Macintosh SE/30 Black Screen

[ortek_service]

Originally Posted by Keith956

The thing these recapping sites never seem to mention is that it's a good idea not to change all the caps at the same time; do it one by one, testing after each change. It's much easier to locate a mistake that way!

Although maybe they are assuming that you are re-capping it because it already doesn't work - due to failure of many of these
- However, if that's due to physical leakage of substance from the capacitors, attacking the PCB, then just changing these (Mostly to Solid-Tantalums, which shouldn't have that issue at least) won't necessarily fix that.

[TonyDuell]

Nor do such sites mention that randomly replacing components without testing first to find what the real problem is rarely a good idea.

I would think that the +5V and ground rails were carried by internal copper planes in the PCB, not (mostly) by tracks on the surface. Of course it's possible there are short traces from vias to IC pins on the outside surfaces of the board for power and ground in some cases.


Anyway. Is it worth testing for continuity between the power connector ground and +5V pins to the ground and power pins of all the ICs?

[ortek_service]

Well it seems there's probably not just power-planes internal to the original PCB, as rather than a 4-layer (that copies have been made using), Apple original used a more-costly 6layer board, according to:

https://hackaday.com/2021/03/24/30-year-old-macintosh-se-30-gets-a-brand-new-logic-board/

So it could be there quite a bit of signal-tracking in the inner-layers.

Although if the changed capacitors were mostly all on power lines then hopefully there shouldn't be any logic signals lost if a via on one of these had got damaged. But maybe thru-plating on holes between layers is more likely when removing thru-hole components so loss of some SMD-pads shouldn't normally break any vias connecting to these (Although repair-attempts could, if trying to poke a wire down the via too forcefully).


Yes, several of us have said that continuity-checks on the connections to all the changed capacitors is recommended, And tis could also be extended to the power-connections on all the IC's - Although maybe it's easier to just check for correct voltage, when powered-up.

I'm not sure whether just +5V supplies would be adequate, or whether +12V / -12V & -5V supplies are also vital and loss of one of these could cause it not to boot.
They did measure OK on the power-input, so it doesn't look like there are any shorts / capacitors fitted backwards and breaking-down (that may trip the PSU, or cause some smoke from these!).

[Tribble]

Thanks both, very helpful.

Keith, I wish I knew that before I started!

[Keith956]

Originally Posted by Tribble

Thanks both, very helpful.

Keith, I wish I knew that before I started!

We've all done it, don't worry!

[Tribble]

Whilst repopulating the board I checked the 220uF capacitor on C21. It checked out fine but there is no continuity between C21 + and +5v rail. The +5v goes to the power and video connector pins 12 and 13. This seems to me to be the no screen issue.

Should I run a +5v bodge wire to C21 positive?

[TonyDuell]

I think you should try that.

[Tribble]

I made a bodge wire to C21. After finishing populating the board I rebuilt the machine.

Sadly no picture, so it's back more diagnostics.

[SiriusHardware]

Unlucky, to find a solid fault like that and yet after repairing it, still have a problem.

It is possible that you have lost the +5V rail in more than one direction, radiating away from that capacitor. With the power off, try metering for continuity between the +5V output of the PSU and the +5V pin of every IC which is meant to run on a +5V supply. Quite a long job, but it may throw up something.

[ortek_service]

Originally Posted by Tribble

Whilst repopulating the board I checked the 220uF capacitor on C21. It checked out fine but there is no continuity between C21 + and +5v rail. The +5v goes to the power and video connector pins 12 and 13. This seems to me to be the no screen issue.

Should I run a +5v bodge wire to C21 positive?

Originally Posted by Tribble

I made a bodge wire to C21. After finishing populating the board I rebuilt the machine.

Sadly no picture, so it's back more diagnostics.

I presume that was a typo and you meant C11 ? (Which is the only 220uF one)
- As C21 is a 0.1uF (Ceramic?) capacitor, and wasn't one of the ones that were changed.

I'd be surprised that those original axial Electrolytics had failed, so I wouldn't expect them to have been on the list of ones to change - which was primarily due to the original dodgy SM Electrolytics physically leaking.

If the thru-plated hole of the +ve of C11 has been damaged (and not somewhere else on the board) and was why there was no +5V supply to it then there could be breaks to other areas that the +5V should go to.

[Tribble]

Yes Ortec, it was a typo. I'm old. That's my excuse anyway.

Looking over the schematic as I'm at work and cannot see the board, it was definitely C11.

[SiriusHardware]

Can you now do as Owen (Ortek) and I suggested - check continuity between the output of the +5V supply and all other points on the PCB which should have +5V going to them.

The theory here is that the connection to a mid-layer track may have been damaged when one of the capacitors was removed. If so, there may have been connections radiating out in more than one direction from that point, so while your wire bodge has fixed one of the missing paths, there may be at least one more break still waiting to be repaired.

[TonyDuell]

Since the replaced capacitors are supply decouplers, one end of each is grounded. Therefore as well an damaging a supply connection you could also have damaged a ground connection.

