From: Paul Keinanen on Sun,May 8 2005 11:54 pm

On Sun, 08 May 2005 10:44:27 -0700, Tim Wescott
wrote:

The basic difference is that with a digital system you either end up
with a clean signal or a useless signal. In an analog system the
character and purity of the signal must be carefully guarded, at
least
until you manage to digitize it. This means that there will be a
much
greater chance that adding a new card to the radio will degrade not
only
the function of the new card, but the function of all the other
cards.

Second, the PC market is a huge one, with great advantages to be
derived
from common equipment and software, and much smaller advantages to be

derived from commonality. This is the exact obverse of the radio
market, including homebrew radios. To make a "card" radio would be
to
define a basic radio architecture, probably down to the IF frequency
(or
at least to the point of forcing you to match your IF and front end).

While improvements could be made within this structure an independent

experimenter couldn't play around with such things as
direct-conversion,
different IF schemes, etc., without extensive modification.

I agree that it would be quite hard to make a good quality radio with
some common backplane structure. However, connecting various
functional modules with 50 ohm input and output impedance could be
used to make quite different radios with good specifications.

That's already been done in the RF industry for a half
century.

As one example, take the U.S.' AN/PRC-8, -9, -10 series
of manpack transceivers covering high-HF into low-VHF.
Still in the vacuum tube era, all of the IF modules
included the IF tuned circuits as well as the subminiature
tube. If the tube filament burned out, the entire module
was replaced. NO alignment tweaking was required. Design
was done back around 1950.

As for standards on control...start with the ATLAS (for
USAF test equipment) and continue on to the IEEE-488
interface. Those standards worked with "modular"
components capable of testing receiver sensitivity down
to noise level with KNOWN signal levels. By the way,
test equipment for RF has been standardized at 50 Ohms
since WW2 days.

For
instance Mini-Circuits also makes various diode ring mixers,
amplifiers and apparently also VCOs that are boxed and have BNC or SMA
connectors. With each functional module in a metallic enclosure,
controlling the spurious radiation between modules is much easier. I
don't know that anyone would make filter modules, which would be
required to build a complete radio. Also SSB-Electronics sold separate
amplifier, mixer, frequency multiplier and crystal oscillator modules
mainly intended for a 10 GHz transverter.

Unfortunately the cost of these modules is quite high, apparently due
to low production volumes and large amount of manual labour needed to
assemble them. If there would be a large demand for such modules, it
would make sense to design them to require less manual labour to
assemble them and hence get the price to more affordable levels.

Define "more affordable." :-)

"Filter modules" have and are built to order by dozens
(if not hundreds worldwide) of companies. The costs
ARE high because they are built TO specifications and
such have to be TESTED to meet those specifications.
Is there comparable KNOWN/calibrated test equipment
in the average homebrewer's hobby workshop that is
comparable...even at "low" frequencies of HF? Actually,
Kaylie's Mini-circuits DOES use calibrated, automatic
test equipment to check out each module, small
quantities to large quantities. Mini-Circuits doesn't
have the market demand to do production runs in the
10,000-lot quantities.

The mystique on L-C filters is largely that...mystique.
Without some good, calibrated test equipment, it is
very difficult to determine what a "filter module"
has for performance. Synthesis (design) of the values
for a particular filter type was arduous until folks
came out with computer-aided design. I have a working
freeware program for PCs on that...send a message in
private e-mail if you want one transmitted to you.

As to cost, just look at a cellular telephone handset.
Those can cost around US$ 50 each, new. They work in
a band roughly centered at 1.0 GHz. Microwaves.
Complete microwave Rx-Tx with synthesized tuning.
For half a hundred US dollars here. A mere 30 years
ago that would be almost inconceivable. Three years
ago the U.S. Census Bureau said that one in three
Americans have a cell phone subscription. That's
roughly 100 MILLION units either out there or waiting
to be used. Market quantity and competition in that
market are the key to bringing down costs. Radio
hobbyists just cannot possibly get close to such
market quantities.

While a backplane would not be suitable for running the RF signals, it
would be a good idea to have a common control interface standard. This
might be some sort of serial interface or perhaps a CANbus interface
as used on some AMSAT satellites.

Who says a "backplane would not be suitable?" :-)
Those PC backplanes carry terribly broad spectra of
RF...from (literally) DC on up to the low microwaves.
No "perhaps" about it. Thing is, the layout can NOT
be done as if it were wire-wrap; i.e., in random
order of wire placement. With broadbanding anything,
every single adjacent trace becomes a COUPLER and
unwitting layouts can produce remarkable crosstalk
effects. Designers have known that for decades and
handle it...all kinds of Application Notes and info
out in public access available for anyone...just too
specialized for the "weekender" small-project
assembler hobbyist.

The IEEE-488 is a mature standard for control and
interface for computer-controlled, interconnected
systems. Would be a bit TOO all-inclusive for a
special-purpose new design. The "interface" does
NOT have to be some kind of "new" thing used on the
latest whatever out in space. It's just a control-
and-response avenue carrying signals of a standardized
kind...a few wires/traces perhaps...laid out properly
if required to be broadbanded or broad in dynamic
signal range. Not a big thing, but needs some
THOUGHT before becoming hardware.