Time to repair the pdp11/05 (how many of these do I have?)

[czunit]

Ok, with the pdp8/L's in a good state I figured it was time to work on the 11/05. This is an old system, ran many years ago when I would use it for RT11. I kept it because it was heavy, and because it has a really cool front switch panel with LEDs!

Anyway, pulled it out from under the marimba and took a look. First job was disconnect the power cables from the Unibus backplanes and test the power supply. And sure enough the supply wasn't working: It would power up, -15 was there, but +15 and +5 were reading zero. So apart it comes....

 

[czunit]

Take apart time....

This is a 10.5 inch box 11/05 and it has an H750 power supply with an additional transformer and a H744 in there as well. Interesting, I guess that was an extra +5 boost for the additional cards in the system. Regardless, the first place to check was the small micro-fuse at the end of the +24 volt source line. In this power supply the 110/220 is rectified down into a 24 volt source supply which is filtered by a pair of large capacitors, then sent off to a trio of regulators. One (the -15) is sourced through a 10a micro fuse and the +5 and +15 are sourced by a 15 amp micro fuse. Sure enough that fuse was blown.

After cleaning things up I decided to wire in a temporary test fuse using an automotive blade fuse and a set of jumper wires to the old fuse location. This would allow me to test without blowing expensive fuses, and also derate things a bit. I went with a 10a fuse which should be able to clear a 24 volt fault. Wired it in, fired the unit up, and sure enough: The +5 and +15 supplies are working. The -15 is working as well, so I'll look to see if I can find a 48 volt micro fuse. Anyone know the part number?

Next step is to see if that H744 is working, it might be working, it might also be supplying an overvoltage that would crowbar the main power supply. Will see...

 

[czunit]

Interesting. Checking the wiring in the box it looks like sockets 1 and 2 (the ones that power the 11/05 CPU core and G series core memory) are powered by the +5 volts from the H750 and sockets 3-5 have their +5 coming from the H744 supply. That's fascinating, because if the 744 isn't supplying the same +5 as the main supply I think the power imbalance would go across the Unibus jumper. That would be... weird.

 

[czunit]

Ok. With the box re-connected and all backplanes unplugged I get all voltages. +5 looks good, +15 and -15 are also good. The second +5 from the H744 is reading 5.1 volts but that's all with no loads. Plugging in the CPU+memory backplane gets me a good set of voltages, however when I unplug the CPU and plug in the DD11 I see there is 0 volts on the -15 line.

That's a problem. Compounded by the fact that in the DD11 is an old Plessy microsystems core module. I used to have two of them, each one is 8k by 16 bits. One was traded long ago for a pair of Sun 386i's (still have those, they were my DNS and mail servers way back when) and the other is in the backplane.

I'll try pulling it later today and see if the voltages go back to normal.

On a related note, I picked up a MM11-d memory to use as the second 16k bank, however the docs on that say it needs to have +20 volts on the Unibus and -5, I'm guessing this module simply will not work in this computer. Just how many different kinds of voltages were there on the Unibus?

C

 

[daver2]

If you check page 94 of http://bitsavers.trailing-edge.com/pdf/dec/pdp11/handbooks/PDP11_BusHandbook1979.pdf in relation to (for example) the M981 you will observe the * related to the +5V on the M981 being only used for the onboard terminator resistors and that the +5V supply should never be interconnected between individual system units via the UNIBUS.

In this case, having multiple +5V power supplies should be OK. As one power supply failure will remove the voltage from one system unit and should not attempt to power any further system units via the UNIBUS inter-connectors.

EDIT: There was probably as many different voltages as DEC wanted!

I seem to remember some of the semiconductor memory being configurable via links to give the user's options...

Dave

 

[MattisLind]

There are plenty of different backplanes for the 11/05. I have three variants. They differ a bit on where the CPU slots and where to put the KM11.
One in BA11-D with two core memory sets in the CPU backplane. Then there a variant that that came in the slim 5.25 inch chassi. Mine has one memory set, CPU and a number of UNIbus slots. Then I have a 11/05S which comes in a BA11-K. It has a more moden memory system which has 32 kbyte per set of boards. This modern set runs off -5 and 20 V while the older uses —15.

 

[czunit]

Ok. I have the old one in a BA11-D box and including a DD11-B which is NOT a SPC/MUD backplane. Problem is other DD11's require the newer power voltages and harness so I may be stuck with 16kw of memory on this thing.

The MM11-D will not work. Anyone need one?

Maybe I did run this thing with only 16kw to test it, and now that I look at it I see I could never get both of those boards and a RX11 in those 4 slots.

On the positive side, the memory seems to be working and I can probably boot up a small RT11 image on it. Not enough memory to run any games or whatnot, but would load the basic OS. Aside from that any ideas on what I can do with it?

Hm. I do have a DA11-F Unibus window that I think has this size plugs. With that I could hook it up to the 11/24, map 16kw of memory from the 11/24 to the 11/05, and boot a full version of RT11. I wonder: Would a DA11-F work with the KT24 memory management unit on an 11/24?

Another option would be to see if I still have a Plessy Q-Unibus backplane and try running one of my Plessy Quniverters to map 16kw of memory over to the Unibus. Anyone ever used a Quniverter, and were they inherently bi-directional?