Topband: PSK-31

[Joe Subich, K4IK]

> 
> >Since the average power with PSK idle is about 50% of the "key down" 
> >or single tone power, 
> 
> That's -3 dB.

You're correct but so what?

> 
> >the proper way to set up PSK (without a scope 
> >and spectrum analyzer) is to back the sound card output slider down 
> >to 50% (about 10 dB), 
> 
> Huh? how does 50% power turn into -10 dB? According to my math it's -3
> dB.  BTW, I measured the relative gain of the slider on my sound card.
> The first notch down from the top is 3 dB below the top. The second
> notch is 3 dB below that. As you get closer to the bottom, the changes
> are considerbly greater (but it does not appear to be voltage 
> linear). 

That's the audio (subcarrier) drive, not RF power.  Your own
measurements 
showed that your sound card required operating at -6 dB before the 
harmonic and spurious content became acceptable. 
 
> Just a minute here. Yes, we want to stay below the ALC. But since when
> is a decent ham rig generating lots of distortion 3 dB below full
> output?  If it is, we ought to be screaming at every SSB operator (and
> every CW operator) who operates his/her radio at rated power!  

SSB operators take advantage of the compression provided by the ALC. 
With PSK (or other multi-tone digital modes) actual intelligence may  
conveyed in the absolute/relative level of the specific tones.  Further 
one does not want the compression characteristics of the system to raise
the (relative) level of the spurious/harmonic content present in sound 
card output.  Finally, without the ability to look at a spectrum 
analyzer to see the actual transmitter behavior, 3 dB of additional 
headroom (above the known peak to average) is a wise minimum to provide 
for transient response.  Operating PSK31 at 3 dB down from transmitter 
peak output almost guarantees that the final amplifier is pushed to the 
limit and is operating at some level of gain compression (if not at the 
very threshold of clipping). 

> >Similarly, as Tom has pointed out, PSK users should be making 
> >use of the narrowest IF filters available (250 Hz) rather than 
> >a 2.4 KHz bandwidth for the AGC loop and then 
> 
> Maybe, but maybe not. Remember that the phase response of a 
> system with sharp filter cutoff can get pretty nasty, and it 
> is the phase modulation that we are detecting! So let's not go 
> too far with this.  2.5 kHz may be excessive receive bandwidth, 
> but 250 Hz may be too narrow. 

It depends on the phase characteristics of the filter.  Most 
250 Hz filters have relative benign filter skirts (1.8 - 2: 1
shape factors ... 2.5:1 for a mechanical filter).  Even a 250 
Hz filter provides twice the "necessary" bandwidth for a PSK31
signal ... without a brickwall transition from passband to 
stopband any problems with phase distortion should be much less 
than those from overload and/or AGC induced gain reduction.   

73, 

   ... Joe, K4IK