Restoration: how far should it go??

From: Tony Duell <ard_at_p850ug1.demon.co.uk>
Date: Sun Jun 15 14:55:50 2003

> Really? I have an H740 power supply sitting inside the chassis of a
> PDP-11/05 that wants to argue with you. As it is, half the +5V regulator
> has been removed, and is going to have to be replaced. It's not going to
> cost that much, but still, I fie you to call that a "replica". The

I see no problem in rebuilding a PSU, but I'd want to keep to the
original design if at all possible...

> important part isn't "Oooh! Shiny old collectible!", it's "watch me make
> the binkenlights do foo after I toggle this program in on the front
> panel." There's a PDP-11/45 that my friend and mentor essentially rebuilt.
> New power supply, grafted-on LTC, etc. Yeah, you can argue that replacing
> a power subsystem is somehow devaluing the system in some way. The system

Yes, actually I would claim that. The PSU _is_ part of the computer.

> will be worthless, though, if you use a period power supply. When linear

Why? My 11/45 (early version ,too) is still running on the original PSUs. All
I've had to do is rebuild one -15V brick...

> supplies fail, they can do interesting things , like toast your logic.

FWIW the 11/45 PSUs (and AFAIK all PDP11 PSUs) have switch-mode
regulators, at least for the +5 and -15V lines. YEs, they can fail, and
yes, in theory they can put overvoltage onto the logic, but DEC put
crowbars everywhere.

I've had mainns-side SMPSUs fail too, and put nasty HV spikes on
the output (an open-circuit or high ESR output capacitor will do that!),
and I've had regulation circuits fail in modern SMPSUs...

Perhaps it's just me, but I've had a lot more problems with relatively
modern 'commodity' PSUs that with the units in 1970s minicomputers. And
at least the latter come with printsets so I have a hope of fixing them.

> What is the point of having a pristine, unmolested logic board of
> burnt-out chips? A different friend of mine recently acquired a PDP-10

That's why you alwys check PSUs on dummy load after installing the
machine or workign on the PSU. And why you check the crowbar (NOT by
triggering it! -- with a dummy load, put a lamp in series with the SCR
anode and then twaek up the PSU output to ensure it will trip -- assuming
the PSU can be turned up that high). In fact I'd probably replace the
crowbar components with new, tested, parts in any really rare machine.

-tony
Received on Sun Jun 15 2003 - 14:55:50 BST