• Please review our updated Terms and Rules here

H724 woes, go figure!

TX_Dj

Experienced Member
Joined
Apr 5, 2013
Messages
229
Location
Sachse, TX
Heya fellas, been a while since I've been in here, but I recall Jack Rubin suggesting this is a great place to come ask questions when I reach my wits end.

Been doing some light cleanup work on my already fairly spotless PDP-8/e, and going thru my initial inventory and checking things out like the power supplies.

This is one of the first pics of my unit *minus* the degraded case cover foam that had begun to rain down by the time I got the rack out the VCFMW last year:
bLYO2FN.jpg


(Well, hopefully that will resize to something on your screen, the preview is forcing it out to full size which is even bigger than my 4k screen)

Anyhow - in this expansion unit, which is where all the core for this system is located, the H724 will not start properly.

The relay energizes, the fans come on, I have 30v on the input to the -15v stage and 15v on the input to the +5v stage, and I'm pretty sure that the +8 and +15 are getting input power too.

SXPDETN.png


The maintenance manual seems to indicate I should be able to read 0/30/30v off the heat sinks on the fan closer to the big transformer, but instead I see 30/30/0, the opposite arrangement. I suppose that makes sense, since the one nearest the transformer has the diodes on it, not transistors.
p5MIanQ.jpg


The four caps at the far end, bus bar'd and jumpered together and marked on the manual's diagram are the input filters for the +5v stage, I've determined that much at least - they are the ones with power there, and the one that says +30V also has it, that's (maybe?) one of them on the -15v circuit's input, however the two big caps closer to the 30v heat sinks and transformer also have 30v on them and the schematic calls for two caps on the input to the -15v section.

No matter what though - I have zero output from any of the regulated sections - the +5, +15, and -15 caps all have no voltage. There appear also to be nothing going on for any of the transistors, either the three that bolt off to the side, or the ones hanging out next to the fans.

All the fuses are good, and like the rest of my system, the fuse holders look as new as the day they came off the line, no corrosion or anything in there, so I'm not suspecting that one of them is a potential loss of power signal to POWER_OK ckt.

I am suspecting the problem is with the POWER_OK circuit because it just simply won't start.and all voltages are non-present from the get-go, like it just refuses to try to start.

Any H724 gurus seen this before or am I making a very incorrect assumption that this thing should be able to regulate without any loads attached? If loads need to be attached, which rail(s) need loads?

I tried putting some small loads on the +5 and -15, since they are the big daddies of the bunch, but still got nothing.

Again, the relay kicks in, the fans come alive, and I have fine voltages out the secondary of the xfmr, so given all I've said now I expect this issue to exist in the A1/A2 card set, but without a riser for those it's really tricky for me to see what they're doing while installed! :)

Thanks for the advice.
 
I think some of my questions, such as "shouldn't it give me voltage even w/o load" will be answered soon, it's been well over a year and a half since I was last poking an 8e, and I don't recall if I checked its voltages with no cards on the bus. I'm moving on to the CPU chassis today, cleaning out the little bit of dust in there and getting rid of the rest of the foam fuzz and making it pretty again, and will be repeating this experiment with its H724, but figured I'd post about this one before I get to that.

Regardless of what happens with the CPU (which is almost full) having a busted PSU in the expansion negates any operation since all core is down there, not in the CPU. Sure, I can shuffle out some cards if the CPU box works, but what fun is that. :)
 
I had the +5 regulator on my 8/e die. The LM723 on the A1 board was the problem.

I soldered lots of wires onto the A1 board so I could see the interesting signals. Pictures starting 1/9/15 here: http://www.ricomputermuseum.org/Home/equipment/dec-pdp-8e/pdp-8e-restoration-log

The 8/e Maintenance Manual Volume 1 describes the operation of the power supply. The +15 supply depends on the -15 supply to work, and the +5 supply depends on the +15 supply to work, and the -15 supply depends on the +5 supply to work.
Pick one supply to start debugging with. I would probably pick the -15V. You could solder wires onto pins 3,4,5,6, and 7 of the VR1 to see what is going on at power up. There should be plenty of help here to get the supply working.
 
So here's something that I had not discovered in my work yesterday... even though I know I checked all fuses prior to testing (all good) and I was certain I checked all fuses again after testing (and I swear they were all good), I find the +15V/1.5A fuse is blown in the unit I was looking at yesterday.

I found this after going to test the supply in the CPU chassis today, and it blew the +15V fuse.

I have to wonder if I'm making a poor assumption in the load I'm putting on the supply. All I want to do is get the PSU started and monitor voltage stability and ripple.

I figured it would be a safe assumption to not load any rail less than 10% rated power, and not go over 50% on any rail.

So I have actually landed that between 15-30% for the regulated rails:

50 Ohms on +15VDC for a 0.3A / 4.5W load (30% load)
10 Ohms on -15VDC for a 1.5A / 22.5W load (18.75% load)
1.5 Ohms on +5VDC for a 3.3A / 16.6W load (16.65% load)

The resistors are appropriately sized for the power dissipation.

Now, maybe this H724 actually has a different condition than the one yesterday - but when I pulled the +15 fuse from today's unit and found it blown (after it tested good prior to firing up!) this caused me to re-test the +15 fuse from the unit yesterday, which I found blown.

I have to wonder if these poor neglected caps have just sat too long, and without reforming or replacement, I'm just going to crater the fuse when the PSU tries to charge the cap...
 
I always start by reforming the caps.

The +15V output feeds the A1 regulator board, so a shorted C14, VR5, or Q10 could load the +15V.

You could take a look at the base, emitter, and collector voltages on Q100 to see what it is doing.
 
So I suppose I can take this to mean that whether I had the unit unloaded, insufficiently loaded, or loaded with a weird offset towards one of the regulated rails, I should still not expect this specific thing to happen.

I think that's the conclusion I've come to as well, I just only have three H724's and one of them is in a system I haven't touched in ages, so I can't remember how it behaves.

I had thought about, since I have two sets of CPU boards and 4KW core maybe I should just use one minimalist setup as a test board set to confirm or eliminate what I said in the first sentence of this update... however, I also know the H724 is supposed to be "smart" enough to prevent itself from trashing the boards in the system... I'm just too used to linear supplies that like to keep on trucking thru a fault, letting all the smoke out, and that distrust of my S100-type machines is what is likely fueling my fear with this setup.... if that makes sense?

I realize the H724 should shutdown if it over/under volts, or goes over current on the draw - it's fused and intelligent, as opposed to the dumb linear and per-card regulator setup of S100 systems that I tend not to trust until proven reliable.

Am I just being unnecessarily cautious in this context and causing myself an issue... I think that's the actual question I'm trying to ask. :D
 
Update...

All appears well with the first one.

I will always be quick to admit I'm no expert. But I don't know why that also makes me a stubborn student.

I have learned my lesson, I'll reform the old big capacitors in these machines before proceeding. Had never had a problem with one until now.

After reforming all the output caps (the only ones not getting voltage in my original testing), I now have steady power on all rails, and a solid power OK signal.

Thanks for the info!
 
Back
Top