PDP-11/44 Restoration

[RSX11M+]

Thank You AK6DN - PM sent.

While I am curious what inspired you to do that work, it is nonetheless very relevant and timely to this discussion.

As a designer, those limits "feel" to me about right for a largely MOS and LS based design.


I'll be in and out all day today watching this thread and will have more time to devote to continuing schematic work tonight, assuming someone else hasn't got you going by then.

You guys are great.

 

[Old Computers]

Thank you all for looking at the schematics. I can read a schematic and see what part is what, but I cannot tell what the circuit does.

I will go ahead and try to load the system more, and I will remove the cards except for the CIM. How much current should I aim for? My multimeter can only handle 10A max. I might be able to get some high power resistors to act as the load, or would some incandescent light bulbs work okay?

Could the ripple in the supply be causing some issues too? I forget what page it is on, but the trace I posted is very similar to the trace I am getting now.

 

[Qbus]

To a man a with hammer every problem looks like a nail. My hammer is still looking for a poor connection in the feed back circuit from the unibus backplane to the power supply regulator, that’s because in playing around with restoring lots of old equipment that’s always been my nemesis. When you are reading the status of the five volt buss there can be an issue with reading it across the two big straps that carry the five volt supply to the backplane that it may be a small amount higher due to the losses that are encountered in the backplane itself. Have been tied up with a thousand projects with the semester getting ready to start and other things going on but will try looking at the manual this week. Think we all have to remember that for a while the system was up and running and then it began to fail again, biased on that I would ten to think that its not as much a component failure as a bad connection and although I have no experience with a 44 switching supply have played around enough with switching supplies as to not buy the idea that without a minimum load they wont regulate and so I am not buying the idea that the problem has something to do with not drawing enough current. Also would like to see the +5 bus between 4.9 to a max of 5.2, would not expect less of the DEC engineers back then.
5.6 or 5.7 just looks too high unless that’s the output driving the backplane, assuming the connection from the power supply to the backplane and the traces on the backplane are around 0.5 ohms you would expect to see about a 0.6 or 0.7 volt drop from the output of the supply to the points where the +5 volt bus is sensed, so the question is what’s the voltage on the return to the power supply? Also do recall that just about everything in all the switching supplies is referenced too the bias supply so that would be a good thing to check. What were you using for the load last time? You will need a good current sink because if you use resistors they are going to be a real small value and are going to develop a lot of heat, remember that at 10A@ 5 volts with a 0.5 Ohm resistor you will develop 50 Watts of heat, at 15 amps using a 0.3 Ohm resistor that will be 75 Watts of heat. I have a huge heat sink with three parallel 2N3055 that I can control the bias on and use that as a torture device for power supplies around the shop and that will work well.
So there you go, that’s the gauntlet. Lot to say and speculate about something I have no experience with or have never seen before but wont be the first time I have said or done something stupid, just ask my wife.

 

[RSX11M+]

...still looking for a poor connection in the feed back circuit from the unibus backplane to the power supply regulator, that’s because in playing around with restoring lots of old equipment that’s always been my nemesis. When you are reading the status of the five volt buss there can be an issue with reading it across the two big straps that carry the five volt supply to the backplane that it may be a small amount higher due to the losses that are encountered in the backplane itself.

Feedback - Yes - my fear too. This is why I say the measurement point has all to do with this. (but you said it better)

What were you using for the load last time? You will need a good current sink because if you use resistors they are going to be a real small value and are going to develop a lot of heat, remember that at 10A@ 5 volts with a 0.5 Ohm resistor you will develop 50 Watts of heat, at 15 amps using a 0.3 Ohm resistor that will be 75 Watts of heat. I have a huge heat sink with three parallel 2N3055 that I can control the bias on and use that as a torture device for power supplies around the shop and that will work well.

Considering it's a +5VB is 120A rated I'd like to load it with something closer to 600W. I was sitting here earlier thinking "what commonly available item would be able to present that kind of load?"

