A guy in our area who seems to be getting good results reports the following in an email:
/Quote/
Audio 'goodness' is related to having minimal error bars, not high percent input level indications.
Here is a comm segment that was perfectly audible. There is some initial settling, but once the quiescent level of 3% is reached, the error bars disappear and the audio is crystal clear.
Note: this is from a strong nearby system.
35%..e:=RRR=RR
25%..e:===
8%....e:
4%....e:=
3%....e:
3%....e:
3%....e:
3%....e:
3%....e:
3%....e:
3%....e:
3%....e:
3%....e:=========R========R==R=============
In my system, if the input signal gain is increased to cause readings much beyond 20%, major error bars start appearing that are filled with Rs, and the audio becomes unintelligible or absent altogether.
Again, the above sequence is from a strong, nearby system. For weak, distant systems the best decode and audio out still are obtained at 3% levels, but the error bars are considerablly longer and audio can be really ragged most of the time.
Decoding success also seems to depend on quality of the discriminator output. Also, inserting a series 100 k resister and 10 uF cap ahead of the audio adapter seems to stop signal sag and greatly improve decoding, Presumably, a buffer amp would be even better, but perhaps not worth the additional complexity.
/EndQuote/
So, from these comments, it looks like the lower the % level, the better, and that unloading the receiver's discriminator output is a good thing.
-rb-
