DEC H7140 (11/40) power supply revisions and general advice?
derschjo at gmail.com
Sat Oct 10 10:45:59 CDT 2015
> On Oct 10, 2015, at 4:15 AM, Mattis Lind <mattislind at gmail.com> wrote:
> 2015-10-10 5:42 GMT+02:00 Josh Dersch <derschjo at gmail.com <mailto:derschjo at gmail.com>>:
>> Hey all --
> First of all I haven't poked around in the PSU of my 11/44 since it worked
> fine when powered up. But the design of the H7140 is a lot similar to the
> design of the H7104, the PSU of the VAX-11/750 which I worked quite a lot
> http://www.datormuseum.se/computers/digital-equipment-corporation/vax-11-750 <http://www.datormuseum.se/computers/digital-equipment-corporation/vax-11-750>
>> Once again, I find myself in over my head debugging a power supply, this
>> time an H7140 from a PDP-11/44. Here's the skinny:
>> I examined the supply physically before experimenting and found a
>> capacitor on the Bias/Interface board that was leaking, bursting and rather
>> burned-out looking (not a great sign) -- this is capacitor C4 in the
>> printsets on Bitsavers (
>> Everything else looked OK physically; I replaced the obviously bad
>> capacitor at C4.
>> When power is applied (plugged in, breaker switch flipped to "On") the
>> relay does not click - based on my readings of the manuals this should
>> happen after the bias voltages are up to spec. I measure 308VDC on the
>> lugs on the top of the memory board, so that's at least something working.
>> Getting to other points to test voltages is a bit more difficult,
>> especially with those high voltages in the way, what a nice design :).
> The previous owner said that it happens that the relays get stuck. The
> relay is there to limit the inrush current so when it clicks it bypasses a
> couple of power resistors. Regardless relay state you will have the full
> voltage over the capacitors. (at least this is the case in the H7104). If
> the relay get stuck then the power resistors get very hot. The relay drive
> circuit detects that the voltage is high enough then turns on the relay.
> Maybe something in this circuitry is bad. In this case you could force it
> to on but then I think you should power it from a variac
Thanks. I should have mentioned, but I tested out the relay and it’s not stuck and seems to be working fine. It definitely looks like the relay control circuitry is not running properly.
>> Switching the front panel switch to "Local" (or any other position) has no
>> effect -- no fans, no LEDs, nothing. I've double-checked all the wiring
>> and everything looks OK.
> Have you verified that you have the 12V BIAS voltage? If not you could
> probably generate the 12V BIAS voltage using a lab power supply just to see
> if the main switcher is working properly.
> I used an insulation transformer and a variac and then forced the relay
> drive to on. In the H7104 the same circuitry also enabled the main switcher
> so unless I did this there were no fun at all. I supplied 12V using a bench
> supply. It made it possible to check everything was working ok without any
> high voltages.
Interesting idea, I’ll look into that. Thanks!
>> Capacitor C4 looks to be involved with the START-UP DRIVE signal circuitry
>> (which drives the relay) so the behavior I'm seeing makes sense if C4 died
>> and took a couple of things with it (or if something else died and took C4
>> with it).
>> Here's where it gets kind of odd -- I spent some time testing diodes and
>> transistors in the related area near C4 and while doing so I noticed that
>> there are four diodes (D1-D4) listed on the schematic that are missing from
>> my board. "Missing" as in someone clipped them out at some point -- there
>> are just nubs of the leads left. I'm not sure why this would have been
>> done, but there were a number of ECOs applied to this board (a few wires
>> and resistors added) and I don't want to assume that if I just put four new
>> diodes in that it won't cause other problems. Anyone know if there were
>> other revision levels of the bias/interface board that would have done away
>> with these diodes? Anyone have an H7140 they can easily crack open to
>> compare? (It's actually relatively easy to get to, if you can get to the
>> Thanks as always,
More information about the cctalk