Again, an interesting observation, Gary.
Location of the filter(s) *should* have no effect on the overall
performance of the IF chain, in fact of the entire receiver, *provided*
that the entire system is operating linearly. In a cascade of linear
systems, the overall transfer function is the product of the individual
functions, etc. The trick, of course, is to maintain a cascade of truly
linear sub-systems.
Placing the filters up front, as you point out, can be an advantage if
they prevent large signals, etc. which *could* drive one or more of the IF
stages into non-linear operation, from progressing beyond a early stage
that can handle their amplitudes linearly. This evidently is the situation
to which you refer in stating a preference for having the filtering come as
early as possible, as in the Corsair IF design. This suggests that the
filters are located "up front" in a deliberate effort to allow the use of
subsequent stages with a lower overall dynamic range than would otherwise
be required.
This is the same situation that prevails with the entire RF/IF system in a
receiver. Provided that the system from antenna connector to some arbitrary
point of delineation in the receiver, such as the demodulator stage input
or the input to the ADC in an IF/DSP architecture, is operated linearly at
all times, regardless of input signal levels, it matters not where the
bandlimiting function is located. Hence, the modern approach with IF/DSP
receivers to concentrate the selectivity in the DSP section subsequent to
the ADC stage. This approach has powerfully attractive performance and
economic ramifications.
Almost every modern commercial and military radio in production today - in
fact, I can think of no exceptions - follows this approach and employs
nothing but DSP filtering for setting passband limits. The extraordinary
and near-mathematicially perfect DSP filter shapes permit a level of
performance essentially unobtainable with analog means. The Ten-Tec RX-340
is a typical example of this architecture at the low end of the
commercial/military receiver market.
On the other hand, if a less than outstanding design and implementation of
the receiver "front end" - i.e., from the antenna connector to, say, the
ADC input for the IF/DSP section - is the case, then despite having near
perfect filters in the DSP section, distortion products generated by
overdriving various stages ahead of that point will be present within the
ADC input and can conceivably not only contribute to signal distortion
post-DSP but also could readily exceed the input signal limit of the ADC
itself, thereby creating the worst possible distortion environment in the
receiver, from which there is no recovery.
It is in an effort to avoid these effects that we find that some receiver
designs (which for whatever reasons - cost, complexity, design effort, etc.
- are unable to achieve the desired linearity ahead of the ADC input)
employ varying levels of selectivity "along the way" as it were with the
use of passive usually crystal-based ladder filters among the IF amplifier
stages in an effort to minimize the forward progress of distortion products
that could drive subsequent stages into non-linearity.
This is probably a reasonable engineering tradeoff between the cost of
analog and software engineering, components, production and so on, required
for a truly "bullet-proof front end" that remains linear under all
operating and signal conditions and the cost of one which is "good enough"
when supported by additional relatively narrow passive filters.
We find examples of both design approaches in the amateur marketplace
today. It will be interesting to observe the direction and rate of progress
of these differing design philosophies as time passes.
Thanks for the information, Gary, and a very enjoyable exchange of views.
73/72, George
Amateur Radio W5YR - the Yellow Rose of Texas
In the 57th year and it just keeps getting better!
Fairview, TX 30 mi NE of Dallas in Collin county EM13qe
K2 #489 Icom IC-765 #2349 Icom IC-756 PRO #2121
Gary Hoffman wrote:
>
> George,
>
> I think it is important to note that a big difference is made by where
> in the receiver chain the filters are located.
>
> In the Pegasus, as you know, the DSP work is done at the end of the chain.
>
> On the Corsair, and many other radios, the crystal filters are in the early
> stages of the chain.
>
> Since they eliminate much noise up front, before it gets fully amplified by
> the following stages, they can be much more effective than the same quality
> filter applied downstream. In fact, since they can help prevent
> fundamental
> overload, they can be dramatically helpfull.
>
> I don't mean to suggest that the IF DSP filter in the Pegasus is a poor
> filter. I only note that it is at the end of the chain, and less effective
> because
> of that fact.
>
> 73 de Gary, AA2IZ
>
> ----- Original Message -----
> From: "George, W5YR" <w5yr@att.net>
> To: "Gary Hoffman" <ghoffman@spacetech.com>
> Cc: "Mike Hyder -N4NT-" <N4NT@charter.net>; "Scanandoah, Alan"
> <ascanand@harris.com>; "Ten-Tec Reflector" <tentec@contesting.com>
> Sent: Monday, October 28, 2002 11:19 PM
> Subject: Re: [TenTec] Receivability
>
> > Very interesting, Gary, and thanks!
> >
> > I have made detailed spectral comparisons between the IF DSP filters in
> the
> > ICom 756 PRO, the Kachina 505DSP of similar architecture and the Icom 765
> > with conventional crystal filters. There is no comparison whatever between
> > the shape factors and overall shapes of the PRO filters compared to any of
> > the others. The PRO filters are mathemetically nigh perfect in shape and
> > performance. I think that you have found that the Pegasus IF DSP filters
> > are not the equal to the conventional INRAD filters - that is very
> > surprising to me.
>
|