Old computer know-how

[eagl]

It's amazing how soon we forget things that used to be common knowledge when troubleshooting computers.

I accidentally pulled out the power cord for my windows home server, so it abruptly shut down of course.  The computer is pretty old, dating back to when a 1.0 ghz Tualatin Celeron was the overclocker's friend, maybe around 2004?  It's so old that the mobo only supports booting from hard drives smaller than about 200GB, so the WHS install is on an 80GB hard drive with 2 larger drives on an add-on PCI sata card (pci, not pci-e).  And no built-in NIC or video either.  AGP port, and the mobo has the premium feature of 5 channel surround audio.  Nice, back in 2004...

Anyhow, after the abrupt shutdown the computer wouldn't boot.  Wouldn't post, wouldn't show anything on the monitor, apparently wouldn't even initialize the AGP vid card, wouldn't beep, nothing.  Repeated off-on cycles, holding the reset button down for a while, nothing worked.  I was thinking I had destroyed the entire computer.  Then I remembered one trick I hadn't tried yet.  I unplugged the power cable again (after turning off the computer by holding down the power switch for 5 seconds), then pushed the power button again.  As the capacitors all discharged, the lights briefly came on for a second then faded.  For good measure, I cycled the power switch on the power supply off and on (yea it is so old the psu itself has a power switch too).  Plugged it all in again, and it worked.

It's dumb but I troubleshot that computer for 30 minutes and was sure I was going to have to completely disassemble it to troubleshoot it including using the bios reset jumper (which could potentially render the sata drives on the add-on pci sata card unreadable), before I remembered that last trick from way back when the ATX mobo and PSU spec was new.  And fortunately when it all came back up the WHS backups were intact and the computer is once again very happy being the slowest WHS backup machine on the planet...

I plan on migrating it to a slightly less ancient computer with a dual core AMD cpu from 2006, but I needed to get it up and running and keep it healthy until I buy another large-ish drive to copy off the backup database.  I have just under 1TB of compressed WHS-style backups and I really like how the old original WHS works, so I don't want to ditch the whole thing and lose what I have just to get a second cpu core, faster NIC, and faster HDD transfer speeds.  It works really reliably now, but its just really slow.

[olds442]

Those power supply switches are VERY useful my friend! even my corsair HX 850 watt PSU from 2012 has one for good reason. The soft power button is hardware mapped. Meaning it sends a BIOS interrupt which during a interrupt storm can cause it to not work. unplugging works too but switching the power supply OFF simply kills power to the system but also it resets the PSUs onboard controller which is handy as that restarts the whole PSU. The PSU turns on when the green wire on the 20/+4 pin connector is connected to the black ground wire. So the controller waits for this. But if the controller got stuck when the power cord came out it tried to restart from where it was (to clear control capacitors and such) and got stuck.

[Bizman]

Thanks for the trick! I just so happen to have two dead older computers which won't boot. I suppose they are beyond that trick, though, since I've already been testing them with another PSU without success. Good to remember anyhow. 

[ROC]

Thanks for the trick! I just so happen to have two dead older computers which won't boot.

Replace the MB Battery   I did that once and a computer I thought was fried got 2 more years of solid life lol

[Bizman]

Replace the MB Battery   I did that once and a computer I thought was fried got 2 more years of solid life lol

Checked with a multimeter, 3.3 volts.

[Bizman]

Replace the MB Battery   I did that once and a computer I thought was fried got 2 more years of solid life lol

After replacing and cross testing every other component I checked the battery for the other dead computer: 1.1 volts! Replaced it and bingo! I owe you a beer. 

Funny how these basics get forgotten...

[icepac]

I did that the other day but still got cmos checksum errors on an industrial motherboard for our dyno.

The dyno guy wanted 15 grand to fix it but I replaced a few components in the hand controller, the motherboard, and a few caps on the controller board on the dyno itself and was back in business for less than 500 bucks.

