Team Vertical's 2010 C6 expedition for CQWW (Kenny, C6ATA, Fred, C6AKX and Walt C6ATR) were disappointed that the K3's ability to distinguish calls in the pile ups was sorely lacking. Kenny greatly prefers his TS850 and I found that I had to turn AGC off (as many other contesters have reported). It seems to me that there are two parameters that control the performance of the AGC as a function of signal strength and thus the dynamic range of the audio signals we hear in the pileup. These are AGC SLP (000-015) and AGC THR (2-8). The K3's dBV signal measurement gives us a convenient way to measure the dynamic range (and audio signal levels) of signals with AGC on and off and to see what affect SLP and THR have on the audio dynamic range.
I used the XG2 signal generator and a simple voltage divider to inject a signal ranging from 1 to 50 microvolts. With the K3's preamp turned off, this gives S1 to S6-7 signals. The DSP bandwidth was 600 Hz with a 1 kHz roofing filter.
AGC vs SLP
SLP adjusts the slope of the AGC transfer curve. As the K3 manual states: "Higher values result in flatter AGC (making signals at all amplitudes closer in AF output level).
Figure 1 shows shows two AGC curves -- SLP = 000 and SLP = 15 (for THR = 8, the highest value).
It is interesting to see that AGC starts to come into effect at about S3 (with THR=8) and for a dynamic range of input signals for S6 to S3 (roughly 20 dB) the resultant dynamic range of the audio signals with AGC F operating is roughly 6 dB. It is no wonder that stations that were running higher SLP values had trouble distinguishing signals in the pileups.
AGC vs THR
AGC THR "Sets AGC onset point; a higher number moves the onset point up." Two AGC curves are shown in Figure 2 -- THR = 8 and THR = 2 (for SLP = 0).
Figure 2. AGC vs THR
As the above shows, THR=8 starts the AGC action at about S3 and THR=2 at about S1 (or below).
AGC for Stronger Signals
I decided to extrapolate Figure 1 to see the affect on the dynamic range for more realistic signals in a pileup -- up to about S9.
Now, for signals ranging from S9 to S3 (an input dynamic range of about 35 dB), the output audio dynamic range for the "best" AGC settings (SLP=0, THR=8) is about 13 dB.
Analysis and Conclusions
The measurements above assume that the AGC OFF transfer function is linear. It may not be but the relative audio dBV signal measurement are probably OK. Further, these measurement reflect the situation of only one signal in the passband, not multiple signals as in a pileup. Again, though, the relative measurements should generally reflect the performance of the K3's AGC.
I have to conclude that the K3 AGC onset with THR=8 seems to me to be too low. I would think that S6 or even S9 signals could be passed by the system without AGC action. That's pretty much what occurs when we operate with AGC off and ride the RF and AF gains.