be required to produce a more rounded knee, nor how 
fast the DSP engine can compute a more complex function - if there is 
not adequate engine speed, things could become quite distorted.

Changing that function should not change the threshold levels - the 
point where the knee occurs just depends on the threshold, and the shape 
of the knee should not influence it.  In use, a softer knee may make the 
use of a lower threshold more practical.

73,
Don W3FPR


Dave - AB7E wrote:
> I also think that the current max AGC threshold is too low, but I have another question.  Why is it necessary for the K3 AGC to have such a sharp elbow at the point where the AGC kicks in?  The Clifton Labs plots show such a sharp bend that it seems multi-signal in-band intermod is inevitable, especially with higher SLP settings.  Wouldn't it be feasible, and more desirable, to have some more gradual inflection point?  I would think that something like a parabolic shape would be better, and the DSP AGC is just the implementation of an algorithm, correct?  I suppose that would tend to blur the distinction between different THR settings, but so what?  
>   
>