[Elecraft] Sherwood Engineering and DSP

[Brian Lloyd]

On Jun 6, 2007, at 9:16 PM, Jan Erik Holm wrote:

> Lets hope Elecraft got it right!
>
> /SM2EKM
> -------------------------------------
>> Sherwood Engineering just put up their Dayton 2007
>> presentation on DSP and how it isn't everything it is
>> cracked up to be.  See:
>> http://www.sherweng.com/documents/Dayton2007w.pdf

The thing is, what he is talking about would be very easy to fix. You  
just need a smoothing algorithm to keep your AGC from responding to  
transients by slewing too rapidly. Smoothing algorithms are easy to  
do. Here is one:

AGC' = 0.9 * AGC + 0.1 * signal

What the above says is:

take 1/10 of the signal level (signal) and add it to 9/10 of the  
previous AGC value to produce the new AGC value. We change the  
response time by varying the ratio. Want faster AGC? Do this:

AGC' = 0.5 * AGC + 0.5 * signal

Want really slow AGC? Do this:

AGC' = 0.99 * AGC + 0.01 * signal

By changing the coefficients depending on whether signal is greater  
or less than AGC you can change attack and decay times. Here:

if (AGC < signal)
     {
     # Attack
         AGC = 0.9 * AGC + 0.1 * signal;
     }
     else
     {
     # decay
         AGC = 0.99 * AGC + 0.01 * signal;
     }

This is a really simple problem and I have a hard time believing that  
this is the real problem. You want the AGC to prevent the A:D from  
clipping but just barely. That will give you the maximum dynamic  
range. If the A:D is allowed to clip then all bets are off.

I just have a hard time believing that any competent engineer would  
fail to understand this.


73 de Brian, WB6RQN
Brian Lloyd - brian HYPHEN wb6rqn AT lloyd DOT com