30 May
2006
30 May
'06
3:19 p.m.
On 5/30/2006 at 1:28 PM Billy Pettit wrote:
YEs, but most of those devices are well-understood, and there may well be
workarounds. If you take a 1920's radio using UV201As, you could probably
kludge FETs in place of them (doesn't some compay actually sell little
PCBs to upt over the pins of dead UV201As to do this?). Germanium diodes
can often be replaced with shottky diodes. And so on.
Some of the odder TTL chips are getting hard to obtain now, but at least
data sheets exist (I will _never_ give up my shelf of TI TTL
handbooks...). I can see myself programming some modern CPLD-thing to
replace an obscure TTL counter/decoder/whatever. Wasteful of resources in
the new chip, sure, but it'll keep my PDPs, P800s, HP9800s, etc running.
Some time, we have to let go and upgrade as much
as it is unaesthetic or
displeasing. When?
Now that's an interesting question.'
I think if "maintainability" is plotted against time, it often looks like
a bell-shaped curve. Near the peak of the curve, we have devices that use
available commodity components (e.g. 7400-series TTL and/or discretes), so,
for example, it's possible to keep a PC XT mobo going for a long time. Way
off on the back end of the curve are devices constructed from things that
were once commodity items, but which are old enough to be rare
(point-contact transistors, germanium diodes, UV-201As, your 160A core, etc.). Off the front end of the curve are those
devices constructed from
devices that, although only slightly out of date, are no longer available
and whose specs may never have been published outside of an internal
company document (try to replace a house-numbered ASIC on a 1990's PC
mobo).
That's the big problem. Even though you probably could replace one of
these ASCIs with a modern FPGA suitably programmed, since you don't have
the specs of it, you have a very long day ahaed of you.
-tony