• Shunt wound motor - like a car starter? [unloaded]
• 12VDC Heater element from a Boat?
• A couple of car cigarette lighters? [not sure if they'd trip or how long they'd take it]
• De-ice strip on a car window

The loads in total need to present more like 0.05Ω (500W) - 0.1Ω (250W)

 

[Old Computers]

For the measurement points I measure all voltages at the different pins on the backplane, not at the PSU's output. This is also true for the Scope. I connect the red multimeter lead to the pins that have the voltage, and I connect the common probe to the chassis. I do, however, connect the dummy load directly to the output of the PSU.

The dummy load consists of two old car headlights' high beam filaments (the low beams are burned out). I have some speaker wire that connects to the PSU's output, goes through the multimeter then to the load, then back to the other side of the speaker wire which connects back to the PSU's return. The multimeter reads around 5.37A then gets a little smaller as time goes on (more around 5.28-5.33A).

As for the load I might be able to find some of those items. I know I have a small precision resistor that is around 0.1ohm, but it is definitely not rated for those wattages. Qbus, your transistor load also sounds good. Is that difficult to make?

 

[Qbus]

You may want to look at EBay for 0.3 Ohm 100 watt resistors or other low value high wattage resistors. Three 0.3 in parallel will give you 0.1 Ohm at 300 watts and just one 0.3 ohm resistor you will be loading down the supply and you can add the second and third to extend the amount of punishment. Just remember that you will need heat sinks for them being they will be dissipating lots of heat, also keep in mind that you will need heavy wire between the load and the power supply. Another thing to think about is # 12 copper wires is about 1 ohm per thousand feet so a 100 foot roll of #12 will look like a 0.1 Ohm resistor and may be easer to connect in. The transistor thing is just a old heat sink assembly scavenged from a linear power supply that I gutted with all the big pass transistors connected with the emitters together and grounded thru the original 0.1 Ohm limiting resistors and all the collectors tied together as the input and all the bases tied together and then use a old powerstat (rheostat) with the one leg tied to the collectors, the wiper tied to the bases and the third leg tied to ground. By controlling the base you can make the entire assembly appear as a virtual resistor that can be adjusted to get the desired load on the power supply under test, once again be aware that lots of heat is generated! You would be surprised by the abuse that those old transistors will stand up to.

 

[RSX11M+]

For the measurement points I measure all voltages at the different pins on the backplane, not at the PSU's output. This is also true for the Scope. I connect the red multimeter lead to the pins that have the voltage, and I connect the common probe to the chassis. I do, however, connect the dummy load directly to the output of the PSU.

Placing the load point near the supply is ok. Where do the sense leads attach? (missed that detail)

Measuring from the backplane - particularly the furthest end is the best point to measure (is that where the CIM would usually be?) however, it would be best to place the minus test lead on the backplane in as close proximity to the positive lead as practical. I know is seems a trivial difference, but with heavy currents running it will be important. (That's why there are both positive and negative sense leads)

Ray's transistor / heatsink dummy load is clever, but it's really only worthwhile if he got those parts by scavenging what he had, not if he had to go procure them.

Another thing to think about is # 12 copper wires is about 1 ohm per thousand feet so a 100 foot roll of #12 will look like a 0.1 Ohm resistor and may be easer to connect in.

This is a good idea for a short term load, but I'm afraid it will heat too quickly and turn the #12 wire into a fuse link. (that's 50amps 250W). I would be much more comfortable if that were #12 Romex, and he used it as a loop.

+5V power in on BLACK at the near end, Wire nut BLACK to WHITE on the far end, and wire nut WHITE to GROUND on the near end (where BLACK is fed) and then the GROUND lead (uninsulated wire) on the far end to the 5V Return side.
​

That would be 0.3Ω (16.6A 83W) and be underneath the wire's safe current rating. (I'd probably still keep the wire loop in a bucket of water or ice)

If that doesn't present a heating problem, you could then progress to a 0.2Ω (25A 125W) load by using the near WHITE as the return point.

NOTE: Do not let water enter the open ends of the wire's jacket or pierce the jacket anywhere. You will still be able to use it for it's original purpose if it's not compromised.

Hope that's clear.

That is probably better than a cigarette lighter, which is only about 1.2Ω hot, I come to find out.

