Macintosh Portable: Twitching in its sleep.

[Stephen_Usher]

OK, here's a strange one. My Macintosh Portable works fine when powered up but after about 10 minutes of use if I shutdown then it "twitches in its sleep". The sound crackles and the display randomly displays things. Pressing the spacebar wakes the machine and it's happy again.

This happens whether or not the power supply is connected. The M/B has been fully recapped.

Here's a video: Twitching

Any thoughts on the component(s) causing the problem?

 

[Stephen_Usher]

Could it be Q11 breaking down? That's the only component I can see which controls that power rail.

 

[techknight]

Never seen this happen before. Anything is possible with this. 

one of the P Channel MOSFETs could be shorted on the motherboard somewhere, or, the PMGR IC is bad. 

 

[Stephen_Usher]

I replaced both Q10 and Q11.

Q10 shouldn't make any difference as it's just controlling power on the IDS connector as seems to be bypassed by as couple of diodes anyway.

No change. So, I just put a 'scope on the /SYS_POWER line and it's floating very close to the 2.7V value. This would why the power is going on and off.

This means that the pull-up resistor R142 must have failed open, or there's something elsewhere on /SYS_POWER pulling it down..

 

[Stephen_Usher]

Nope. I put another 100K resistor over the top (making it 50K effectively) and the voltage rose slightly. However, this means that it must be some other component dragging the line down.

As far as I can see, /SYS_PWR only goes to the power manager, video chip, CPU GLU,  MISC GLU, a AC244 (U13C), an AC153 (U11K, SCC) plus another couple of MOSFETs (Q4 - external video, Q13 - floppy).

 

[Stephen_Usher]

Can someone measure the resistance (in circuit) between the two sides of R142 on their machine? This should at least give me a baseline of what I should expect. (Given the logic I would think probably close to the 100K pull-up.)

 

[Stephen_Usher]

I've added a 10K ohm resistor on top of R142 now as a test.

/SYS_PWR is now mostly ~4V.

A lot of the time the output is stable, but then there are bursts where it changes, which looks a lot like some sort of leakage from another logic line somewhere:

Video of /SYS_PWR

The thing is, what is actually running when the machine is asleep which could be producing these signals? If I can find out then this may localise the problem.

 

[Stephen_Usher]

I've been monitoring the machine all afternoon.

For most of the afternoon the /SYS_PWR line settled down to 3.6V (with a few glitches up to 4V).

I then booted from floppy and shutdown and it returned to 3.6V, but with more times at 4V.

However, after a few quick cycled of sleep-wake-sleep there was a lot more glitching on the line, sometimes almost getting up to 5V for brief periods before it dropped down to averaging 3.2V, so the up/down/up/down caused whatever component which is failing to get worse.
 

 

[Stephen_Usher]

And back up to 3.8V, with (rare) glitches up to 4.98V. So it's not really getting worse, just switching between stable levels.

 

[techknight]

sys_pwr should be going to the PMGR. 

Keep an eye on the PMGR they are a known problem in the portable. 

 

[Stephen_Usher]

sys_pwr should be going to the PMGR. 

Keep an eye on the PMGR they are a known problem in the portable. 

I've also seen many 74xx244 chips go bad in strange ways, including sending stuff down the inputs.

 

[techknight]

True, but i've worked on at least 50 of these motherboards of the years, and I had ran out of PMGRs on parts boards due to failures like this. I have not seen this "exact" failure so your treading on unknown terrain here anyways. 

something to keep in mind. Also, personally, I have never seen a 244 and 245 fail. But I dont usually service computer equipment, most of the stuff I serviced was analog. 

Also I dont know if you have the backlit, or non backlit version, but there is a cluster of 74 series logic ICs on the non-backlit version that are to the right of the CPU that tend to spontaneously short out. ive had to change a number of those as well. 

 

[Stephen_Usher]

Yeah, it's a non-backlit. From what I remember those are a set of three 244s. 244 buffer chips are notorious for going faulty.

 

[techknight]

That's odd. Maybe its the luck of the draw. I have yet to come across a bad 244 on anything ive ever serviced. 

Now, Muxes, and 166s, things similar to that, I see bad all the time. Especially in Arcade level stuff. 

also the set of 3 are above the CPU. not to the right of the CPU. there is a 74LS02 and a handfull of other ones. Ive seen them blow out on several occasions. and spontaneously. 

 

[techknight]

bump...

You figure this one out yet? Just curious as I am as well... 

 

[Stephen_Usher]

I'm afraid not. Without starting to remove chips one by one, which is dangerous in of itself, until I find the one pulling the line up there's not much more I can do.

Given that the machine is not going to be powered up and sleeping much (it's going to be mostly packed away with all batteries removed) and only shown as a curiosity really, the 10K pull-up work-around is "good enough". If the machine were to be spending a significant amount of time asleep or shutdown I would consider it more of a priority due to the increased risk.