Macintosh LC III only booting when system hot/warm

[smiba]

Hey everyone,

I'm working on a Macintosh LC III for a friend that does not post when you power it on. (There is no startup "chime" and it basically just acts dead)

However if you let it be dead for a while, while keeping the system running it eventually starts to work! If you wait for the ICs and the whole board in general to warm up it will eventually post and try to boot its system.

This however is not a very reliable way because even if it has booted, it may still just randomly reboot, especially when its just getting warm.

My experience with computers with issues that slowly dissapear when heated is that there is a tantalum capacitor that is leaking away current to ground. However with no (good) schematic and no bad looking tantalum caps to my naked eye underneath the microscope, its gonna be a bit of a challenge!
I've already replaced all the electrolytic caps (which were all off spec)

Does anyone have any tips? What would be the fastest way to check all the 10 to 20 tantalum caps? I really woudn't even mind replacing them all but I neither have the appropriate values of them.

[rf+tech]

A dead PRAM battery is well known to cause boot issues on these 680xx classic Macs.

https://www.mouser.com/Search/Refine?Keyword=TL-5101-SBP

RF+ Tech

[smiba]

A dead PRAM battery is well known to cause boot issues on these 680xx classic Macs.

Yeah no worries about the battery, I currently don't have a replacement but I've hooked my bench supply on it with 4V.
Its not just an issue with the battery, since it also just spontaneously reboots. (If the PRAM battery is dead usually it just doesn't boot, but once it does you're good)

[Rasz]

have you washed it?

[james_s]

Have you replaced all of the SMT electrolytic capacitors? That is the very first thing I do when working on a 68k Mac. Usually I have to do some cleanup of corroded traces and occasionally repair a via. The board will need a thorough washing to get all the leaked electrolyte off too.

[smiba]

have you washed it?

Not extensively (I did a bunch of brushing with 99.9% 2-propanol), but all pins look shiny and good.

The issue goes away with heat though, which again tells me something like a oscillator or the more likely tantalum capacitor is bad.
Any idea how to easily localize the capacitor thats faulty? Any caps on this mobo that are prone to fail?

Have you replaced all of the SMT electrolytic capacitors? That is the very first thing I do when working on a 68k Mac.

I did! They have not really leaked much, but have lost most of their capacitance. They have been replaced with quality replacement SMD caps.

[james_s]

It sounds like you've done the obvious stuff then. Next thing I'd suggest is get a can of freeze spray (or air duster held upside down) and once the machine is all warmed up and stable, start hitting small sections of the board methodically with a small shot of freeze spray until you find the spot that makes it crash. That should narrow it down to a fairly small section.

[smiba]

Managed to narrow it down to the Y1 crystal oscillator



Ordered a replacement and will update this thread if I succeed

[james_s]

Oh nice job! Do you have a scope? I'd be curious to know what's actually happening, maybe the frequency is way off?

[tooki]

A dead PRAM battery is well known to cause boot issues on these 680xx classic Macs.

Yeah no worries about the battery, I currently don't have a replacement but I've hooked my bench supply on it with 4V.
Its not just an issue with the battery, since it also just spontaneously reboots. (If the PRAM battery is dead usually it just doesn't boot, but once it does you're good)

I still recommend going on eBay and buying a proper battery for it, they’re just a few euros. (And I’m not sure why you’d apply 4V, when those batteries are rated at 3.6V and rarely exceed 3.7 or 3.8V.)

[smiba]

Oh nice job! Do you have a scope? I'd be curious to know what's actually happening, maybe the frequency is way off?

I have a simple digital scope but I won't be able to measure the crystal because just connecting the probe wires will be enough to disturb its oscillation :p
(I've quickly tested it but as far as I could see the oscillation didn't even happen anymore even at 10X just because of the 50cm of probe wire)

]I still recommend going on eBay and buying a proper battery for it, they’re just a few euros. (And I’m not sure why you’d apply 4V, when those batteries are rated at 3.6V and rarely exceed 3.7 or 3.8V.)

Oh yeah of course, Its just that when I was working on this device I didn't have any left and I didn't feel like waiting days for one to come in. When its one of those days where you just want to work on projects workarounds have to be found

I selected 4V for no particular reason other then that a fully charged Li-Ion cell often is about 4V (depending on chemistry, production and type). I could've selected 3.6V, it would work as well.

[tooki]

It’s not a lithium-ion battery, dude. It’s a primary (disposable) lithium thionyl chloride (or similar) cell. At no point would the circuit ever be expecting 4V.

[smiba]

It’s not a lithium-ion battery, dude. It’s a primary (disposable) lithium thionyl chloride (or similar) cell. At no point would the circuit ever be expecting 4V.