Also, I don't think I'd add that load after the unit was powered. I'd have it attached before hand. If the supply trips immediately when you try it, don't try again.

Found a TM on the BA11-A (I think you guys were already reading it)

 

[Old Computers]

@Qbus: Thanks for the resistor tip. I think that is what I will probably go with.

As for the wire idea I will also consider that, but I am thinking that the resistors will be easier to deal with.

@RSX11M+: Would measuring the voltage at one slot, but using a different slot's ground be okay? Since the pins are so close together I am afraid of shorting out the 120A line.

For the sense leads I actually do not know where they are. I just know that they are present and it comes from the backplane. I suspect that the flex print cables also carry them based on the block diagrams I have seen, but I am probably wrong.

So I should switch the load on after it is powered up? How long should I wait? I could probably get a breaker or something similar to connect the load through.

The technical manual has been quite an asset!

 

[RSX11M+]

...So I should switch the load on after it is powered up? How long should I wait? I could probably get a breaker or something similar to connect the load through.

Please re-read:

I don't think I'd add that load after the unit was powered. I'd have it attached before hand. If the supply trips immediately when you try it, don't try again.

@Qbus: Thanks for the resistor tip. I think that is what I will probably go with.

Best solution if you can find em - but remember they'll have to be cooled somehow. These usually look like brown ceramic spools with large connection lugs at each end.

As for the wire idea I will also consider that, but I am thinking that the resistors will be easier to deal with.

Not a bad improvised load. [agree - resistors are better]

@RSX11M+: Would measuring the voltage at one slot, but using a different slot's ground be okay? Since the pins are so close together I am afraid of shorting out the 120A line.

Sure adjacent slots are fine. Actually, in an empty BUS - devoid of cards, it won't matter. If you're that nervous about it, I suggest you make up some insulated probes or tac solder a permanent extension for your testing.

I'm not quite yet to the fine detail of knowing where the sense leads are either. It will come.

 

[Old Computers]

Ah okay. Thanks for pointing that out.

Yes the load is attached at power up, and I will do the same thing for the new load I make.

 

[RSX11M+]

Ok... bigger post is updated.

 

[Gerardcjat]

@RSX11M+

Quote : "Found a TM on the BA11-A (I think you guys were already reading it) "

Where can I get this TM, in order to fully "follow/understand" the ( very ) technical points discused here ??

Many Thanks.

 

[Old Computers]

@Gerardcjat: You can find the manual at Bitsavers. I cannot remember what folder it is under at the moment though.

I obtained some high power resistors today. I have a 0.1ohm resister rated at 140W, two 1.5ohm rated at 25W, and two 0.2 ohm rated at 25W.

I did some calculations and using just the 140W resistor would draw 50 amps at 250W. Should I use a current limiting resistor so as to limit the wattage so the big resistor can handle it? Also is there any combination of these resistors I could use that would present a good load yet the wattage would not exceed the rating? I am also going to think of a cooling system for the resistors.

Now I just have to figure out a suitable wire gauge and connector. Using this setup I will definitely not be able to measure the current being drawn.

 

[Gerardcjat]

@ Old Computer

Quote : "I obtained some high power resistors today. I have a 0.1ohm resister rated at 140W, two 1.5ohm rated at 25W, and two 0.2 ohm rated at 25W."

Great, I was about to suggest you buy some as I just did.

From China ( Sorry for that ... ) they are affordable, as I just bought Five 1 Ohm 50 W for around $12 ( w/ one extra free ( 0.01 ohm 50w ) and free shipping ).
From an other file, you probably have understand that my plan is to dummy load a PDP 11 Power Supply
So, now, we are almost on the same ship

 

[Gerardcjat]

OK, now some maths ....

From what you get, it is clear that you CANNOT put directly the 140 w Resistor to the 5 V.
it is going to take 250 W and will not last long !!!
My best suggestion, so far :
- put the TWO 0.2 Ohm in parallel --- you get 0.1 Ohm 50 W
- then put the 0.1 Ohm 140 W in SERIE with that assembly
Now, you have : Total Resistance : 0.2 Ohm Under 5 V. that is 125 W LOAD ( Drawing 25 amps ) that will "split" this way :
65 W for the 140 W resistor ( so OK !! )
32 W for each 25 W resistor This is not very good, but at least it's better than nothing.
You have to put all that stuff on a large heatsink and have a fan blowing strong on it.
I "think" that if you stay under 4 minutes test time, the 25W resistors may stand it.

