Wow, I go away for a week and it takes me most of the following week to get
through PICLIST post.  But it's all good stuff.  Thanks to all out there.
Couple of comments.

1.  Old teletypes.  No-one's mentioned the real oldies I remember from the
UK GPO in the sixties - painted dark green and weighed a ton.  They had the
interesting feature that the printing was always a character behind.  As
each 5 bit code came in, it printed the previous character then set up the
printhead for the next.  Was this the first example of PIPELINING?

2.  PLDs.  My favourite is the 18CV8.  This is a 20 pin device with 8
outputs.  Key feature is that each output macrocell is truly universal.
That is, polarity, feedback type (registered, combinational or pin) and
output type (reg or com) are independantly set.  Main drawbacks are
single-sourced, not supported by most current design tools or cheap
programmers.  I therefore recommend the 22V10 against the 16V8, if only
because its output macrocell is 'clean' - the 16V8 is a mess because it was
designed to be a universal replacement for 16R4, 16R6, 16R8 and 16L8, where
the com and reg outputs had different feedback and output enable
arrangements.  The 22V10 also has the advantage that it has lots of product
terms - 16  on two of its outputs.  Final benefits are that everyone makes
one, and many CPLD and FPGA manufacturers quote their products in terms of
multiple 22V10 equivalents, so you get a good feel for what you'd be getting
into.

3.  Overvolts protection.  This comment refers to recent posts on both MOV
(etc) devices and in-system programmable logic.  I had a bad experience a
year ago with Lattice ispLSI parts, when I managed to blow up several.  It
took a while to work out what was going on, and I urge anyone thinking of
using in-system programmed ICs of any type to read carefully.  The system
under development was a prototyping aid.  In use, it is possible to short
out Vcc (5V) and Gnd.  OK, so there's no need to design the system so that
such a user error won't cause problems, but the system MUST be undamaged by
it.  Lattice ispLSI WAS damaged.  The mechanism is that a modern PSU can
respond pretty rapidly to changes in current demand, and can push out a
hefty lump of amps for very short periods.  If the short is not a nice,
hard, permanent thing, but the sort where two bare wires brush together, the
Vcc line will bounce as the short re-opens temporarily - and any inductance
around will boost the bounce.  Testing (without the Lattice CPLDs in
circuit) showed excursions on the 5.0V Vcc line of up to 10.4V relative to
0V, with typical peaks of 8.3V.  These lasted for several microseconds.
There was absolutely NO adverse effect on other components including 74LS,
75ALS, 74HCT, 74ABT and an 18CV8.  The Lattice chips have an on-chip charge
pumps and related features to provide the programming with only 5V
available, and it appears these features are the reasons they failed with
very low resistance between Vcc and Gnd pins when Vcc pulsed too high.  If
you look at the data sheet for almost any 5V logic IC, you'll find the
absolute maximum Vcc quoted as "7V".  In the case of Lattice ispLSI, that
absolute maximum is a true, hard, cross-this-line-and-you-die limit.  OK.
So now I had to stop my 5V line exceeding 7.0V.  Not 7V, 7.0V - quite a
difference in practise.  ALL the standard voltage clamp products, even those
rated as 5.6V, actually have a rounded "knee" on their characteristics, and
don't actually clamp hard by conducting amps until around 10V.  The ONLY
device I could find which worked was the SGS-Thomson 1N5908.  By fitting
this, and a 100uF axial tantalum to slow down the speed of the voltage edge,
there have been zero failures in use.  SGS-Thomson used to make devices
which (on paper) would have been even better, the GMP5 and GMP5B, but these
ceased production seveal years ago.  To summarise, in-system programming is
great, but you never get something for nothing.  The cost of 1N5908s in all
your products might soon exceed the cost of buying a programmer for a
non-isp 22V10!

Tim Forcer            Tel: (+44) (0)1703 593362
                      Fax: (+44) (0)1703 592053
                    email:  tmf@ecs.soton.ac.uk
Department of Electronics & Computer Science
Room 3005, Building 35
The University, Southampton, SO17 1BJ     UK