VOGONS


First post, by stepleton

User metadata
Rank Newbie
Rank
Newbie

Greetings, it's nice to join you all.

A new-to-me 5170 worked for about 24 hours before the PSU failed. (I believe it's the PSU because all the drives work in other machines and the mainboard shows no sign of shorted voltage rails --- a ATX to AT cable adapter arriving soon will hopefully confirm my suspicion.) The symptoms are that the PSU makes power for a fraction of a second and then shuts down. The hard drive (still the original Type 20) will click and the fan inside the PSU will twitch. Initial suspects include broken regulation leading to overvoltage and shutdown, or broken monitoring triggering shutdown prematurely.

I'd like to troubleshoot the PSU, but for that a schematic would be helpful. It looks like there were several PSUs for the 5170, and the one whose schematic is easily found online is a different model to the one I have. This one is an Astec AA13240. Does anyone know if it's possible to find a schematic for this unit?

5170 PSUs are said require a lot of load to work properly (including seven amps on the +5V rail!), but there seem to be multiple PSU implementations, and I would be surprised if they all needed the same minimum load. Is this high requirement universal to all 5170 PSUs?

Thanks for any help!

Reply 1 of 9, by Horun

User metadata
Rank l33t++
Rank
l33t++

I cannot think of any older AT PSU that "requires" 7 amps on the +5v. Can you point to a reference where it says that ?
Most very old ones require between 1 and 3 amps minimum (sometimes stated on the PSU itself, have a few that do).
As for schematic probably never find one. Nearly all AT PSU are designed the same just use diff caps, coils, etc....
As a rule I always connect a dummy HD (old scsi with bad sectors are good for this) when testing older AT boards with older AT PSU to make sure the load is ok.
Your 5170 will require -5v IIRC which most ATX to AT adapters do not supply, unless the ATX does.

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 2 of 9, by stepleton

User metadata
Rank Newbie
Rank
Newbie

I cannot think of any older AT PSU that "requires" 7 amps on the +5v.

The time when you could say this with confidence is nearly at an end! 😀

Can you point to a reference where it says that ?

I direct you to the table at the top of page 3-4 of the IBM PC AT 5170 Technical Reference.

Your 5170 will require -5v IIRC which most ATX to AT adapters do not supply, unless the ATX does.

You're right that the adapter I ordered won't supply -5V, but I'm hoping it's not really necessary. This person with a really authoritative-looking user icon says you don't need it. But if it is necessary, I'll see if I can kludge in -5V somehow. I'll report back tomorrow evening when I know for certain!

Reply 3 of 9, by the3dfxdude

User metadata
Rank Oldbie
Rank
Oldbie

I've never seen a motherboard critically utilizing the -5V. It's only for a few ISA cards that exist that use it. So for the purposes of diagnosing, it is very unlikely it will be needed here.

Did you check for power good? Did you check the voltages from the supply? Shorts are possible, but so is an aged power supply. If you reduced down to a minimum of motherboard (and loaded the 12V line with a dummy load) and the problem persists, then you are looking at swapping the motherboard or power supply to isolate the problem further.

Reply 4 of 9, by stepleton

User metadata
Rank Newbie
Rank
Newbie

I have measured it and the power supply is producing: no volts. I did not check the power good line since power bad actually 😀

What happens is the PSU flickers briefly into life before shutting down, so you get a little burp of voltage for less than a second before it's 0V everywhere. What I ought to do is set up the oscilloscope and see what this flicker looks like. If the +5V line goes substantially over +5V, then this suggests that the PSU has lost the ability to regulate and is just opening the taps all the way --- then the voltage monitoring in the PSU shuts things down. Or, if the +5V never gets close to +5V, then the voltage monitoring might be tripping prematurely.

(Apply same reasoning to other rails.)

With the equipment I can be bothered to set up right now, I can't load the PSU properly unless it's plugged into the 5170. There's a chance there's some kind of mischief happening on the mainboard, but I don't think so --- I'm not measuring a short across any of the rails. The adapter cable tomorrow will let me try to run the mainboard off of an ATX PSU, which will hopefully rule it out as a factor. Only the 5170 PSU will be left.

(I haven't done the oscilloscope test I mentioned above yet because the PSU is currently cracked open on my bench.)

Reply 5 of 9, by stepleton

User metadata
Rank Newbie
Rank
Newbie

Update: the 5170 boots and runs just fine without -5V from the power supply. A cheap 20-pin-ATX to AT adapter does the trick for me.

I suspect it won't be possible to get schematics for the PSU. I'll try to follow various online guides like this one, which is older but perhaps more contemporary to the kind of supply I'm dealing with here.

Also, although shot-in-the-dark recapping isn't cool anymore in the retro scene, I've still seen or received several suggestions just to replace the bulk capacitors on the output rails: if these aren't working correctly, I'm informed that they could cause voltage irregularities that trip the overvoltage protection apparatus. We'll see if I have replacements to hand or if I have to make an order.

Reply 6 of 9, by Horun

User metadata
Rank l33t++
Rank
l33t++

That is good news, now back to this:

stepleton wrote on 2022-08-31, 17:28:
The time when you could say this with confidence is nearly at an end! :-) […]
Show full quote

I cannot think of any older AT PSU that "requires" 7 amps on the +5v.

The time when you could say this with confidence is nearly at an end! 😀

Can you point to a reference where it says that ?

I direct you to the table at the top of page 3-4 of the IBM PC AT 5170 Technical Reference.

Interesting as a 5ohm Dummy load on 5v is 1A, on 12v is 2.4A. So how does that equal a 7 amp load.... inquiring minds are curious and goes back to my thought that a 1-3A load on those is sufficient.
Maybe they meant: If connected to motherboard w/o a harddrive ??

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 7 of 9, by stepleton

User metadata
Rank Newbie
Rank
Newbie
Horun wrote on 2022-09-02, 01:39:

Interesting as a 5ohm Dummy load on 5v is 1A, on 12v is 2.4A. So how does that equal a 7 amp load.... inquiring minds are curious and goes back to my thought that a 1-3A load on those is sufficient.
Maybe they meant: If connected to motherboard w/o a harddrive ??

Yes, I'm pretty sure this is what they meant. You can see a picture of the IBM-supplied dummy load here: http://minuszerodegrees.net/5170/misc/5170_load_resistor.htm

Unfortunately, the recap did not fix my PSU. I'll have to be more clever than that if I'm going to fix this thing.

Reply 8 of 9, by Horun

User metadata
Rank l33t++
Rank
l33t++

Thanks, and sorry to hear it still does not work.
On my other comment that means that is one poorly designed PSU even for the time. To require 7 amps load (7x5 is 35w but 7x 12 is 84w, so split it for say 60 watts to be stable ??) on a ~180 watt PSU is very poor even in very early 1990 standards...
Just saying maybe better to replace the guts with a better AT PSU board than try to fix a poorly designed one. Did a transplant on a Turbo XT PSU that worked out well.

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 9 of 9, by mkarcher

User metadata
Rank l33t
Rank
l33t
Horun wrote on 2022-09-03, 03:23:

On my other comment that means that is one poorly designed PSU even for the time. To require 7 amps load (7x5 is 35w but 7x 12 is 84w, so split it for say 60 watts to be stable ??) on a ~180 watt PSU is very poor even in very early 1990 standards...

According to the technical reference, they require 7A @ 5V and 2.4A @ 12V, so that is 35W + 29W = 64W specified minimum load. That's close to your 60W figure, but you don't arrive there by distributing 7A over both rails.