PDA

View Full Version : Intel Mini Abruptly Reboots




Makosuke
Jul 4, 2006, 02:26 PM
Well, crud. I have a brand new Mini that I was ALMOST done getting set up for work, when all of a sudden it develops a hardware problem. The thing keeps just rebooting, as if the power had flickered off for a second.

Got the OS updated, no problem. Installed 2GB of RAM myself, no problem (but MAN is it tight in there--almost as bad as putting RAM in the first-gen iMacs). BootCamp and Windows installation, no problem. Try to update Windows... and the issues start.

It's way more prone to doing this under Windows (at first when I tried to do anything with the network now it won't even finish booting), but after half hour of running the MacOS it also does this.

Hardware Test CD of course says it's fine. And I tried resetting the SMC and PRAM.


So... since I have to tear the thing apart again to put the factory RAM back before sending it in for service, is there anyting in particular I should look at while I'm in there? Maybe something I could have knocked loose or something? (Though I was pretty darn careful.)

It'd be nice if it was something obvious and I didn't have to wait another week for service (and hope that AppleCare doesn't notice I goofed with the RAM--no, I'm not certified to do it, but only because I haven't been able to justify the money for the course/test yet).



yellow
Jul 4, 2006, 02:29 PM
As I note in the other thread.. memtest.
Could it be overheating?

Fleetwood Mac
Jul 4, 2006, 02:29 PM
It could be that one of the sticks you put in wasn't seated perfectly.. Just for the hell of it, I'd put the Apple RAM back in and see what happens.

Fleetwood Mac
Jul 4, 2006, 02:32 PM
As I note in the other thread.. memtest.
Could it be overheating?
That's possible. This program works for Intel Macs.
http://www.bresink.de/osx/TemperatureMonitor.html
My CPU rarley goes above 60

Makosuke
Jul 4, 2006, 03:13 PM
I'll try memtest, but I would expect bad RAM to cause kernel panics/bluescreens, not abrupt restarts. I'd also expect it to be just as prone to failure during the Windows install, but then who knows.

I've already checked the temp with Temp Monitor, and it's not hot at all (at least the processor cores aren't--both under 50C--and the case is barely warm to the touch). The HD is reporting an unchanging 50C, but I'm guessing that's because the app isn't being able to read the temp from the SMART sensor, or that drive just doesn't report/measure temperature.

yellow
Jul 4, 2006, 03:22 PM
That would be my expectation too.. but since that's your 3rd party option and both OSes are showing the same problem, that'd be my first culprit. Of coruse, it could be some sort of EFI ****-up. Is there an EFI equiv of NVRAM reset?

Makosuke
Jul 4, 2006, 03:38 PM
The closest thing to an EFI reset that I'm aware of is resetting the SMC, which I've tried.

...although it's interesting: I had left it unplugged for a while, then rebooted to turn the BSOD back on in Windows just to see if it coughs up an error there instead of the abrupt reboot. And it seems to be working right now, at least for the last few minutes.

I'm running the updates to see what happens, but this would mean it either is a heat issue (seems odd, given that it's not hot, and it gets a lot hotter grinding away on the CD during Windows' installation), it takes longer than Apple's claimed 15 seconds of being unplugged to reset the SMC, or I jiggled something into place while moving it around.

If it goes back to restarting again in the next few minutes, I guess that'll mean it's heat-related, one way or another. If it doesn't, then I'll have to jiggle it to see if that causes it to fail again. And if that doesn't, then I guess leaving it unplugged for a good while is necessary to REALLY reset the SMC.

Makosuke
Jul 4, 2006, 03:39 PM
...and 10 seconds after I posted, a restart. HEAT?!

livingfortoday
Jul 4, 2006, 03:42 PM
Same thing happened to me on my Mini when I upgraded the RAM. It passed all hardware tests, but in OS X it would randomly restart after thirty seconds to a minute. I pulled the new RAM chip out, and the problem went away. I would put in the stock RAM and see if that gets rid of the problem.

bousozoku
Jul 4, 2006, 04:15 PM
Unfortunately, it sounds like poor RAM. From what people have been saying, it doesn't seem like the firmware of the Intel Macs disables poor RAM. It lets it stay and Mac OS X hiccups on it, the way Windows does.

NoNameBrand
Jul 4, 2006, 04:18 PM
Hey, my mini did the same thing - RAM upgraded to 2GB by the reseller, it would get random reboots, usually temperature related (if the CPU hit 80°C).

