Print

[warmech]

I've been speaking to folks across a few different communities about my Lisa 2's PSU issues and was recommended to post a thread here, so I'll try to condense the story so far into a concise post (apologies in advance for almost assuredly failing to be concise).

Several weeks ago I acquired a Lisa 2 with a dead PSU (and battery-bombed I/O board, but that's a story for another day) that I have been working on ever since. The PSU had a blown fuse and some cap leakage, so all electrolytic caps (and the fuse) have been replaced. With the I/O board removed and the remaining PCBs cleaned and repaired, I reseated the PSU in the Lisa and it came to life... briefly. The font power switch lit up and I could hear what I thought was the whine of the CRT, but am now pretty sure was just the transformer in the PSU. It had signs of life for several seconds prior to shutting off once again. Inspection of the PSU showed that the fuse had blown again and caps C18 and C21 had very puffed-out tops (the familiar stink of dead caps also hung in the air). Before proceeding with additional work, I consulted a few different forums and eventually built a load tester for the PSU (see attached images) and got the following results when testing voltages:

+5v rail: +8v
+12v rail: +18v
+33v rail: +53v
-5v rail: -5v
-12v rail: -18v
+5 standby: +5v

Aside from these results, the following represents the body of testing/work I have done around the PSU:

- Tested all Q-series transistors - Q1 was the only weird one which returned .462v when testing base to collector and .525v when testing base to emitter. I must admit that I'm not familiar enough to know if those are truly within spec for the type of transistor Q1 is, whereas all the rest of the 3904s test in the .6v range and look completely fine. Also, I went to check if Q1 was hot to the touch after powering off the PSU (it was still plugged in, just off) and got a slight tingle from the case - should the collector be energized like that even when the PSU is "off"? I also am feeling weird about Q2; is this the regulator that should be producing +12v? If so, around +18v is coming out of it even though it appeared to test normal.
- R29 is operating within spec - 0-500 Ohms when tested; adjustment in any way does nothing to the outputs.
- CRs 8, 9, and 11 appear to test good on my end, but I'd like to make sure I'm testing correctly and that my values represent correct/expected ones. CR8 showed .155 from both anodes to the central cathode pin and CR11 showed .122 from both anodes to the center pin. CR9 was .680, which should be well within spec. I have no idea how to correctly test CR10, as I don't really know what the heck it is nor can I find a datasheet for it.
- All other normal diodes have been tested (including the three diodes attached to R17 an R19) and their results are below - I don't think anything is bad based on what I found, but it's always nice to have a second set of eyes.
- I have attempted to test CR17 and CR21, but I'm not terribly sure how they work with relation to being tested. They aren't transistors I'm used to working with, so I'm not sure what I should be seeing. All pin combinations result in OL except for 3 to 2, which results in .690 on CR17 and 1.236 on CR21.
- In one of the attached images, the four transformer pins all have continuity, which is slightly odd to me as it means that all three pins boxed in red in the top-left also all have continuity. As for the lower left-hand box, those four pins also have continuity between Q1's base pin and the trace to its left. Is this a sign of a short within the transformer? With the transformer removed these continuities go away, so I know it's local to the transformer, at least.

As stepleton on the 68KMLA forums described the issue: "we know the switching power supply is switching, it just doesn't seem to be regulating". What in the world am I missing, lol?

Diode testing results:

CR1	.525/OL
CR7	.450/OL
CR8	.155 |>-o-<| .155
CR9	.685/OL
CR11	.122 |>-o-<| .122
CR12	.450/OL
CR13	.205/OL
CR14	.450/OL
CR15	.525/OL
CR16	.525/OL
CR18	.450/OL
CR19	.450/OL
CR20	.460/OL
CR22	.520/OL
CR23	.450/OL
R17/R19 Diodes	.480, .450, and .450/OL

Unfortunately, my knowledge of PSUs this complex is scant and I feel I've waded in to the deep end at this point. Any help forward would be extremely appreciated.

Edit: Some additional info... Got a response from a user on TinkerDifferent (rdmark):

Another tip that I read in a thread somewhere on Lisalist, is that the optocoupler on U3 may go bad with age and cause voltage issues. It has an LED inside the package with wears out with heavy use. And supposedly (this is 2nd hand information) it is in charge of decoupling the hot and cold side of the PSU and a failure mode is that the voltage on the cold site get ramped up beyond spec. I preemptively replaced the one in my PSU with one from this eBay seller.

