I don't think much of the discussion has looked very closely at what I
think is envisioned here -- a mainframe which would accept various
"cards" from numerous vendors. As I detailed in an earlier posting, it's
tough enough (and costly) to make a robust interface when a single
company has full control of the mainframe and plugins. But let's think a
little about the problems of making a mainframe which could accommodate
cards from various vendors -- cards which have different performance
characteristics.

The first question is, who will define the interface? Who will dictate
modifications as they become necessary?

Then let's consider a vendor who wants to make, say, an audio amplifier
card. It has digital signal processing with a dozen different modes.
Each mode has considerable adjustment range, for example the width of a
bandpass filter. The interface would have to have pins dedicated to
these functions, and the front panel would have to have switches and
controls for them. How about an oscillator? One might be digitally
tuned, another analog. There are bandspread and RIT to accommodate in
addition. What do we do about T/R switching and timing if it's to be
used in a transceiver? How about shielding specifications so it won't
interfere with other cards?

The only possible way I can see something like this being even possible
is for a virtual "front panel" being done in software and appearing on a
PC screen; only in that way could each card be sure that the necessary
controls would be present. Some sort of serial bus with expandable
protocol would be used for all controls.

Then the question becomes, who will define, develop, and maintain the
software? I can tell you from experience that it's no easy matter to
keep any software working properly as new operating systems, protection
software, and hardware appear. Add the necessary hardware interface to
the equation and the job gets tougher yet. Oh, and what do you do when
key components of the interface become obsolete and no longer available?

It's common for people who've never had to design something which will
be reproducible by the thousands and operate without error, to say how
easy something will be. As one of those people who spent a career
designing just such equipment, I'd bet serious money that the cost of
development and maintenance of the interface would never pay itself back
in sales. Unless, of course, it's done by volunteers. My question is:
Why don't folks like "John Smith" get off their duffs and do it?

Roy Lewallen, W7EL