• Please review our updated Terms and Rules here

Restoring a DEC PDP-11/05

Here is the relevant part of the schematics.

00DBEC70-BB2E-47F6-BFC9-2822B5799934.png

You need extenders to properly access the signals. I think it is a pretty good investment to ease the debugging. The 11/05 doesn’t seem to have the microaddress on the backplane for debugging purposes like the 11/04.

I repaired a bunch of 11/05 boards myself and just think it is interesting to solve the mystery of a non-working machine. So it is me that should thank you for providing such an interesting problem to solve! A problem that I haven’t seen before.
 
Here is the relevant part of the schematics.

View attachment 1244158

You need extenders to properly access the signals. I think it is a pretty good investment to ease the debugging. The 11/05 doesn’t seem to have the microaddress on the backplane for debugging purposes like the 11/04.

I repaired a bunch of 11/05 boards myself and just think it is interesting to solve the mystery of a non-working machine. So it is me that should thank you for providing such an interesting problem to solve! A problem that I haven’t seen before.
So here’s an odd update:

I appear to have lost my +5v and +15v supplies going to the backplane. The front panel is okay, as it’s fed from a different power supply. The odd thing is that I didn’t actually touch the supply or any adjustment screws at all…

I simply powered the system up this morning to take some readings and saw the 7261 wasn’t getting warm like it usually does. Upon probing some of the VCC pins on certain ICs it became clear that they weren’t getting any power.

The issue seems to be localized entirely to the PS1 (5409728) power supply. I’m getting -15v, but only 1.3-1.4 volts where 5 and 15 should be…

I am hesitant to touch the voltage adjustment pots in case my issue I’m experiencing is transient and I run the risk of overvolting something…
 
You should use a dummy load and check the supplies. It has been discovered that the power supplies behaves very badly if the output capacitor is bad/low capacitance.

It is a switching power supply. But with low capacitance it starts acting as a linear supply. The guys at RICM as researched this issue.

Check if the supply is switching or not.
 
You should use a dummy load and check the supplies. It has been discovered that the power supplies behaves very badly if the output capacitor is bad/low capacitance.

It is a switching power supply. But with low capacitance it starts acting as a linear supply. The guys at RICM as researched this issue.

Check if the supply is switching or not.
Funny you would mention that. I saw the thread about the 5409728 regulator board, which is the one I have. I read somewhere else that no +15v and +5v is
indicative of the bridge rectifier diode D14 failing.
 
Just a little update after the weekend;

I ordered a new bridge rectifier for the power supply which should arrive by Thursday. Additionally, I ordered three double-width riser cards for troubleshooting the 7260 & 7261 modules. When I bought the 11, I was handed a box of assorted unibus and two omnibus modules. I have seven M7800 asynchronous serial line modules- one new in the (opened) box, three M7860 general interface modules, and one M792 bootstrap(?) ROM module. For PDP-8 Omnibus modules, I have one M8650 teletype interface and a W967 wire wrap module (which appears to be new-in-box). Not exactly sure what to do with all of these, though the W967, even though it was intended for a PDP-8, fits into the backplane just fine on my 11. It's completely empty/unpopulated so I was thinking of connecting a few of the pins on the edge connector to test leads to serve as means of measuring rail voltages while the machine is in operation...
 
FWIW, Guy at Shiresoft reworked the KM-11 maintenance boards to a modern form a few years back and those are pretty nice to have when doing work on the 11/05. I don't know if he has any left or perhaps the board art is available. I think I still have the code for the LED separator in my library.
The M792 is hilarious. 32 words of straight up diode ROM. Human-readable and rewritable with a soldering iron. Just enough to boot a device if the device was simple. I have a few and always was tempted to set one up to do some little thing or just be a "killroy was here" somewhere in the I/O page.
 
About the W967. Check the ground and power pins. I believe that card is pre-wired for the GND and power to the bus and I am reasonably sure that those pins are not the same on the PDP-11 bus and the Omnibus. DEC liked to tie all the ground pins together on the cards and this could cause no end of problems if you power up the 11 with one of these cards installed. I don't have one of those in my collection and have been looking for one. I would guess $100 to $150 is what it would sell for on EBay.

About the M8650, it is a highly sought after card these days. In the last year I have seen them go for $292, $172, $140 and the most recent one was listed as a Buy It now for $125 and it sold before I saw it.
 
Looks like it’s PDP-11/05 power supply side quest time.