Thanks I had no idea! Oops, won't do that again 

[dexters_lab]

Just an FYI but the caps in the PSU are notorious for leaking too, i've just fixed up a LC machine where the PSU was in much worse condition than the SMD caps on the motherboard, they are all through hole in the PSU so very easy to change. At the same time i relocated the battery off the board and used a CR2032 instead as they rarely leak so there should be very little chance of future leak damage to the board.

there are three videos about the repair on my channel which maybe of help
https://youtu.be/XJRvM7hezAE

[tooki]

Long-term, is a CR2032 (alkaline) any less likely to leak than the 3.6V lithium cells? I've never had one of the latter leak myself, luckily, but from what I hear, they never leak before about 10 years. (In contrast, I've had alkaline batteries leak many times, including a few button cells.)

Edit: Never mind, I was confusing CR (lithium) with LR (alkaline).

[helius]

CR cells ≠ alkaline. CR uses LiClO₄ electrolyte in organic solvent, whereas the alkaline electrolyte is KOH in water.
They sometimes do leak, but the risk of damage is much less than alkalines, because there is less liquid and it's much less corrosive.
You can identify a leaked CR cell by a fine white (nitre like) dust left around the (–) side. By contrast, leaking alkalines create a sticky mess which quickly corrodes metal into blue and brown oxides.
The primary lithium cells used in the original equipment were type ER: SOCl₂ (thionyl chloride). Thionyl chloride is quite toxic and also a chloride ion donor, which means it can react with moisture to form HCl—not the best thing for electronics.

[helius]

The reason the ER cells were specified is their high capacity, low self discharge, and high pulse current capability, which meant the computer designers were never constrained by battery performance. A 1/2AA ER cell has 1200 mAh capacity, compared to a CR2032's 220 mAh; and the latter can only supply small steady currents.

[tooki]

CR cells ≠ alkaline. CR uses LiClO₄ electrolyte in organic solvent, whereas the alkaline electrolyte is KOH in water.
They sometimes do leak, but the risk of damage is much less than alkalines, because there is less liquid and it's much less corrosive.
You can identify a leaked CR cell by a fine white (nitre like) dust left around the (–) side. By contrast, leaking alkalines create a sticky mess which quickly corrodes metal into blue and brown oxides.
The primary lithium cells used in the original equipment were type ER: SOCl₂ (thionyl chloride). Thionyl chloride is quite toxic and also a chloride ion donor, which means it can react with moisture to form HCl—not the best thing for electronics.

Ack, you're right, I had a total brain fart there! I was confusing it with LR button cells, which are indeed alkaline, and it's those button cells I've had leak (producing the blue oxides just like you said). I don't think I've ever had a CR cell leak.

[smiba]

Even after replacing the crystal the system experienced similar issues, all the circuitry around the crystal came back fine.
My final conclusion was that the silicon itself has aged, so I decided to replace it with an oscillator.

Very quick video:



Its not my most beautiful work (wish I had smaller AWG wire around) but it does the job

[james_s]

Wow, looks like you need some Kynar wire, or the enamel wire used in solenoids and motors and whatnot works too.

So that solved the problem? Was the original oscillator made of discrete parts or is it integrated into an ASIC?

[smiba]

Wow, looks like you need some Kynar wire, or the enamel wire used in solenoids and motors and whatnot works too.

Absolutely! Wish I had more appropriate wire, could've ordered it but then I would've needed to wait for it to arrive haha

So that solved the problem? Was the original oscillator made of discrete parts or is it integrated into an ASIC?

It did solve the issue! The original circuit is just a crystal, no oscillator.


Removed the two resistors and the capacitor on OSC1. Then connected the output of the oscillator directly to OSC1.

[james_s]

Well you can always rewire it later.

I wonder if one of those capacitors has gone temperature sensitive? I suppose it could be a fault in the IC, good thing it's not in an area that prevents the clock input from working.

[smiba]

They weren't, I've also tried heating the capacitors but that did not solve anything at all.

99% confident this was an issue with the IC itself not being able to get the crystal stable anymore.

[paul8f]

...a can of freeze spray (or air duster held upside down) ...

Yeah, this method works very well. For even better isolation accuracy (especially for ICs and transistors), wet the top of the suspect component with Isopropyl Alcohol. Then as this evaporates it will cool just that component.

[james_s]

Here's hoping it doesn't degrade further.

Years ago I worked on a Bosch dishwasher with a problem. I eventually figured out that it would work if I added a pullup resistor to one of the MCU output pins which was apparently unable to drive itself high. That was probably 8 years ago and as far as I know it's still working.