[Elecraft] Regarding the K3 and high QRN levels

[Guy Olinger K2AV]

The defensive hardware AGC is hardware AGC because you can see the entire
circuit for this defensive AGC in analog components.  It cannot be turned
off.  It is an AGC that in all circumstances would be better if you didn't
need to use it, though this is clearly not possible in some circumstances.
This is only one of various reasons for using the narrowest roofing filter
that fits the mode and usage.

I could call hardware AGC simple-*ssed AGC, because it is, and it has only
one purpose: don't overdrive the d*mned analog to digital converter (ADC).
Hardware AGC seems more polite. ADC-defensive is more specific.

DSP AGC is entirely done in algorithm in the number soup bit bucket.

The degradation of NR performance occurs when the ambient noise level is
allowed to hit the ADC at a high level, so that both the noise and desired
signals are at the same level going into the ADC, with the defensive
hardware AGC acting as the leveling device.  The follow-on digital
algorithms then have to accept as fact that the noise and the desired signal
are now the same level.  Perhaps with an ADC with another 30 or 40 db range,
and similar increases in non-military cost processors are available, a
defensive hardware AGC will not be needed.

The manual solution to the quandary of the prior paragraph is to turn off
PRE, perhaps turn on ATT, and perhaps reduce the RF gain pot as well until
the ambient noise is only moderate at the K3 output.  As clear as this might
seem to some, attempts to correctly relate a "roaring noisy" K3 to
PRE/ATT/RFgain settings for a band have often not been understood.

PRE/ATT/RFgain are not dynamic adjustments, they don't vary in millisecond
time ranges as found in hardware AGC.  They vary at most at hand-twiddle
speeds as they are manual.  When the PRE/ATT/RFgain have been turned down so
that ambient noise is in the LOW range at the ADC, then following digital
algorithms have true information to work with, that has not been compressed
by the hardware AGC.  The DSP's "AGC" can do the stuff that can't be done
with simple circuits.

>From emails it is clear that proper manual reduction of PRE/ATT/RFgain to
suit band and conditions is not catching on in a significant customer
slice.  My suggestion is to take these users at face value, that they want
the setting of PRE/ATT/RFgain done automatically for best performance.

If someone insists on calling that AGC, then there would be *THREE* AGC's.
Hardware ADC-defensive AGC, digital level management in DSP algorithms, and
a third, crawl-speed RF string management AGC.

I'm thinking that the DSP can tell the difference between noise and discrete
signal. It therefore can over a period of time (a second or two, or five or
ten?) determine the best setting of PRE/ATT/RFgain and slowly engage the
preamp, attenuator and set the resting gain of the post filter IF ampl based
on that determination, rather than manual input.

This latter automatic mode for RF string management could be engaged by
reducing RFgain to minimum. To get out of it the user could advance the
RFgain. At the point matching the automatic setting, the manual control
would re-engage.  The PRE/ATT settings would be left as is, because those
would probably be correct.  But they would follow the buttons if depressed.

73, Guy.



On Fri, Jul 9, 2010 at 11:16 PM, Wes Stewart <n7ws at yahoo.com> wrote:

> I'm sorry, but you can't have it both ways.
>
> "RF" (i-f) gain as implemented in the K3 operates on the post-filter (8
> MHz) i-f amplifier.  The control voltage is derived at least two (perhaps
> three) ways.
>