I replaced the suspected faulty D14 rectifier and no change. Still no +5 or +15. I double-checked every connector to see if it was loose and found everything connected properly. Fuses F1 and F2 aren’t blown, and I’m getting 28-28-59 volts on the connector from T2.
No signs of electrolytic cap leakage or failure. When the power supply stopped working, there was no smoke or flashes…



Edit:

I probed around with my multimeter and double-checked the fuses. I got ~39 volts across F1. Looks like it could be blown. Initially, I only checked continuity, and it likely passed because of the resistor installed in parallel with it. Or, at least that's what I think might have happened...
 
Last edited:
Update: I removed the suspected blown fuse, F1, and verified with my DMM that it was indeed blown. I didn't have any subminiature through-hole fuses on hand, so for diagnostic purposes, I connected test leads to a standard size 15A glass fuse and connected that across where the old fuse was. Upon switching the system on, it blew immediately. Looks like the issue has to be with the +15 and +5 volt supply, as I'm still getting -15 volts and the new bridge rectifier is working properly.

I haven't read too much in-depth about how the H750 PSU works, but I'm wondering if it has a crowbar circuit that might be kicking in when I turn it on. Of course, I'm going to make an effort to rule out anything that might just be causing a dead short to ground first.

Has anybody else experienced something like this before? As I mentioned before, there were no explosions or smoke, I just turned the system on one morning to find that the PSU wasn't working correctly...
 
Is it connected to the machine? Verified that there is no short-circuit in the machine?

Use a bench PSU and try to feed it from. It should work quite well from 20 VDC. Use dummy loads. Not too much loading for initial testing.

If it draws a lot of power with little loading I would suspect fried Q22, Q1 or D5. If Q22 is short circuit the crowbar will trigger.
 
Is it connected to the machine? Verified that there is no short-circuit in the machine?

Use a bench PSU and try to feed it from. It should work quite well from 20 VDC. Use dummy loads. Not too much loading for initial testing.

If it draws a lot of power with little loading I would suspect fried Q22, Q1 or D5. If Q22 is short circuit the crowbar will trigger.
It was only connected to the key switch on the front panel. The power distribution board for the backplane is unplugged from the socket on the H750's regulator board (Connector J2). I was thinking of trying to see how it would behave when connected to a bench power supply. Would I be connecting the +20V to the "Load" side of F1 and the ground to J1 pin 1? (Center tap)

I've attached the drawing I've been working off of; I know it says H740, but the regulator board referenced is the same one that I have.

PSU regulator board.PNG
 
There are some Jerry Walker videos that are quite helpful for testing the power supply. In this series he's restoring a PDP-8/M but the power supply is extremely similar to yours. The Kunkin loads that he uses are very useful. You can find them for a good price on eBay.

I think Jerry starts talking about power supply testing in part 3, and it extends into part 4 and maybe also into part 5.
 
The 8/M (and 8/F) uses the same type of supply.

+5 and -15 is switched PSUs. +15 is linear. The switching is hysteresis switching so the output will be a sawtooth overlaid on the nominal voltage.

Before doing anything, check all big semiconductors.
I have had 2N5302 that shorted.

Good old incandescent car lamps are good as dummy loads.

Just attach your bench supply to the AC input terminals.

EDIT: I know see that you need to connect the bench supply to the upper ac input and center tap. You will not get -15 but that one works so we don’t need to debug it.
 
Last edited:
There are some Jerry Walker videos that are quite helpful for testing the power supply. In this series he's restoring a PDP-8/M but the power supply is extremely similar to yours. The Kunkin loads that he uses are very useful. You can find them for a good price on eBay.

I think Jerry starts talking about power supply testing in part 3, and it extends into part 4 and maybe also into part 5.
Wow! That’s nearly identical to mine. I followed through his video on it testing my regulator board as he did. Sure enough, Q7 (D45H8 PNP power amplifier) was shorted. I’ve ordered a replacement which should arrive early next week.

The 8/M (and 8/F) uses the same type of supply.

+5 and -15 is switched PSUs. +15 is linear. The switching is hysteresis switching so the output will be a sawtooth overlaid on the nominal voltage.

Before doing anything, check all big semiconductors.
I have had 2N5302 that shorted.

Good old incandescent car lamps are good as dummy loads.

Just attach your bench supply to the AC input terminals.

EDIT: I know see that you need to connect the bench supply to the upper ac input and center tap. You will not get -15 but that one works so we don’t need to debug it.

As for testing the +5/15 volt side, I had discovered that the aforementioned Q7 was faulty. Whenever I would power up that side from the bench PSU, it would just short.
 
