using new technology on old machines. Was: PDP-12 Restoration at the RICM

[tony duell]

> >
> > Would you put plastic handles on a piecc of antique furniture? Would you
> > make the seatboard for an antique longcase clock from MDF?
> > Both are easily reversable, BTW.
> 
> Sure! Temporarily and reversibly, of course, and I'd hope to replace them with proper stuff when possible. But 
> to bring up an old computer system right now, I'll kludge in what I have available to get it running. In that 
> respect, an Arduino-based baud rate generator could be considered test equipment rather than a component.

Ah, 'there's not the time to do it properly, but there is the time to do it again'.

Why not do it properly first time? What is the rush in bringing up a classic computer? And for a test,
use the TTL pulse generator you have on your bench. Or even an NE555 astable (yes, with a decent capacitor
it is stable enough for a baud rate generator, I've used it). Heck, I've worked on machines that used a 
2 transistor astable multivibrator for the baud clock. Surely you have 2N3904s in the spares box?

Incidentally, if certain horologists heard you would use MDF in an antique clock, you would be
going home with a pendulum rod shoved where the sun don't shine ;-)

> >> If you have the ttl logic bits lying around and know how to use them, fine.
> >> Still would probably need debugging.
> >
> > FWIW I have made programmable dividers on a couple of occasions recently
> > (one was a 100/120 flash-per-second stroboscope, the other was the transmitter
> > half of a modem to talk to TDDs). Both of them worked first time. I guess it's just
> > what I am used to.
> 
> Exactly. And for somebody who doesn't already have a full stock of TTL parts on hand, a different solution may 

'Full stock of TTL parts' ??? You make it sound like I am suggesting using lookahead carry generators,
parallel multipliers, Excess 3 to 1-of-n decoders and the like (all of which exist(ed) in TTL). No, I am suggesting
using some very common counter and gate ICs.

How are you going to fix a TTL-based machine like your 11/730 without spares and without knowing
what the ICs do?

> present itself. I play with gear from WWII military radios up through thoroughly modern electronics. When I 
> work on a WWII radio, it might be considered cheating to poke at it with my Fluke multimeter, Tek DSO, HP 
> spectrum analyzer or HP synthesized signal generator (the latter two of which are slaved to my GPS-disciplined 
> frequency standard), but those are the tools I have on hand, so those are the tools that I use.

It is cheating :-)

More seriously, tools are one thing. And had it been suggested that as a quick fix you took the TTL output
from a sig-gen or took the output and clipped it to TTL with a transistor buffer then that would IMHO be
reasonable (even if said sig-gen contained many times the number of components of the rest of the machine). 
But to make a custom solution that is over-complicated IMHO is the wrong way to do it. 

You might well inject your HP sig-gen into the mixer (first detector) stage of your WW2 radio to get it going
if the local oscillator had failed, or if you didn't have the right crystal, or whatever. But I hope you wouldn't
replace the valve (tube) based local oscillator in such a set with a digital synthesiser as a permanent 'repair'

-tony