While I'd rather not spend $30 on a relay/OEC, I will if I need to; is there some credence to this? The schematics don't otherwise look like it should cause power to ramp up on the cold side in the event of a failure, but I'm not exactly an expert on all this. Heck, I figured U3 was just the relay to power on the machine.

One last thought - R29. I replaced it just to see if I was crazy and got the same result - zero change in voltage whatsoever no matter how it's adjusted. It seems like the sense line (and the pot, by extension of that) is just not even factoring in at this point. I pulled Q2, Q3, and Q4 again just to check them one more time and they all test good. CR8, 10, and 11 aside, the only other thing I can't test is CR21 (one of the MPU-131s), but I'll have a NOS one to drop in sometime this week. Are the values for CR8 and CR11 acceptable as I've tested them above? Or are they out of spec? And as before, if anyone can fill me in on how the heck CR10 is supposed to be tested I'd be very grateful. It's like the shutdown circuit isn't even functional, so I'm curious if CR10 is just shot.

[stepleton]

Hello there over here.

Thanks for the measurements. I think we can focus the search a little bit more.

We have the theory that the PSU isn't regulating, which means the secondary side (the circuitry to the right of the transformer in the schematic) either isn't telling the primary side (the stuff to the left) to tap the brakes once the +12V rail gets to +12V... or maybe the secondary is phoning home but nobody is taking the call. Or maybe someone is taking the call but they aren't able to do anything about it.

The mechanism that does the regulation feedback is all in the bottom third of the schematic. One of the tricky things about switching power supplies like this is that you need to communicate this control information back to the primary side, but you don't want any electrical connection between secondary and primary. A connection like that is potentially a way for mains electricity to get into the electronics, and that's a no-no, so there has to be some other kind of bridge. In the Lisa PSU, it's the transformer T4: the information crosses the gap magnetically, just like power crosses the gap in the other direction in T3.

One thing you might do is start from T4 and imagine what might happen if components connected to it weren't working as they should. You might stumble across an idea that predicts the same symptoms we've been observing.

But CR7, CR8, CR11 --- these probably aren't involved. They have to do with converting the AC power coming out of T3 and rectifying it back into DC, which is then smoothed by all of the bulk capacitors downstream between the diode and the connector.


Lastly, you asked about CR9 and CR10. These parts are part of the crowbar circuit, which should detect overvoltage and shut down the power supply. We're pretty sure this isn't working for you, since your power supply is perfectly happy to pump out +18V on the +12V rail until your smoothing caps puke.

CR9 is a zener diode (note how the cathode bar is crooked). Zeners are special because when you reverse bias them (put the positive and negative voltages in the "wrong" direction, as they are here), they don't allow any current to pass until a precise voltage threshold is exceeded. (You can see why this would be helpful in an overvoltage circuit --- a zener is the fault detector!) When current does flow, it goes into the gate of CR10, a silicon controlled rectifier (or SCR for short). SCRs are special because they are like transistors, but once the valve has been opened, they start conducting like a diode (i.e. one way) and just ignore the gate input until there's no more current flowing through them.

So, if the +12V rail goes over a threshold, the SCR is supposed to short the +12V rail to ground and keep it that way until the power supply shuts itself off. Why would grounding the +12V rail shut things down? Because it cuts off power that ultimately gets fed into T4 through Q2, which is controlled by Q3. Hmm... maybe this is a clue...

[warmech]

Heya again!

So I've done a bit of reading about how the crowbar works and I definitely understand it much better now. Thank you for explaining it as you did - that's maybe the most concise explanation I've read yet and it makes the most sense (read: you got my slow-as-molasses brain to comprehend it finally). That having been said, since it's not shutting down the PSU by way of shorting +12v to ground (and we know CR9 is within spec)...

Q2, 3, and 4 all seem to test okay. so I'm assuming they can be ruled out as culprits. None of the tantalums are gone either, seeing as they haven't exploded, and the electrolytics are all new so they're probably able to be ruled out as well. Resistors can be easily checked (which I will), but I would guess are unlikely. The only secondary-side active component is CR21 which tested kind of wonky but I have a replacement en route for (a few, actually). Then there's T4 itself. How do you even verify a transformer is good or bad? Probe the 3-turn and 6-turn sides and if continuity on each they're okay?

