On Mon, 23 Feb 2009 13:57:32 -0500, Steve Ellington wrote:

>Is this correct?
>If a cw signal falls within the DSP passband it should and will pump the 
>AGC. If a CW signal falls outside the DSP passband it should not pump the 
>DSP's AGC. The problem here is defining what the DSP's passband it. If a 
>signal is just outside the audio range of the DSP (can't be heard) then I 
>would consider it "outside the DSP passband" and it should not pump the AGC 
>however this is never the case. I see moderate signals just slightly outside 
>the audio passband that pump the AGC. This is partly what is confusing us. 
>Signals that we can't hear pumping the AGC worry us. What we hear coming 
>from the speaker doesn't match how the AGC is responding. If my WIDTH is set 
>for 100hz and a signal is at 110hz, I won't hear him but my S-meter responds 
>to him as well as my AGC "desenses" (reduces gain).
>All of this has nothing to do with the roofing filter.

There are several HUGE gaps of understanding in this question, so I'm going 
to attempt to respond to this tutorially. 

First, the shape of the skirts. You seem to be viewing the response of a 
filter as if it were square -- that is, no skirts. When you build a filter 
for a bandwidth of 100 Hz, 100 Hz is the width between the -3dB points on the 
filter's response. Depending on the design of the filter (how many poles and 
how they are aligned), the response may be -20 dB at a 200 Hz width, -40 dB 
at 400 Hz width, and so on. The filter's cutoff may be sharper or narrower, 
depending on the design. To get a feel for how filters work, go to the Inrad 
website and poke around until you find graphs of the response of some of 
their filters. Inrad makes high quality after-market IF and roofing filters 
for a lot of ham receivers, including the roofing filters used in the K3. 

Second, the -3dB points of a 100 Hz filter are 50 Hz above and below the 
center of the filter. Third, let's say that the signal you're trying to copy 
is 50 uV and the interfering signal is 5 mV. That's 40 dB difference. To make 
those signals equal in level in your earphones, the filter must be 40 dB down 
at the frequency of the interfering signal. Many of the stronger signals on 
the band may be 60-100 dB stronger than the weak signals we're trying to 
copy. If one of them is within the passband of the roofing filter or the IF, 
the skirts of both filters come into play. 

A fourth issue is that the all the filters in a system act in cascade -- that 
is, the total response is the sum of the responses of each filter. Let's say 
you have a 400 Hz roofing filter installed and you set the DSP IF to 400 Hz. 
The roofer may be -3dB at 200 Hz off frequency, the DSP is -3dB at 200 Hz off 
frequency, so the cascaded response is -6dB at 200 Hz off frequency. At 400 
Hz, if each filter had the same shape (unlikely) and was -20dB, the two in 
cascade would be -40 dB. In general, 8-pole filters have steeper skirts (that 
is, their response cuts off more sharply as you move off center) than 5-pole 
filters, so a 5-pole 200 Hz filter might have the same response 200 Hz off 
frequency as an 8-pole 400 Hz filter. 

All of these filter shape and cascading concepts apply whether the filter 
uses crystals or DSP. From "outside the hardware box," the primary 
differences between them are their degrees of adjustability, how much signal 
it takes to make them non-linear, and what they do when they overload. One 
major reason for using a roofing filter is to protect the DSP filter from 
being overloaded by strong signals. The other major reason is the cascading 
it provides. 

Now, let's look at the AGC. I haven't taken the time to study design details 
of the K3's AGC, but from what Wayne and Lyle have posted here, I think I 
understand that there are two AGC functions, one in hardware and one in 
software, that sense signal levels at different points in the signal flow and 
control gain at different points in the signal flow.  The AGC in hardware is 
not adjustable directly, but we CAN (and MUST) prevent it from pumping by not 
overdriving it. The AGC in software (that is, the last IF) is VERY adjustable 
from the CONFIG menu. 

Another way of looking at the problem is to understand that the AGC turns 
down the system gain if the signal at the output of the IF is too high. If we 
let the AGC do it, the AGC will pump on a strong signal that gets through the 
filter skirts. But we can prevent most, if not all, of that pumping by 
turning down the RF gain with the controls on the front panel of the radio -- 
that is, turning off the preamp and turning on the attenuator and turning 
down the RF gain controls. That's how W4ZV operates his radio, and it's how I 
operate mine. We're both quite happy with how our radios are working.

73,

Jim Brown K9YC


______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:Elecraft@mailman.qth.net

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

Reply via email to