Another iMac G3 Mystery

[tmtomh]

Hi all,

I'm familiar with all the woes that afflict iMac G3s, but I have a situation that's a variation on the theme, so I'm hoping for some help, or at least some perspective.

The basics:

- iMac G3 slot-loader; 350MHz upgraded to 450MHz via a mobo from a 450MHz unit whose PSU died.

- I upgraded the mobo, added RAM, updated the firmware, installed OS X Tiger, and swapped in a larger HD. Somewhere relatively late in that process, the machine became unstable, often shutting itself down right after startup, and when trying to wake from sleep. The original 350MHz mobo sometimes had wake-from sleep problems, but these never included a shut-down; and with a 400MHz mobo swapped in, the machine had some video corruption (not from the analogue circuitry but from the mobo itself), BUT it ran like a tank otherwise, always waking from sleep and always starting up properly.

- Back to the current situation with the 450MHz mobo - here's the really weird part: I can always get it to start up properly if I zap the PRAM. The PRAM battery is good though, reading 3.57V on my volt meter. I've also reset the PMU via the mobo CUDA button.

- I've swapped HDs in and out, swapped all the RAM (modules and slots), checked, adjusted, and replaced the internal power button (sometimes they tend to get stuck), swapped keyboards, and tried running without a PRAM battery installed. The situation is always the same though - 9 times out of 10 the unit will not start up properly unless I zap the PRAM; 9 times out of 10 it will turn itself off when I try to wake it from sleep; and if it does wake from sleep properly once, it will not do it a second time.

- The common behavior when it shuts down - whether upon startup or upon trying to wake from sleep - is that the HD does not spin up when it is supposed to. So when I turn on the computer, the HD does its initial spin-up, then the grey Apple appears, but then right when the little "spinning gear" should appear and the HD should start loading the OS, the screen gets a darker shade of grey and the computer shuts down. Similarly, when I try to wake it from sleep, the pulsing amber light changes to solid green (as it should), but the HD doesn't spin up, and then the whole machine shuts down.

Any ideas WTF is going on here?

I would chalk it up to the PAV/analogue circuitry problem - but the fact that I can always startup the computer by zapping the PRAM, and that the HD seems to be involved (and keep in mind it's the same with two different HDs), makes me wonder if there's something else going on that I'm missing.

Could it be a hopelessly crashed PMU? Is there any possibility that the IDE cable could be to blame? I realize I'm grasping at straws here, but before I give up and part this thing out I want to make sure there's nothing I'm missing or overlooking.

TIA!

Best,

Matt

 

[equill]

Assumimg that your hardware mix from different sources has not assembled incompatible (in ROM/firmware terms) bedfellows, only one potential fly in the ointment strikes me. You have already hinted at it. The PMU. The PMU is a 'puter-within-a-'puter, and has a 'mind' of its own. When it is reset you must be careful not to let the switch 'bounce', because the second contact can undo the good work of the first. A single push of 1-2sec is needed for reset. One sign that the first attempt has gone agley will be that your nice plump PRAM battery has declined from 3.6V-odd to zilch in a matter of days.

If this is the case, try resetting the MLB by removing all power sources with the mains cable still attached to the switched-off wall power socket to provide a reliable ground, and then lastly remove the cable. Allow half an hour unpowered and unconnected, and then start again from the ground up, with a spanking new 3.67V PRAM battery, keeping the old paws well away from the PMU area, yourself well grounded as you work, and with that unpowered power cable well connected. Insert the PRAM battery just before you reset the PMU, and then close the patient up promptly. Cross fingers.

de

 

[mac-man6]

Could it be a heat problem?

 

[tmtomh]

I found a solution - though I'm not exactly sure why it worked.

I did an erase-and-install of Panther (someone at 'fritter suggested downgrading to Panther, saying "iMac G3s don't like anything higher than that." That wasn't my experience, but I figured what the heck).

Since I did that, it's like a new machine - rock-solid. It boots up properly every time, wakes from sleep, warm restarts okay, and boots properly after being unplugged for a little while.

So my conclusion is that the problem is/was one of three things:

(1) Tiger didn't like the RAM and Panther is more tolerant;

(2) Panther is somehow more tolerant of a wonky PMU;

(3) Some combination of the above affected the firmware update I did to prepare for OS X, and for some reason Panther is more tolerant of whatever's going on in OF than Tiger.

(4) Something strange happened with the original Tiger install and/or update to 10.4.11 and associated updates, that perhaps corrupted some System caches or kernel caches.

I would be inclined to think it was number 2, except a messed up PMU would, I imagine, cause problems with any OS X version; and because I was very careful when resetting it, and even removed the battery and let the mobo sit for awhile before doing so to make sure it worked (I put the battery in before actually pressing the button, as Apple prescribes for this model).

It's very strange though, because no matter what the problem is, the difference is so black-and-white - not only would Tiger crash, but it would shut the computer down, while on the other hand Panther is totally stable, without even a hint of any problems.

The other strange thing: Panther runs much, much faster. Usually on these 350-500MHz G3 machines (iMacs and iBooks), Tiger is about as fast as Panther.

Oh well, who knows - the end result is that this iMac is now running great!

M

 

[Da Penguin]

I have had a similar track record with my upgraded cube. It has a 1ghz upgrade in it that just fights with leopard. It won't sleep properly and it freezes once a week (kernel panic screen 50% of the time).

When it runs in panther, 90% of the issues disappear. There is still the occasional freeze, without KP screen, making me think it is a heat issue.

That doesn't help much with the rhyme or reason, but reinforces the weird panther-ness.

Another item which I found helped a good bit, was doing the good ole nvram reset in OF. From my understanding, PRAM zap should clear this as well, but I have always had better success doing both than just one or the other. If I do this on the cube about once a week, it keeps it from having it's issues almost entirely. I feel like a car mechanic on a very picky sports car.

TBird