Primary side is where things get iffy for me. CR13 I am now putting two and two together about and realizing it's measuring way lower than should be expected for a diode. I pulled it again and it measures .180-.190-ish/OL. That's... not sounding good. Only issue is, I can't identify the diode. The case markings are 2N:-004 (or ":-004 2N") but 2N isn't a diode, right? That's a transistor indicator, I thought (and the symbol on paper is a diode, marked 40V/1A). Also, googling that combo has led nowhere and I'm kinda scratching my head on that. Aside from CR13, there is also CR17 (another MPU-131 like CR21, of which replacements are on the way) which I don't really know how to check. We know that Q1 has to be working to some extent, as it's switching power, and that it's neighbors CR24-26 are testing good as well (the diodes looped around R17 and R19). Its other neighbor, CR15, is also testing okay. Adjacent to the stream of components between Q1 and T4, CR16 and Q5-6 both test okay; they lead to the relay/opto-isolator, so I wouldn't think they'd be part of the issue (unless rdmark was right and that thing is the missing link, in which case I wouldn't know whether to laugh or cry).

Have I way overstepped or over-thought all that? Or could CR13 just be faulted and throwing a wrench in the whole process left of T4? If so, lol, how do I even identify what diode that is?

[sigma7]

Heya again!

It looks like CR9 tested ok as a rectifier diode, but not the Zener characteristic.

If it seems safe to operate the PSU briefly with the dummy load, then measuring the voltages to ground from each end of CR9, and the +12 output, may be enough to reveal which component (of the crowbar circuit) has failed.

If you can't operate the PSU, then measure the zener voltage of CR9 (either out of circuit or with one lead out of circuit).

The A version of the Apple schematic says it is 1N5292B, which is apparently a current regulator diode, however later versions of the schematic show it as a 12V Zener without any part number, so my guess is that production models used a zener instead of 1N5292.

It seems to me that "12V" is too low for that zener, as it is often possible to adjust the PSU so +12 is quite a bit higher when there is no load. I don't know off-hand what voltage/part it is ... anyone have the number?
U3 is a solid state relay that switches the PSU on/off under software control via the I/O Board COP; it doesn't have anything to do with the voltage regulation. Solid state relays usually have internal opto-isolators so that may be where the confusion about what it is came from, but "solid state relay" is what you'd be looking for if you need a replacement, not an opto-isolator.

If your dummy load has a switch to turn on/off the PSU via pin X on the card edge, and that switch works, then U3 is working.
When measuring a transformer winding, it is normal for it to have a low resistance; that doesn't mean it is shorted (still could be, but not a common failure).
"Schottky" diodes have a low forward bias voltage, 0.2V is in the right ballpark for these diodes. On modern schematics these have a square hook on each end of the bar, but the missing hook on the Apple schematic isn't conclusive.

[warmech]

I have no idea why I didn't think to test that - it's +18v on either side of CR9. Is that to say that the zener aspect of CR9 has failed? (I apologize if this is a stupid question - I genuinely don't know )

I checked the number on the diode and it's a Motorola part: 1M12ZS2 - I assume this is just a generic 12V zener diode? I have no idea about any of its other attributes, like current rating, tolerance, etc. as, once again, no datasheet and google comes up with the slimmest of pickings to provide no info whatsoever.


Good to know regarding U3, Schottky diodes, and the transformer - that's a load off my mind. It (U3) is presently working, so that's good to know.

[stepleton]

For what it's worth, I don't think we know if the Zener has failed --- I think sigma7@ was saying that we just haven't tested whether it's still Zenering. For example, we haven't tried to reverse bias it with a voltage and see if it conducts if the voltage exceeds a threshold. The easiest way to do that would be to take it out of circuit, which you could do by lifting a leg as mentioned. Bear in mind that you wouldn't just want to apply a voltage directly to both sides of the diode: you want a resistor to limit the current in case the zener does start to conduct.

The fact that in-circuit it's +18V on either side could mean that it's working fine, but that the SCR next door isn't responding to its gate.

[sigma7]

it's +18v on either side of CR9. Is that to say that the zener aspect of CR9 has failed?

I checked the number on the diode and it's a Motorola part: 1M12ZS2 - I assume this is just a generic 12V zener diode?

Since the voltage is +18 on both sides, that means the voltage across the diode itself is ~0V, which means it is either a dead short or there is only leakage current through it. Since you measured it as looking like a diode, then it isn't shorted, so no current.

Since there is +18 at the cathode, there is a circuit to the +12 rail.

Since it is +18 at the anode, it seems there is no circuit to ground on that side. That could be due to R13 being open. However, +18 at the anode should trigger the SCR, applying the crowbar. Since there is no crowbar effect, that's a second clue (or problem), so both could be due to no circuit from CR9 to both (R13 and CR10). Another possibility is that CR9 is connected to both R13 and CR10, but there is no connection to ground for R13 and CR10.