The computer has been replaced. I just called AppleCare and then took it to the reseller. No problems whatsoever.


If you need to get at stuff on the mini first, to wipe it or whatever, you can do the following to stop the rebooting:

If you follow the SMC reset instructions for the G4 mini (unplug everything, wait 10 seconds, plug it in while holding the power button (hold for a while), and then release the power button and boot the machine), the machine will come up with the fan running full blast, and won't reboot, so I think it's clearly a heat triggered fault in the SMC/logic board (or RAM). This SMC reset is in the Mac mini (Intel) service manual; the one on Apple's site will result in the fan behaving quietly.

Makosuke
Jul 4, 2006, 09:02 PM
Thanks for all the suggestions.

I was almost positive that it was specifically the RAM overheating, but now that I'm doing more tests, I'm sort of baffled:

Running memtest on the Mac caused a spontaneous reboot after a while, and also resulted in a grey screen with some graphical glitches after the restart. That seemed to point to the memory being screwy, since it uses shared memory.

I let it cool, then just turned it on and let it sit for a while, with everything (airport, bluetooth) off. No reboot. Ok, makes sense, no heat being generated.

So then I crank up both cores to max with a couple yes > /dev/null commands, and get the processors heated up into the 70s and the fan cranking. Sits happily in this state for over an hour. Ok, not JUST heat, must be the RAM.

So then I run Memtest... and it completes without problem this time.

So... not RAM? I turned Bluetooth on and let it sit for a while, and that didn't do anything, so the only variable left is Airport.

I've now turned Airport back on (crappy reception now, perhaps indicitive of something), and it didn't crash just sitting there anyway, so I'm running Memtest again.

I'm doing all this testing now, because it's such a hassle to tear apart I want to know what to be looking at when I get in there. If it's just the 3rd party RAM, that's nice (I don't need to deal with an Apple return), but it is a little surprising, since it's decent DMS stuff, and I've NEVER had a problem with their RAM before. In fact, my MBP has a stick of the same stuff in it, and is quite happy with it.

If it were the RAM, and I could reliably generate crashes, I might try one stick, then the other, to see if both need replacing or just one. I would like to be able to reliably generate reboots SOMEHOW, though (and not just in Windows), so I can test if it's ok with the stock RAM, and if not tell Apple how to reproduce it (I have no Apple store within 300 miles, so I can't just carry it in, and it's an EDU institutional order, anyway).

Counterfit
Jul 4, 2006, 09:27 PM
I would have given up by now. :o

bousozoku
Jul 4, 2006, 09:29 PM
The one thing I've noticed with cheaper RAM is that it's less resistant to electromagnetic interference. I'm sure it's the support chips on the stick and not the RAM chips since they're fairly standard.

Were watching t.v. or listening to music when it failed? Perhaps, you were using your mobile phone?

Makosuke
Jul 5, 2006, 12:55 AM
The one thing I've noticed with cheaper RAM is that it's less resistant to electromagnetic interference. I'm sure it's the support chips on the stick and not the RAM chips since they're fairly standard.

Were watching t.v. or listening to music when it failed? Perhaps, you were using your mobile phone?Nope. Sitting all by its lonesome on a stool, not really near anything at all. But Minis are pretty well shielded, anyway, so I'd be rather surprised if ESD had anything to do with it.

DMS qualifies as "cheaper", but it's not exactly cheap--they use decent chips, offer a lifetime warranty, and specifically guarantee Mac compatability. They're not as fancy as Crucial, but they're frequently recommended along with OWC around here to save a bit without sacrificing much quality. I've been using DMS at work for about 7 years now--everything from ancient Macs to my MBP--and never once had an issue until now.

So anyway, I'm not about to blame the brand, assuming it's the RAM it's just bad luck and a bum stick or two.

I have the thing torn apart now, so we shall see.

Makosuke
Jul 5, 2006, 11:39 AM
I figure I'm beyond the point of help now, but I need somewhere to vent. This is maddening!

I rip it apart and try it with just one stick of the DMS 1GB. No apparent problems. So I try putting both back in... and no apparent problems. Until I try it in Windows.

So now, with the top open (not seeing how that could make that much difference for cooling--if anything, it should have less directed airflow), I CANNOT get the MacOS to cause a spontaneous reboot. Tried running Memtest three times in a row, downloading a 200MB file, nothing. Windows also ran fine for about half an hour, then while I was just reading MS tech docs it suddenly blinks out again.

