[Elecraft] AGC Pumping - NOT an issue in Orion or K3(?)

[Bill Tippett]

         Regarding comments by W3FPR and N6KR below.  I've been
thinking about this and wondering why I have never heard this
in my Orion in nearly 4 years.  I've operated in many contests,
including the CQ 160 CW every year with only an Inrad #762
4-pole 600 Hz roofing filter.  This environment is far worse
than virtually any other situation you are likely to encounter,
typically with S9+++ signals spaced every 500 Hz.  I have
*never* heard classical "AGC pumping", which sounds like the
noise floor is being modulating up and down in a normal rig.

         From an RF front-end standpoint, the K3 is a virtual clone
of Orion.  Both are down-conversion (8.2 MHz first IF versus 9.0
MHz) followed by narrow roofing filters.  Both DSP stages (2nd IF
in K3 and 3rd IF in Orion) are protected from over-driving by an
analog AGC which activates at approximately S9+30 (not sure of the
exact point for K3).  This prevents the DSP stage from going into
fibrillation (over-driving) which sounds like machine gun fire
beside your ears...nothing like classical analog AGC pumping.

         Assuming signals entering the DSP are not over-driving it
(i.e. less than S9+30 dB or -43 dBm using the classical definition
of S9 = -73 dBm = 50 uV), the DSP creates a filter around the
signal of interest.  According to Doug Smith (original architect
of Ten-Tec Orion), the DSP stage has approximately 100 dB of dynamic
range (reference below).

"As the resolution and speed of ADC hardware improve, digital radio 
receivers become less dependent on analog AGC to meet dynamic-range 
requirements. State-of-the-art 24-bit IF-DSP converters, such as 
those used in the Ten-Tec Orion, produce about 100 dB of dynamic 
range. That means a receiver can handle signals from the noise floor 
to almost 40 dB over S9 without analog AGC. Above that level, analog 
AGC is still necessary to maintain the linearity of analog circuits 
and to prevent overload in the ADC hardware."

                 http://www.doug-smith.net/dspdynamics.htm

         Assuming I have a very strong interfering signal (less
than S9+30 which is controlled by the analog AGC limit), the DSP
filter should be able to eliminate it without any AGC pumping.
I repeat from Doug Smith above:

"That means a receiver can handle signals from the noise floor to 
almost 40 dB over S9 without analog AGC."

Assuming S9+29dB = -42 dBm, subtracting 100 dB for the dynamic
range of the DSP, the result is -142 dBm, which is well below
the noise floor.  To me this explains why I have *never* heard
AGC pumping in Orion...hopefully K3 is the same!

                                                 73,  Bill  W4ZV

W3FPR:
AGC pumping is the problem I notice most with data modes - the 
roofing filter must keep out of passband signals at a sufficiently 
low level to allow copy of a weak desired signal.

Also my opinion on filters used for data modes is that they would 
ideally have good group delay characteristics, and to me that is more 
important than steep skirts or a flat passband. (steep skirted 
filters with very flat passbands usually exhibit poor group delay 
characteristics). Low group delay also means more pleasant sounding CW.

Steep skirts are not really needed with a roofing filter because the 
DSP can create the desired skirt selectivity. Again the roofing 
filter must only be narrow enough to keep strong offending signals 
out of the 1st IF passband (or sufficiently reduced to keep the AGC 
pumping problem at bay). Perhaps a Gaussian to 6dB filter would act 
as a better roofing filter than a steep skirted filter - but I have 
not done any measurements - just thinking out loud.


73,
Don W3FPR


wayne burdick wrote:
Let's suppose the only crystal roofing filter you have installed is 
2.8 kHz wide. You put the rig into CW (or data) mode, then dial the 
DSP bandwidth down to 400 Hz. Everything will be fine -- unless an 
extremely strong signal appears that's *outside* the DSP's passband 
but still *inside* the crystal filter passband.

This can cause what's called "pumping," meaning activation of the AGC 
or desensing. Many transceivers with I.F. DSP don't have the ability 
to do anything about this, so they get "trashed" on receive. Such 
transceivers typically have only a 3 kHz (or wider) roofing filter, 
allowing strong signals to cause trouble.

The K3 has roofing filter bandwidths as low as 200 Hz, giving you a 
means to completely eliminating problems with nearby signals. If 
installed, these will be automatically selected as you narrow the 
DSP's bandwidth using the shift/width/hicut/locut controls. Or you 
can select them manually.


We'll add this to the FAQ.



73,

Wayne

N6KR