I suppose another failure mode could be that CR10 is shorted from gate to anode, cathode open, in which case R13 would burn up. So check for a short from the gate of CR10 to +12 rail if R13 is open.

The leads on CR9 look corroded, so check the diode lead is truly connected to the circuit board. ie. not in a hole with solder around it that doesn't actually make contact. This isn't normally a problem with boards that used to work, but if you've removed and re-installed the corroded component, the new solder joint might not be good.

Once you get the crowbar working, you may want to disable it to figure out why the regulation isn't working, but check the capacitors etc. will tolerate the high output voltages.

Come to think of it, since the regulation isn't working, the PSU may continue to drive all the current through CR10 rather than going click click click, causing CR10 to fail. So make very brief tests and check the temperature of CR10 (don't burn your fingers!) I think the fuse may save you from this condition, but not certain.


I think 1M12ZS2 is probably a 12V zener. Closest datasheet I could find easily is in 1988 Motorola Rectifiers and Zener Diodes Databook (Thanks Al!) https://archive.org/details/bitsavers_motoroladaRectifiersandZenerDiodesData_33936696/page/n393/mode/1up

FWIW, 1N5292 datasheet at https://archive.org/details/bitsavers_motoroladaRectifiersandZenerDiodesData_33936696/page/n437/mode/1up

edit: found 1M12Z5 at
https://archive.org/details/1965MotorolaSemiconductorDataManual/page/n58/mode/1up
Where
1 = 1Watt
 M = Motorola
   12Z = 12V Zener
        5 = 5%

No indication about S2, but does mention "series matched sets" with better tolerances were available, so perhaps that's 2%.

[warmech]

Excellent news - the crowbar is now working! It was this line right here:

there is no connection to ground for R13 and CR10

that finally clicked in my head. Sure enough, testing for continuity showed zero connection between that whole trace along the edge of the PCB and ground. The issue ended up being a bad trace coming off a capacitor which has since been fixed and the PSU is now (un)happily chirping away as expected due to an overvoltage issue. Now, the secondary ground connection coming off of R29 was serviced by that trace so it's connected correctly again, but there is no change in voltage when attempting to dial it in either way. It was too hot to stay in my shop any longer, so I'll test more after work is over and the A/C has some time to cool things off out there.

Not out of the woods, but certainly seeing daylight through the canopy now. Thank y'all both so much - we've got to be in the homestretch now.

Edit: I'd like to state for the record that I feel like a complete idiot for missing this, lol.

[sigma7]

... the PSU is now (un)happily chirping away as expected due to an overvoltage issue.

Great progress!

Now, the secondary ground connection coming off of R29 was serviced by that trace so it's connected correctly again, but there is no change in voltage when attempting to dial it in either way.

... confused by that... R29 (the voltage adjust trimpot) doesn't connect to ground, does it? Resistor on one end, transistors on the other end and the wiper, no?

[warmech]

My bad - that was to say that the secondary ground coming off of C33 (that is tied to R29 by way of R27 - and the +5v sense circuit) was connected to that. Fingers typed faster than my thoughts could keep up with.

[sigma7]

While reviewing the May 1983 Hardware Reference Manual for another PSU thread, I see it refers to U3 as an "opto-isolator" so I guess YMMV and tomayto / potaahto.

[warmech]

Well, update for y'all. I took the PSU to a friend's place who is infinitely better at this than I am (this is his post-retirement day job, so he knows what the heck he's doing) and let him take a look at it. We went over as much of the PSU as he had the equipment to test and basically ruled out every component but the transformer. Nothing shorted, nothing blown - everything checked out. However, while testing, it appears that CR10 failed, as the crowbar has reverted to a failed state and our subsequent testing confirmed as much. All that having been said, I think this PSU is a lost cause at this point. I have another lead on a replacement (working) PSU, so that may be the path forward from here.

Adjacent to this, however, I've been trying to test the CRT to make sure no damage occurred when the system took a hit during the brief overvoltage it took when the PSU was fired up the first time. I have a bench supply putting out +33v and +12v and have wired up a fingerboard running both to the video board. Additionally, I've wired the fingerboard to the old PSU's focus/contrast controls, so the CRT is wired up with all it should need. All that considered, I get no life out of the CRT - no neck glow, no high-pitched whine you normally get out of CRTs, nothing. I tested a couple of points on the video board while it was mounted, and it's getting just under +12v/+5 (11.8-ish/4.9-ish) and also very slightly under +28v out of the regulator, so I know it has power and that the regulator wasn't fried from overvoltage. I'll pull the board to start testing components tomorrow, but does the CRT require anything else to power up? I have the faceplate on the front to engage the interlock, but does the motherboard/CPU board have to be connected for it to power up, or am I missing something else? The hardware manual doesn't mention anything specific in regards to requiring the CPU to turn the dang thing on, but I do know a good part of the video control is on the CPU board.