So it's not fixed, but happening WAY less frequently now, and I can't even reproduce it ouside Windows. What is Windows doing differently with the hardware that's tickling the problem? I guess I'll try running Windows for a while on the factory RAM and see what that does. I just hate not being able to point to a specific situation that causes the issue, and if it's just my DMS RAM, then there's no reason to get it serviced... but if I can't reliably reproduce it, how the heck can I tell?

CanadaRAM
Jul 5, 2006, 11:42 AM
I'll pitch my vote in for a defective power supply. Random brief failure to deliver voltage. You seldom see a RAM or software-based complete, instantaneous powerdown. Even more seldom at random intervals.

And of course, AHT will report everything is fine -- as long as it can complete running. Ditto with Memtest.

Why Windows? Maybe the power management under windows is different? Or Windows consumes more background CPU? Dunno?

yellow
Jul 5, 2006, 11:43 AM
but if I can't reliably reproduce it, how the heck can I tell?

You're a smart fella.. you know you have to put the Apple RAM back in and see if you can get it to blink at all. That's the only good way for you to tell at this point if it's your RAM or something else. Stop resisting.. :)

Makosuke
Jul 5, 2006, 11:49 AM
I'll pitch my vote in for a defective power supply. Random brief failure to deliver voltage. You seldom see a RAM or software-based complete, instantaneous powerdown. Even more seldom at random intervals.You know, that's what I thought at first, and it makes WAY more sense (as I said at some point before, you'd think bad RAM of any sort would just cause BSOD/panics) except it seems to be triggered by some particular set of circumstances in software (or did, until now), one of which was not hitting the power supply heavily--I ran the thing for hours at full load, and it was fine, where a few minutes in Windows would bring it down. I'd say that Windows isn't doing any power management, except the fans don't seem to indicate that--they're pretty quiet most of the time.

But maybe I'm chasing the wrong ghost here. Maybe it's internal, but it'd be nice if it was just the external brick--maybe I wouldn't even have to send the CPU back in for replacement. Wish I had a second Mini power supply around that I could test with.

NoNameBrand
Jul 5, 2006, 12:08 PM
I could get mine to reboot 100% of the time by running two instances of Chess.app, CPUvsCPU games on strong AI; only once did it manage to last long enough to finish a game. Kicking in Dashboard or ExposÚ when the CPUs were close to 80░C would usually push it over the top.*

I would try running it with the fan on 100%, see the last paragraph of my first post in this thread, and see if you can stand the noise - there's enough airflow to play a harmonica with (I tried it :D ).

Makosuke
Jul 5, 2006, 07:01 PM
Thanks for the suggestion, NoNameBrand. That did indeed eventually cause a spontaneous restart, but the processor cores never got above 60C, and it seems like they basically won't with the fan running as it's supposed to.

I'm trying a couple of yes > /dev/null commands to max out the processor combined some 3D running in a window (Enigmo 2 title screen--has a bunch of lighting effects), plus running Memtest--that's basically taxing every subsystem but the disk to the maximum. And I'm 99% sure it's not the disk specifically, since it happily went through the entire Windows installation multiple times, which includes a huge amount of disk access, without complaining. No problems until I actually started RUNNING Windows.

...and with either one of the 1Gig sticks, I can't get it to crash so far. The processor is under 60C, and it's running happily. The floor brick is barely warm, for that matter.

ARGH.

Makosuke
Jul 6, 2006, 03:38 PM
Ok, I think I've narrowed it down, and it may have been my fault. I think it's the audio circuitry.

I first noticed that the Mac was acitng as if it had no sound output. I then realized that while I had been temporarily plugging in the flat brown cable that goes to the audio board but not locking it in place, it apparently was a good enough connection for the Mac to start, but not to have the board behave properly. And since it won't boot at all if that's not plugged in, it could well cause spontaneous reboots.

It would also finally make sense why Windows would be twitchier than the Mac, but not its installer--the Installer doesn't have any driver for the audio out, while Windows by default has the optical output on full-time, which the Mac doesn't.

So, I reconnected it carefully, put my 2GB of RAM back in and was absolutely unable to get the Mac to fail--ran it for hours under heavy load with sound blasting away the whole time. Started it in Windows... and failure 20 minutes later. Dang. Must be a hardware problem that I was exacerbating with the bad connection.

Put the original RAM back in, and Windows wouldn't fail. Ran it all night playing music with the iTunes visulaizer playing, and it was fine. Hmm...

