• Please review our updated Terms and Rules here

PDP-8/e H724 Power Supply

m_thompson

Veteran Member
Joined
Jul 8, 2014
Messages
1,985
Location
Rhode Island, USA
About 10 months ago the +5 output in the H724 power supply went out of regulation, tripped the Crowbar, and blew the 25A fuse. The LM723 regulator IC on the A2 regulator board had died. Replacing the LM723 fixed the +5V output.

Last week while I was replacing the front panel plexi and bulbs the -15 supply died. The -15 supply also uses an LM723. This one had rusty leads, so I replaced it. It didn't change the -15V behavior so I connected a clip lead to each pin on the LM723 to measure the voltages. After removing the A2 board for about the fifth time the -15 started working. Maybe the connector for the A2 regulator was dirty?

The system is back together and working OK. Now I can go back to work on the RK05 drive, and the RK8E controller. The M7104 board in the RK8E had a 100ns delay line glued on as an ECO. The delay line was ripped off before I got the board, but I have a tube of them. Debugging the RK8E should be lots of fun.
 
Mike, try running the power supply isolated from the omnibus and see how long it runs. You then may try load testing the -15 volts, but just connecting a resistor. This should tell you whether the fault is in the power supply or on the bus. Good Luck Mike
 
I did all of my debugging today with the power supply out of the chassis, so the Omnibus is OK. The two final stages of the -15 supply, Q300-Q304 are all OK. The voltage on the base of Q300 is way to low. That voltage comes from the LM723 on the A2 board. Now I need to find out why.

Hopefully when I remove the supply again it will stay broken long enough to find out what is intermittent.
 
The power supply is out and back on the workbench.
Of course it is working OK now.
I put a resistor load on the -15 and will run it for a while.
 
Have you looked at R5? The voltage adjusting pot? Sometimes the wipers on old pot's will lift and open up. With your resistor load adjust the pot and see how smooth the voltage changes. Mike
 
For the few times that it was on the workbench and in a failed mode I tried adjusting the output voltage. Changing the Trimpot didn't make a difference in the output voltage so I suspected the Trimpot. We have replaced Trimpots in other DEC equipment, but this one measured OK. I need to see the voltages on VR1 pins 8, 6, 2 & 3 when it is broken.
 
I resoldered everything on the A1 regulator board and it is still broken. The output from the LM723 regulator is 3.64V, so I would expect Q300 to conduct a little. That would get Q301-Q304 to conduct a little, which would increase the -15V output, which would lower the ground (-15V) on pin 5 of the LM723, which would increase the LM723 output on pin 6, and eventually get the output up to -15V. That is not happening. I will pull the A1 regulator card and connect a current limited lab power supply to the base of Q300 and inject a little current. That should let me regulate the output to -15V. This experiment should tell me if the problem is on the A1 regulator card, or in Q300-Q304.
 
I spice modeled the -15V power supply so I could see exactly how it works. The LM723 is a very good regulator. Spice made some strange looking waveforms because the + side of the -15V and the - side of the +5V are common. The +5V powers the LM723 with the ground of the LM723 connected to -15V.

Now that I understood how the -15V supply worked, I started measuring voltages again. I used the -15V connection for ground to make interpretation of the measurements easier. The voltage on the base of Q300 was 3.64V, and 2.91V on the base of Q301-Q304. Very low, but the transistors look OK. The voltage drop emitter to collector on Q301-Q304 looked reasonable considering the low base voltage.

I measured voltages relative to the common ground for all voltages. The -15V was about 2.75V, so nothing was shorted. The emitter of Q301-Q304 was also 2.75V. Um, that's not right, it should be ground. The -15V fuse F300 was OK. I could see 2.75V on both sides of the fuse holder. I chased the ground connection from F300 to the bus bars, and found that the screw that connected one of the +15V capacitors to the bus bar was loose, and it also had the ring terminal for the -15V ground on it. That explains why the power supply was position sensitive.

A little over a year ago the +5V regulator died. When testing the repair I used the capacitor screws to attach lots of ceramic resistors for loads. I forgot to tighten one of the screws, and it happened to be the ground for -15V.
 
It is running again. It just booted OS/8 from a floppy.

SPICE is an electronic simulation tool. I use the free LTspice. I entered the schematic for the -15V part of the 8/e power supply, fed it 30VDC, put a load on it that starts at 0 Amps, jumps to 8A, and then back to 0A. I can probe the schematic and see current and voltage waveforms. You can actually see a lot more than you can with a 'scope. Using this tool you can really learn a lot about how parts of these systems behave.

We used LTspice to model the whole H740 power supply that is in a PDP-12 DW8E Omnibus expansion chassis. It is the same power supply as in the 8/f and 8/m. We were able to determine that the H740 is actually a switcher and is very sensitive to the quality of the output capacitors. That let us understand why the power supply was misbehaving and fix it.
 
Could you please share your SPICE model? I've just repaired the +5 regulator in one of my RK05 drives (replaced blown pico fuse) and it seems to "wobble" a little - voltage is 5.002 but occasionally seems to drop about .5 volts for a few seconds then pops back up. The blown fuse was the result of a user-induced short while attempting to measure ripple on the +5v bus of the drive. :>(

Maybe I can model the supply and get a better idea of where the failure might be.

Thanks,
Jack
 
Back
Top