I think as well as checking that the +5V rail goes to every place it should, you should do the same for ground.

[Keith956]

If it were me, I'd use the fact that the video sync signals were not present (and hence the display fault). Narrow the search to the video RAM and associated PALs, checking their supplies and determining whether the clocks in that part of the circuit are working. Still tedious but might reduce the searching.

[TonyDuell]

I suggested checking signals (and some useful ones to check) back in post #90 and was shouted down for it.

[SiriusHardware]

I suggested checking signals ... and was shouted down for it.

Not exactly:

What we have here is a straightforward cause and effect fault - the unit was originally essentially working except for a relatively trivial display fault (display offset, as far as I remember).

The unit was then recapped, after which it wasn't working at all. Common sense dictates that this was due to something which happened during the recapping process.

From there, there are two equally valid approaches:

One, go back over what was done and see if it is possible to identify what went wrong during the recapping process, either visually or by making simple continuity measurements. This I think is the better initial approach for a relative newcomer to the game.

Two, treat it like any other 'was working but has now failed' fault and examine power rail voltages / trace signals logically until the point where they disappear. This is the method likely to be preferred by people who have years of experience of troubleshooting electronics, including myself if I had the machine in front of me now.

It was Tribble himself who decided to go with approach #1, which did actually pay off in so far as it has already located one stretch of missing +5V rail. I think he should continue with this approach until it has at least been established that the power rails (including 0V, as Tony rightly pointed out) are fully intact. If that approach alone doesn't find and fix the remaining problems but does establish that the rails are intact and going everywhere they should, it will then be time to go back to the beginning with approach #2.

[TonyDuell]

Could somebody please explain to me how failed decoupling capacitors on the digital board and/or failed capacitors in the sound circuitry can produce a correct but offset display? In other words, given the orginal fault why were these capacitors replaced?

I agree that getting the supply and ground rails right is a good thing to do now. No IC can work without power and we now know that at least one track was damaged.

[ortek_service]

Hopefully, as this is apparently a 6-layer board, the power (as well as ground) will be mostly on large-area planes rather than relying on a single long track.
So failure of a thru-plated hole / via connection between layers will only take out the supply / ground very-local to that and not other areas on the board.

And worth checking to start-with that all the replaced capacitors of the 13 C1-C13 that should be connected between +5V and ground, have 5.0V across them when powered-up. As if the connections to these are all still intact, then there hopefully won't be too many other places that could have lost the +5V power.


C1 (47uF) = +5V Supply reservoir/decoupling on Power Connector Sheet (8) - just shown in block of decouplers
C2 (470µf) = +5V Supply reservoir/decoupling on supply to both Audio Amps
C7 (47uF) = +5V Supply reservoir/decoupling on Power Connector Sheet (8) - just shown in block of decouplers
C11 (220uF Axial Al-Electrolytic) = Main Reservoir for incoming +5V on J12 pins 12&13 supply input on Power Connector Sheet (8)
C12 (47uF) = +5V Supply reservoir/decoupling on Power Connector Sheet (8) - just shown in block of decouplers
C13 (47uF) = +5V Supply reservoir/decoupling on Power Connector Sheet (8) - just shown in block of decouplers

And maybe also check these on the other non +5V supplies, as lack of those could cause issues:
C8 (47uF) = Main Reservoir for incoming -5V on J12 pin6 supply input on Power Connector Sheet (8)
C9 (47uF) = Main Reservoir for incoming -12V on J12 pin7 supply input on Power Connector Sheet (8)
C10 (47uF) = Main Reservoir for incoming +12V on J12 pin14 supply input on Power Connector Sheet (8)


These ones won't actually have a supply voltage across them (but should probably normally measure a (< 5V) bias voltage across)
C3 (47uF) = UB10 (L.Ch) Audio Amp Output DC-Block
C4 (47uF) = UB11 (R.Ch) Audio Amp Output DC-Block
C5 (47uF) = UB10+11 Audio Amp's VRef pin(15) decoupling

[TonyDuell]

Looking at the photo in post #17 it would appear that many of the ICs are surface-mount, either the square PLCC packages or SOIC (like small DIL, but surface mounted). Connections to pins of those chips must be made by traces on the top side of the board. It's possible a power or ground trace from a via (so connected to a power or ground plane inside the PCB) to the actual IC pin has been damaged.

[ortek_service]

Yes, although if these tracks to the PLCC etc. IC's are not near the changed capacitors then hopefully they are not too-likely to have got damaged.

And you'd normally want to keep the supply / ground connections to IC's as short as possible from the planes so hopefully they would have vias very-close to their power / ground pads.

However, if any of the changed capacitors are decouplers for specific IC's (rather than just general distributed capacitance), then you'd normally want the supply/ground connections of the IC's to route straight to the decoupler (as it would be less-effective at decoupling, if it was out on a spur with the added track inductance / resistance).