Thinking maybe I just plugged it in better this time somehow without realizing it, I put the 3rd party RAM back without disconnecting that cable, and it's now run for four hours under load playing music in Windows from a network volume (to prompt some network access) all without a hiccup.

It would appear I've fixed it, though it's now so severely irregular that I'm skeptical. I suppose it's better that it was my fault, since I can fix it myself, though I'm still wary that there's actually something wrong with the audio board and I was just making it worse by not connecting it properly.

I do have one question:

On those flat brown cables, what's the proper method of plugging them back in? I've assumed that pushing it all the way in to the white line, then pushing the brown locking thingy down while holding it in place is how, but is that wrong? Anybody know how this is supposed to be done?

GageParker
Jul 7, 2006, 11:16 AM
I am getting behavior very similar on a Intel Mac Mini with 2GB of RAM.

My wife and daughter who have their own (non admin) accounts are not having the problem. Mostly they do iTunes.
I have had the problem when I stress the system by:
playing dvd and MS Powerpoint
downloading Google Earth and simultaneously playing the Intro to Google Earth
and most recently
playing reversi.
The problem comes and goes. After shutdown for a few hours it takes about 45 minutes to recreate the problem. Once it starts it is very easy to reproduce. Most of the time I get a reboot sometimes I get the grey screen with video anomalies, sometimes I get repeated restarts.


I am able to reproduce the problem on wife and daughter accounts by doing one of the above activities. Applecare is telling me to replace the 3rd party RAM installed by my reseller.

I am noting that I am having a sound problem (only right speaker will play).

Makosuke
Jul 7, 2006, 09:05 PM
Well, your sound problem sounds like somethiing worth sending it in to Apple for--that's obviously an issue not caused by RAM, unless the people who installed it screwed up the audio cable while they were in there.

In my case, the audio thing appears to have been just another red herring; I hadn't connected it properly and it was causing problems, but it wasn't the root. After a lot more methodical testiing it basically boils down to this:

If I put in the stock Apple RAM, it will run under as much load as I can put it under without issue (we're talking maxing out both processor cores, doing a RAM test to exercise the RAM, getting the 3D chip going with a game, playing some iTunes over the network to get some Airport access going, and playing a movie in the background to generate disk access). It'll do this for hours on end, and the processor cores never get much above 60C.

Likewise, if I put either of the 1GB sticks in individually, it will run whatever I throw at it for as long as I want it to run.

If, however, I put both 1GB sticks in and boot Windows, it will spontaneously reboot 10 minutes later, even if I don't do anything. The MacOS is a little more irregular about timing, and load seems to accelerate it, but it's the same.

I'm guessing it's in some way heat related, since it was more stable when I had the case open (this didn't change the temperatuers of the processor cores measureably, but might allow more airflow past the RAM, I suppose). I'm waiting to see what DMS tells me about an RMA--if this is any sort of a common issue, surely they'll have heard about it before. Certainly hoping it's just bum RAM, since I can't replicate it with the stock config so it's going to be hard to get Apple to look at it.

Makosuke
Jul 8, 2006, 12:55 AM
I just did one further test to confirm that it is heat related: I tried closing it up and pinning the fan on full blast using the tip above, and indeed, no restart when I just let Windows sit there. I didn't bother to put it under load, because knowing that the fans help was enough.

So I guess it's the RAM overheating, and only when there are two sticks stacked on top of each other--I'm sort of assuming that only one leaves enough airflow to keep the temp acceptable. What I CAN'T figure is why A) the RAM would overheat to begin with, since it shows no other problems and B) The mini would spontaneously reboot instead of just panicking.

Perhaps it's because the graphics chip shares main memory, so when the main memory goes screwy it freaks out and just pops off?


I suppose just one of the sticks may be bad (so statistics dictate), and I could figure out which by pairing each with one of the stock sticks and seeing which causes reboots. Ugh, more testing...

After G
Oct 18, 2006, 11:50 PM
I know it might be too late for the people on here who have had their problems, but I wanted to tell people about my experience with this problem, as it may save somebody a LOT of headaches in the future.

I had the "abrupt" rebooting problem as well. I thought it was bad memory as well, after reading what people did. I tried doing the "yes > /dev/null" thing, and running memtest. The problem seemed to be replicatable by running memtest, and the computer would randomly reboot.

