Print

[stepleton]

Hi all,

The Datapower 1.8 PSU from my 2/10 is broken! It's shutting down just after power-on, presumably with some kind of voltage protection circuit engaging.

SMPS aficionados can probably diagnose and fix the problem in a flash, but I am a weekend warrior here. I do like James's reverse-engineered schematics on Ray's site --- they seem very cleanly laid out to me, with lots of helpful annotations. I feel like even I ought to be able to figure it out, and the fact that so much control circuitry is on a separate plug-in module suggests that I can remove the module and do a good amount of troubleshooting without poking my fingers around rectified mains current, a real bonus!

...But of course I've got plenty of projects already. Does this failure sound familiar to anyone? Maybe there's a component that's known to cause this kind of problem that everyone just replaces automatically. If anyone doesn't mind me looking at their homework, I'd love to see what happens if I just copy your answer 

[rayarachelian]

Mine did this. Did it it buzzes or ticks while you pushed the power button in, and power light is on until you release? if so, yeah, over voltage crow bar is what's happening.

Check for shorts if it works in another Lisa, if not just replace the big capacitors including the RIFA ones (that's all I did to fix mine.)

[stepleton]

This PSU was recapped in 2017 and had been working fine until a few weeks ago. I think this is a different problem. I've got a (very buzzy) spare 1.2A PSU working as a backup in the 2/10 at the moment (I'm not running the Widget --- there's a much more efficient hard drive in there now ), so the problem seems to be isolated to PSU.

More applause for the schematics --- I'm finding them a lot easier to read than the Apple ones. I'm trying to think my way through a plan of attack now.

[compu_85]

When my Datapower was doing this (the condition I got my 2/10 in) the dual diode had shorted out. It's one of the TO3 package devices with a big heatsink on it, I believe CR20 on the schematic.

Here's a snippet from the old list,

very very very likely that one of the secondary diodes has a shortcut. So before you solder out all the caps, first try to make some measurements instead of a try-and-error-repair :-)
There's only one big dual-diode (looks like a transistor in the metal-housing) which is probably the bad one if you describe that clicking-sound. It's overcurrent shut-down, because that diode has a shortcut. Try to measure with an ohm-meter one of both contacs at the soldering side with the housing of that diode. There must be one good and one with less than 0,5 ohm. Else try all the diodes vertically mounted, they could be bad too

I've found the traces on the DataPower supplies are very easy to damage - moreso than the Apple supply. So work carefully!

You can see the failed dual diode from my machine here: http://compu85.homeip.net/stuff/Lisa/Fixed!/DSC02354.JPG My friend was helping me do the repairs, the dead part is by the base of the desk lamp.

-J

[stepleton]

Your post made me very hopeful about a possible fix! Unfortunately, no luck for me this time --- CR20 measures okay, and so do other secondary diodes. I spent part of the evening with a much more electronics-savvy friend, and none of the devices roused too many suspicions in him except for C19, which was hard for my meter to get a good reading from (50uF with polarity reversed, no good measurement with leads in the proper orientation.

He's also jokingly suspicious of SCRs like CR1, but mainly he doesn't know what "good" looks like for those off the top of his head, and also the switching controller Z2 could be malfunctioning (fortunately not so uncommon a part). More obvious alternative solutions were not obvious yet, in other words.

I'm not sure where you would find a replacement for CR20 if that were the problem: my searches for it were not so successful, and it seems to me that you might need to engineer an alternative.

So, the investigation will continue...

(ETA: Electronics-savvy friend was very patiently watching over video-conference and telling me what to measure )

[compu_85]

The replacement for CR20 in my system is a more modern 3 legged package. Finding a heatsink was tricky, I didn't want one too heavy which would put stress on the leads.

[stepleton]

Still slowwwwly working on this. I thought I'd make a dummy load first to support the investigation, and also for inevitable future repairs. Behold: attached to a working PSU, the Lisa 0.

(I guess I have to get around to fixing the thing now.)

[stepleton]

My power supply is repaired! Thanks to James for reverse-engineering the schematics and Ray for hosting them.

Consulting with a friend, we tested things on the PSU whilst it was powered down and basically ruled out most of the passive parts on the board. All of the diodes appeared to test OK, too (or at least didn't lead us to suspect they were bad). This focused our attention on the various active components that were left over.

I didn't feel confident about troubleshooting a live power supply with my experience and equipment, so, I basically shotgun-replaced all of the comparators, many of the commodity transistors (all of the 2N4404s), the SG3524 switching controller, and the crowbar SCR. Happily, this seems to have worked. I saved everything I replaced, and this afternoon I ran some tests on those parts. Everything that's easy to test appeared to be just fine, which leaves only the SG3524. This is a complicated IC (relatively speaking) and seems like the kind of thing that would take a fair amount of preparation to examine, so for now I think I'll just set it aside. I did hook up Vcc and ground and checked the reference voltage output, which seems fine, and the chip doesn't get suspiciously hot, either. If anyone has any ideas of a test they'd like me to try, I'd give it a shot in my spare time if i have the parts.

I can't say my other tests were comprehensive, but for now I suspect the SG3524. It's nice to have the old PSU back; let's hope no part fails that's harder to replace (the SG3524 is still available for sale).

[rayarachelian]

Congrats on a successful repair!

[sigma7]

Old thread, but since it is the only one with "Datapower" in the subject...

Al Kossow located an official schematic for the DataPower control daughterboard (posted to bitsavers... thanks Al !); which is annotated as Rev. D.

The reverse engineered schematic of 2002-01-12 is annotated as derived from Rev. C.

The only differences I have found are some (surprisingly high) voltage specifications on capacitors, and the note that R15 (8K25 @ 1% on the reverse engineered schematic) is in fact not a specified value but factory selected (presumably to set the VR reference voltage with better than 1% accuracy).

If anyone finds other differences, please append here!