• Please review our updated Terms and Rules here

H740 Power Supply, what's your Q1?

gwiley

Experienced Member
Joined
Nov 12, 2021
Messages
249
Location
San Diego, CA, USA
The simplest circuit in the H740 supply (PDP-8/M), the +15V output, is giving me trouble. I'm seeing ~20Hz oscillations when the +15V output is loaded above 0.5A. Q1 in both of my supplies is a 2N3791, a PNP power transistor (not a Darlington). However, the schematic identifies Q1 as part number MJ2500, which is a Darlington. I'm curious what Q1 part number others have in their H740 supplies.

I've tested all of the transistors in the +15V circuit and swapped Q1, the pass transistor, with the same part from another supply. I noticed the circuit that drives the base of Q1 has an asymmetrical impedance. Q3 pulls it high but a 10K (R38) to ground pulls it low. I tried reducing the value of R38 to 3K, and the oscillation now happens at a higher load of around 0.8A. Reducing R38 by a lot isn't a good solution though because 39 volts squared / R38_small can dissipate a lot of power. I'm thinking that swapping Q1 with maybe the correct MJ2500 is the right thing. Seems like the MJ2500 Darlington is the right choice here. With a 10K pulldown it needs a pretty high beta to drive an amp on the +15V output.

I checked Roland's post: https://forum.vcfed.org/index.php?threads/what-did-i-do-to-my-pdp-8-today.77577/page-33#post-1247620 and I see an MJ2500 in his 11/10.

Does anyone else have a 2N3791 in their supply? I wonder if there was an ECO to change this transistor and I'm just not finding it.
 
Both H740 in my PDP-11/05 and in my ME-11L have MJ2500s. Both supplies contain rev D 54-09728 regulator boards.
 
Both those supplies came from the same "repair" shop, so all bets are off as to suitability. Do they look like they've been replaced?
I inspected the joints before desoldering and they looked really clean. Kudos to the repair shop for their workmanship 😀

Thanks for commenting about this… helps solve the mystery of the non-Darlington. I was thinking there was maybe a cost reduction or design improvement ECO that I missed.

I have some MJ2500’s on order which should arrive early next week. I temporarily installed two 8.2K resistors in parallel with R38 which works up to about 0.7A.
 
That power supply needs a darlington to reach the specified 1A on the +15V. But since the +15 was used for the RS232 output and the pull up resistors on the bus load, that 1A was probably never reached. Can you check R33? Is that 0.39 Ohm? This resistor is the current limiting resistor. When the voltage across this resistor is about >0.6V, Q2 starts to conduct and limits the output current. It would start to limit at 0.6V / 0.39Ohm = 1.5 Amp.

If that resistor is modified to a higher value the current is limited to a lower value. For example on 0.5A the resistor would be approximately 0.6V / 0.5A = 1.2 Ohm. So I wonder if that resistor is tweaked too... If that resistor is original and you have problems at 0.5A, I would think that Q1 is not original. Especially when R38 is 10K, this is too high to drive a single power transistor.

So I would definitely search for a good replacement darlington. This is not too critical. But also check the resistor values in the circuit.

Regards, Roland
 
That power supply needs a darlington to reach the specified 1A on the +15V. But since the +15 was used for the RS232 output and the pull up resistors on the bus load, that 1A was probably never reached.
True. Even at half an amp it seems marginal without a Darlington.
Can you check R33? Is that 0.39 Ohm? This resistor is the current limiting resistor. When the voltage across this resistor is about >0.6V, Q2 starts to conduct and limits the output current. It would start to limit at 0.6V / 0.39Ohm = 1.5 Amp.
For debugging, I unsoldered the collector of Q2 to disable the current limit function. The first thought when I noticed the oscillation was that the current limit circuit was causing the problem, but it turned out not to be the case. I did check R33 and confirmed that it's close to 0.39 ohm. It was difficult to measure accurately because the value is so small, but the meter read 0.4 or 0.5 ohms
If that resistor is modified to a higher value the current is limited to a lower value. For example on 0.5A the resistor would be approximately 0.6V / 0.5A = 1.2 Ohm. So I wonder if that resistor is tweaked too... If that resistor is original and you have problems at 0.5A, I would think that Q1 is not original.
R33 looks like it's the original part, but note that I also thought Q1 was original too. But since current limiting was disabled, the R33 & Q2 combination wasn't a factor, for the moment.
Especially when R38 is 10K, this is too high to drive a single power transistor.
Definitely. R38 supplies only about 3mA so Q1 needs to have huge gain to source 0.5A or 1A.
So I would definitely search for a good replacement darlington. This is not too critical. But also check the resistor values in the circuit.
I wanted it to be as authentic as possible, so I found some MJ2500 transistors on ebay.
Based on the comments above I started looking for other PNP Darlingtons in TO-204 and variations (aka TO-3 to some of us). @intabits has a TIP645 in one supply which got me looking for other substitutes. I bookmarked them for the future, but since the MJ2500's are on order I'll just go that route for now.

Regarding other substitutions in the circuit, there are some:
1. R56 was changed from 10K to be 680 ohms in parallel with 2200pF. I'm planning to change it back to 10K without the capacitor when the MJ2500 is installed.
2. C6 is missing. It's a 0.1uF on the +5V output. The +5V output was tested up to 16.4A using the electronic load (constant R load). Steady state seems good but I plan to install a 0.1uF in that location before putting everything back together to minimize the supply switching noise on the +5V rail.
3. C18 in the +15V circuit is 1000pF but the schematic says 680pF. I was planning to leave that one alone, so go with the 1000pF that's there now. It's not far from the value in the schematic.

I was planning to take a schematic and highlighter pen and compare every part on the board to confirm that the value is as expected.
Kind of a newbie question: is there a source that describes the ECO's to the H740? (A similar thing would be useful for the Omnibus boards, too.)
 
> R56 was changed from 10K to be 680 ohms in parallel with 2200pF
That modification sounds to me that they wanted to prevent some HF oscillation... The same is for a higher value of C18... The question is if this was necessary in the original configuration or with the modified power transistor.

> is there a source that describes the ECO's to the H740?
It is a good idea to check for these indeed. But I'm not sure all revisions and modifications are in the online documents. I have H720 schematics which don't seem to be online. So there is a bit of work for me to do too...
 
Last edited:
Back
Top