After reseating cables again and again, resetting the SMC and PRAM, swapping sticks, etc., I happened to notice a picture on a Mac mini take-apart guide. (I was using the guide to help me take apart the Mini. It seems there is a hardware button to "reset PRAM" (according to the site, it resets BIOS). It is circled in violet in the picture. Push this button ONCE, and all your problems should be solved. I know about the keyboard shortcut, and tried it multiple times before opening up the mini (it's a pain to use the putty knife).

60159

If anyone wants to read the guide, it's at http://hdtv.o0o.it/mini/

Makosuke
Jan 8, 2007, 03:19 AM
I'm re-dredging up a really old thread because after being away for a while I noticed the dredging-up post, and realized I never reported on my final solution.

So, for point of reference if anybody sees something similar:

In my case, the problem was indeed this specific pair of DMS RAM sticks. After seemingly endless iterations of testing, either stick would, WHEN NOT ALONE, eventually cause a spontaneous reboot.

I contacted DMS, they did an advanced replacement with a new pair of sticks, and those worked absolutely flawlessly from the get-go. A VERY weird problem, and not one most people are likely to ever encounter, but at least in my case it was indeed defective RAM.

Kudos to DMS, however, for agreeing to replace it and even advance-shipping me replacement RAM with a prepaid label to return the defective stuff.

rsucre
Feb 1, 2007, 10:01 PM
So with the new RAM, the Mac is not overheating? Or it does overheat (reaches 80C) without rebooting?

Makosuke
Feb 13, 2007, 10:26 PM
I assume the previous poster meant 80C, not 80F (which is basically room temperature), but yet another delayed, for reference reply:

The mini did NOT overheat with the replacement RAM--it maintained a totally reasonable temperature--and worked fine.

...but it wasn't registering temperatures any higher with the bad RAM, nor did it feel subjectively any hotter--the RAM sticks didn't feel unusually warm. I'm assuming that the problem wasn't that they were generating too much heat, but that they flaked out once they hit a certain ordinarily acceptable temperature.

I can say for sure the underside of my MBP, which uses essentially the same RAM, gets considerably hotter than anything in the Mini did.

Point being, the issue was NOT overheating or any kind of unusually high temps in the Mini, it was RAM that couldn't handle perfectly normal temperatures.

rsucre
Mar 3, 2007, 09:26 PM
Thanks to Makosuke, my abrupt reboot issue is now solved. I took the mini to the place where I bought (and installed) the RAM, they replaced one of the modules and the problem is gone! It used to reboot in every memtest. I was able to run > 5 full tests without a reboot.

xpetex
Jul 28, 2007, 07:07 AM
Thanks to Makosuke, my abrupt reboot issue is now solved. I took the mini to the place where I bought (and installed) the RAM, they replaced one of the modules and the problem is gone! It used to reboot in every memtest. I was able to run > 5 full tests without a reboot.

I had 2 gigs in my mini, and had these problems too. The store was completely unhelpful, so i ended up getting two 512 mb circuits from cruical, and so far it hasn't rebooted after 5 days or so. used to happen every day, and i could repeat it using memtest.

The reset pram thing seems interesting, perhaps i will try it to see if it makes any difference. it works pretty well with just 1 gb, but it'd be nice to have the 2 gb working.

thanks guys for making this thread. atleast i have a stable computer now.

sucks i had to use money from my own pocket to solve something that should be covered by guarantee, but i didn't want to be without the computer another 3-4 weeks like the 2 previous times when i took it in.

xpetex
Jul 28, 2007, 11:08 AM
The reset pram thing seems interesting, perhaps i will try it to see if it makes any difference. it works pretty well with just 1 gb, but it'd be nice to have the 2 gb working.


thansk for the tip, but it didnt' seem to do the trick for me, so it was back to 1 gb. well, that's fine.

Cougar428
Apr 6, 2008, 12:28 AM
Interesting problem. I know this is an old thread, but it's the exact same problem I'm having with my Mac Mini 1.66 Core Duo. If I put 2gb in it, it randomly restarts, with either of the 1gb sticks by themselves, it's fine. It's also fine with both sticks in, but the top one not plugged in correctly (so it's not detected). I ran it with just a single 1gb stick for about a year, not a single issue. I upgraded the hard drive the other day, and thought I might try plugging the second stick back in, and the problem may have returned. (Only had it reboot once so far, but that's once too many times).

Memtest works fine, passed 10 tests without issue, but I know it'll reboot itself at some point. A really annoying problem. I'll try with a 1gb stick and a 256mb stick and see how that goes.