Edit: After craning my neck a little further over, there is in fact neck glow and I just wasn't looking well enough. Still no high voltage, though. I'll test transistors in the morning and see what things look like.

[sigma7]

... test the CRT ... I have a bench supply putting out +33v and +12v and have wired up a fingerboard running both to the video board. Additionally, I've wired the fingerboard to the old PSU's focus/contrast controls, so the CRT is wired up with all it should need. All that considered, I get no life out of the CRT - no neck glow, no high-pitched whine you normally get out of CRTs, nothing. I tested a couple of points on the video board while it was mounted, and it's getting just under +12v/+5 (11.8-ish/4.9-ish) and also very slightly under +28v out of the regulator, so I know it has power and that the regulator wasn't fried from overvoltage. I'll pull the board to start testing components tomorrow, but does the CRT require anything else to power up? I have the faceplate on the front to engage the interlock, but does the motherboard/CPU board have to be connected for it to power up, or am I missing something else?
...
Edit: After craning my neck a little further over, there is in fact neck glow

Without any video signals, I think all you'd get is the dim glow of the filament.

The horizontal sync is what drives the flyback transformer and without that, there will be no whine or HV.

All of the video signals come from the CPU board (the video data come from the slot RAM, then through a shift register on the CPU board to make the video bit stream).

The front panel interlock switch disables the PSU like the back panel interlock (so not a concern if you are supplying power otherwise).

Since the PSU was generating all the voltages (albeit too high due to apparently no feedback), I don't think the transformer is a likely suspect, and not surprising that CR10 didn't like sinking all the energy. I've been pondering how to troubleshoot this particular design and will let you know if I come up with a potential procedure.

[patrick]

(...) and basically ruled out every component but the transformer.

In your first posting you write that you get voltages on the secondary side. Your voltages are 50% too high, but the ratio of the 5V, +/-12 V and 33 V is correct. -5V does not count, this is regulated by a separate 7905. That means primary drive circuit is working, transformer is working, secondary rectifiers are working. But your regulation is not working.

At first: when operating the PSU outside the Lisa, make sure pin N (+5 V sense) is connected to pin 11, 12, M (+5 V output)! Otherwise your controller will measure 0V and continue outputting power until the crowbar triggers.

You already replaced R29, so I assume it is working. This is a common failure, the wiper can have intermittend contact. The Q3 will never conduct, the controller circuit thinks the output voltage is much too low, same as above. Rotate R29's wiper to the Q4 side, then the controller gets the biggest actual value available, which means it should set the lowest output voltage. Target value is 0,7 V at the base of Q3 when +5 V and -5 V have equal magnitude.

If the thing still bangs into the crowbar: disconnect mains, apply +5 V from a lab PSU to the +5 V rail and -12 V to the-12 V rail. Do not apply -5 V to the 7905 output, these don't like to be reverse-powered. Put an 1k resistor in series with the +12 V rail and apply 12 V. The resistor is necessary to limit the current through CR21 when it fires.

Now start with R29 away the Q4 side. Rotate it slowly and see what happens with Q3 and Q2. The voltage at the base of Q3 should rise until it reaches 0.7 V. Then Q3 and Q2 should conduct and the PUT triggers. That means the voltage at pin 3 goes from 12 V to almost zero.

If this works, connect another 1k resistor from pin 3 of CR17 to 12 V line. It might be necessary to disconnect pin 3 from the PCB, so that only the bias voltage is connected. Now repeat the procedure above and check if CR17 triggers, too. Your fault is likely to be somwhere in this area.


The only PUT you can buy today is the 2N6027 from OnSemi. This works as a replacement for the obsolete MPU-131.

[sigma7]

At first: ...

That's a most excellent & helpful troubleshooting process, thanks Patrick!

Perhaps you can tell us half-voltage folks... when this PSU was configured for the European market (ie. 230V), how was the compatibility of the standby transformer managed:

The schematic shows a jumper, but no notes for its use... was a different transformer part number used, or something (resistor, diode, inductor, capacitor?) used in place of the wire jumper that we have?