Good news everyone-

The power supply works again. Q7 seems to have failed and triggered the crowbar circuit.
Last night I swapped it out for a new (old stock) D45H8 and connected a 15A glass fuse with test leads. I jumped the leads where the front key switch connects, plugged it into my variac, and slowly brought up the voltage.
 
There are some Jerry Walker videos that are quite helpful for testing the power supply. In this series he's restoring a PDP-8/M but the power supply is extremely similar to yours. The Kunkin loads that he uses are very useful. You can find them for a good price on eBay.

I think Jerry starts talking about power supply testing in part 3, and it extends into part 4 and maybe also into part 5.
I watched a few of Jerry Walker's videos and will watch more as time permits. These videos are very good and very informative. Thanks for the link.
 
Just a quick follow up and following @MattisLind’s advice on psu testing-

Since I don’t have a fancy dummy load on hand, I’m using an 18W 12V incandescent lamp (the kind you would use with garden lighting or reverse lights on your car). With no load, the 5V output measures
~5.2 volts. With the lamp connected, I’m reading 4.9 volts. Is this too much of a voltage drop? I’m not actually sure what the total load on the psu is under normal circumstances, so I figured I’d ask here before resuming testing.

Another thing, since fuse F1 was bad, had anyone actually found a replacement for these? I have searched for subminiature through-hole fuses, but can’t seem to find an exact match for what I have. Granted, there were no markings on it either.

I was thinking of getting a fuse holder socket to place on the board, so if I were to have a fuse blow on me again, I don’t have to de solder it to replace it…
 
Since I don’t have a fancy dummy load on hand, I’m using an 18W 12V incandescent lamp (the kind you would use with garden lighting or reverse lights on your car).
Before getting the Kunkin electronic loads I was using a bunch of 8 ohm 50 watt resistors with leads that I purchased on Amazon. I crimped piggyback quick-connect terminals on the leads so I could reconfigure the resistors somewhat easily in various series and parallel combinations. Using resistors is convenient because the load is more constant and easily known.

Power resistor values around a few ohms are also inexpensive and readily available. These are sold as audio loads and as load resistors so automotive turn signals work properly when LED bulbs are used in cars with bi-metallic blinker modules.
With no load, the 5V output measures
~5.2 volts. With the lamp connected, I’m reading 4.9 volts. Is this too much of a voltage drop? I’m not actually sure what the total load on the psu is under normal circumstances, so I figured I’d ask here before resuming testing.
Seems like a rather large change between no load and some load, but maybe it's possible. The lowest current that I used was 50 ohms, so 0.1 A. The test results of the H740 in my PDP-8/M might be helpful for you to compare with, which is very similar to your power supply. Ignore the efficiency and calculated efficiency columns. They are wrong because the unloaded 15V regulator also draws some current from the input supply. The input voltage for these tests was 29.48 V (cell A1 in the spreadsheet). When measuring larger currents, be careful to use solid connections and large wire so unintended voltage drops don't affect your measurements.
H740 5V table.png H740 5V output vs current.png H740 5V sw freq vs current.png
Another thing, since fuse F1 was bad, had anyone actually found a replacement for these? I have searched for subminiature through-hole fuses, but can’t seem to find an exact match for what I have. Granted, there were no markings on it either.

I was thinking of getting a fuse holder socket to place on the board, so if I were to have a fuse blow on me again, I don’t have to de solder it to replace it…
I was lucky and didn't have any blown fuses, but I hope I haven't jinxed myself by typing this, LOL. Consequently, I've not searched for replacement fuses. The fuse holder sounds like a good idea, though. You could still put through-hole fuses in the circuit later, if desired.

Oh, there's also this... you might discover burned quick connect terminals and burned Mate-N-Lock connector pins on the wires leading from the +5V output to the backplane. At least two of us have run additional +5V and GND wires in parallel with the originals.
Roland's solution for a PDP-11/10:
Here's what I did in the PDP-8/M, inspired by Roland's fix above:
 
Here is the relevant part of the schematics.

View attachment 1244158

You need extenders to properly access the signals. I think it is a pretty good investment to ease the debugging. The 11/05 doesn’t seem to have the microaddress on the backplane for debugging purposes like the 11/04.

I repaired a bunch of 11/05 boards myself and just think it is interesting to solve the mystery of a non-working machine. So it is me that should thank you for providing such an interesting problem to solve! A problem that I haven’t seen before.

Which specific pins on these chips should I be looking for activity on?

I have my extenders fitted and ready to begin further testing. I put the 7260 and 7261 back in today and adjusted the +5 volt output to exactly 5.0 volts with both cards installed.
 
Back
Top