About current measure : that depend on how accurate are your resistors .... Let's say they are 5 %. Measuring voltage at one resistor level, will give you the amps draw ( at +/- 5 %, which seems acceptable ).
At least, you will, thru that measure, have an idea of what is "lost" ( dissipated ) in the connecting wires, I mean "The connecting wires reducing the calculated theorical drawn current "

 

[Gerardcjat]

CAN'T you get an other 0.2 Ohm 25 W ??

It would be just PERFECT, connecting all three 0.2 Ohms in parallel ( and then in SERIE with the 0.1 Ohm, as previously said ), you will get :
156 W LOAD ( 31 Amps drawn )
with 97 W on the 140 W resistor
and 19 W on each 25 W resistor

 

[RSX11M+]

@Gerardcjat: You can find the manual at Bitsavers. I cannot remember what folder it is under at the moment though.

EK-BA11A-TM-003_Aug83.pdf

I obtained some high power resistors today. I have a 0.1ohm resister rated at 140W, two 1.5ohm rated at 25W, and two 0.2 ohm rated at 25W.

I did some calculations and using just the 140W resistor would draw 50 amps at 250W. Should I use a current limiting resistor so as to limit the wattage so the big resistor can handle it? Also is there any combination of these resistors I could use that would present a good load yet the wattage would not exceed the rating? I am also going to think of a cooling system for the resistors.

Is it too late to up your order of those 0.1Ω resistors? (BTW- Where did you find them?) With 4 you could make a nice series parallel bridge that could draw 50A and sustain 250W (although 75W 0.1Ω resistors would do as well) . Do it again (8 resistors total) and that would be sufficient load to test this supply to any limit I would consider.

The two 0.2Ω, even in series will still be over their power limit. (~31.25W each) Three in series would be ok. (~14W each)

On Series dropping resistors - Don't forget that dividing the resistance also divides the power.

Gerardcjat's 0.1Ω in series with three paralleled 0.2Ω would survive, but only draw 38 amps. Better than your existing load, but not yet what I was hoping for.


That 100' #12 Romex looking any better today?

Now I just have to figure out a suitable wire gauge and connector. Using this setup I will definitely not be able to measure the current being drawn.

Let's decide on the components first, then the wire - ok?

Measurement will be calculated indirectly by voltage drops. That's the handy thing about using resistances of 1 or .1 or .01 etc... makes readings direct except for the decimal point.

 

[Qbus]

I have an advantage being that I have a well-stocked electronics junk yard at the house so heat sinks are no issue. Maybe what you can do is look for old UPS systems like the big rack mount APC 2200 or 2500 watt units and you can get them for almost nothing after the internal batteries are dead. There you will find lots of high current wire with connectors and large capacity heat sinks, all the parts you need for building a power supply torture device.

 

[RSX11M+]

I have an advantage being that I have a well-stocked electronics junk yard at the house so heat sinks are no issue. ...

I'm curious how you decide what to cannibalize?

I'm in much the same state of general equipment collection (well, perhaps worse) but I don't seem to have the heart to tear up stuff. (most is, or was, working when taken from service)

For example - I must have 10 UPS's around here. All worked, but eventually their batteries failed. (got those too) I continued saving them because as the economy fails, I expected batteries to eventually cost less than new units. But this hasn't happened in over 20 years now.

Another example is a disused rear projection HDTV. It's 57" and still "works" but was never a good receiver. It has no modern inputs (no HDMI) so it's can't be hooked to anything made today. Yet I can't bring myself to salvage it for parts.

Etcetera ad nauseum.

Just a little semi-related thread hijacking till activity resumes.

 

[Old Computers]

I could get some more, but the price for those resistors was a bit steep. I might go back to where I got them and see if they could lower the price any or have any similar resistors there.