TenTec
[Top] [All Lists]

Re: [TenTec] One complicated radio! [Long}

To: tentec@contesting.com
Subject: Re: [TenTec] One complicated radio! [Long}
From: "Grant Youngman" <nq5t@comcast.net>
Reply-to: tentec@contesting.com
Date: Mon, 24 Nov 2003 16:19:20 -0600
List-post: <mailto:tentec@contesting.com>
> SO BOTTEM LINE, MY QUESTION IS:
> Is ORION OK? or are we doing all this tweaking beacuse there are 
some
> problems making it work the way we all want AND EXPECT it to?
>

One opinion (of many).  In short, yes the Orion is OK.

There are a number of design elements in this radio that are quite 
different than most other radios out there.  Digital AGC is one, and I'm 
sure it's something we'll see in most new radios.  For example, the block 
diagram of the coming Icom 7800 indicates AGC loops within the DSP.  
The PRO-II on the other hand, appears to use a digital level to control 
only analog AGC, but that isn't the same as having an AGC loop fully 
implemented in DSP.  It doesn't behave in quite the way we're used to 
analog AGC behaving, and I think that's the root of some of the 
complaints about "noise", etc.

But that in itself doesn't imply there's something wrong with the radio.  It's 
a new paradigm.

These comments are based on Doug Smith's articles, and my own "blind 
man feeling the elephant" observations, because I haven't dug through 
the schematics, and certainly haven't seen any DSP code .... so I can't 
guarantee that what I think is entirely correct ...

The control flexibility -- of threshold, decay, and hang rates -- are the 
result of containing the primary AGC loop in the DSP.  It seems to me 
this could also be done in an analog AGC loop, but perhaps not quite so 
easily and with as great a range.  At least there isn't a (ham) radio in my 
memory that has had all of these control functions sitting on the front 
panel.  You're lucky in most cases to have a variable AGC "speed" 
(decay) control rather than just a S/M/F/OFF switch.

The RF Gain control is actually an IF gain control, and controls one or 
more gain/attenuation blocks in the DSP.   It isn't controlling the gain of 
an analog RF/IF stage in the radio.  The ATTN does that by brute force, 
and of course there is an analog AGC loop, but that only kicks in on 
signals over 9+30 or so.  I think (If I recall Doug's writeup correctly) that 
the analog AGC is really there primarily to prevent overdriving the ADC, 
less so to manage overall receiver gain.

The overall objective is to maintain high signal levels at the input to the 
ADC -- but with enough margin that you don't overdrive the stage and 
clip.  That maximizes the dynamic range of what you can get through the 
ADC and into the DSP IF for processing.  So the front end runs wide 
open unless the levels need to be cut to protect the ADC, or you punch in 
some attenuation.

And the digital AGC -- unlike typical analog AGC -- seems to have a gain 
component on the curve as well.  At least that is what I feel on the 
elephant, whether it's quite correct or not.  But what is the reduced 
attenuation other than "gain" relative no a nominal unity point.  

The way the behavior of "threshold" is described by Doug Smith, signals 
below the threshold are basically left "as is" by the AGC -- at least to 
some point. So if you have the threshold set to 3 uV, then signals below 
that level are not affected by AGC, and will be less "loud" than signals 
above the threshold -- that is, they retain their expected dynamics -- weak 
signals are less loud than slightly stronger ones, and then as the 
threshold is reached, everything gets compressed to a more or less 
constant level.   As the threshold is lowered, the noise level appears to 
incease and weak signals tend to come up in level, implying to me that 
what we observe as increased noise is actually an increase in "gain" of a 
DSP gain block.  It's true that strong signals are compressed, but weak 
ones (if there's no stronger signal to capture the AGC) seem to be 
increased in level -- all with the objective of equalizing apparent gain.  As 
the AGC threshold is increased, the background noise level drops -- 
which is consistent with signals below the threshold not being affected by 
AGC, and vice versa.

Here's a guess -- that what sounds like front end "noise" and what some 
have assumed means a poor noise floor  is the result of what I've just 
described.  Case in point -- the high level of white noise that appears in 
the radio's output when no antenna is connected, and everything is 
basically running wide open with threshold set to the minimum value.  
That noise level is really unimportant, and says nothing about receiver 
"noise" -- it's an artifact of what the DSP with DSP-AGC does with no 
input signals -- just trying to run everything wide open -- and what you're 
hearing is extremely low level random noise at the ADC input being 
amplified through the DSP.

This really is something that would be interesting to have a better 
understanding of.  The details of some or much of it may be proprietary.  
But it would help in a better overall understanding of what's going on in 
the box.  Better than just stroking the side of the elephant, anyway :-)



_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec

<Prev in Thread] Current Thread [Next in Thread>