I keep sharp on all the old school skills because most of them still work on modern equipment and have a sweet hakko desoldering station I use for modifying automobile engine ECUs.

Nice to see you back up and running but I still can't figure out why my old Iwill scsi motherboard from 1995 still hasn't suffered a "battery failure" but it might have something to do with it being part of the dallas real time clock.

[Bizman]

Replace the MB Battery   I did that once and a computer I thought was fried got 2 more years of solid life lol

Just out of curiosity I changed the battery of the other computer, too, although my multimeter had previously measured a solid 3.3 volts for it. Click, beep - actually more beeps, since the ram stick was missing. Ram in place, click, beep, what the heck! Green led on the monitor, picture and text saying keyboard error. Unplug, reinstall more components, click - beep - windows. Shutdown, unplug, rest of the components back from the going to be a replacement rig, plug, click, beep, windows... No problems whatsoever.

Time to recheck the original battery: 7.4 - 8.3 - up to 8.4! Turning the knob between 20 and 200 volt area gave a short measurement of 3.3, but it went back high. No wonder the computer wouldn't boot.

ROC, I owe you another one.  


@icepac: Motherboard batteries seem to last longer if the computer is being plugged into mains all of the time. The stand-by power will take care of keeping the settings. I've noticed that computers behind an extension cord with a switch need battery change more often than the directly plugged. Old rigs might also have less things to do with the battery.

[ROC]

No way!  That is too cool 

[Bizman]

No way!  That is too cool 

Too cool? The beer? My wife's cousin used to get the worst hiccups from cold beer, he had to order it microwaved. Really.

Seriously speaking, I'm beginning to wonder how many mobos I've doomed to be defect without changing the battery. Until now I've been in the belief that a failing mb battery either shows

• at startup on the first screen saying something like "your settings have changed, press F1 for settings, F2 to continue with your own risk"

or

• while in Windows, the Security Center nags about your freshly updated antivirus, after which you notice that the Windows' calendar and clock lay five years behind

[Bizman]

Blimey! The latter stopped working again, so it wasn't a battery issue after all. The original battery started showing a solid 2.9 and the new one 3.3 volts.

[ROC]

OH man Biz, that bites.
Well it did work with a fresh battery, which leads me to wonder, have you cleared the cmos settings, using the jumper on the motherboard?
Changing the battery might have caused it to reset, if you clear it, and it works again, that narrows down the possible problems.

Did you add new hardware or software recently that might be conflicting with an old bios?  Have you updated the bios?

Now I'm curious.  I hate it when things work just fine...sometimes.

[Bizman]

Prior to changing the battery I had done every possible trick, including resetting the cmos. The mobo in question doesn't have a jumper for that, but I shortcut the connectors inside the battery tray. After it failed today again I reset it again, several times. I also fiddled with the power cord (no switch) to no awail. The only difference between before and after resetting the cmos was that the pulse of startup attempts changed pace similarly both with the old and new battery. It looks like the psu were signalling the mobo but the mobo wasn't answering. I tested it with several psu's. The original one makes a clicking sound, another one has a led indicator which flashes in the same pace as the other one clicks.

The setup is highly original, it's an elderly Fujitsu-Siemens Scenic with a new hard disk being the only alteration. The latest bios is from 2006... After a thorough checking and fresh reinstallation I sold it to the customer about a year or maybe more ago to replace his even more ancient rig. It just stopped working one day after being totally co-operative the previous evening. I was already prepared to move the hdd to a slightly younger F-S Esprimo with similar specs, but the XP repair installation wouldn't work. To avoid starting from scratch I gave the battery a final try, which to my big surprise worked! I had already called the customer this morning to arrange for delivery, but luckily he had some family plans for the day. I had already unplugged the rig when I thought I'd better boot it once again. Good for me I did... So I cancelled our date and continued with plan A. The Esprimo is now downloading the latest 100+ updates for XP SP3, after which I'll install some applications and copy the essentials of his profile from an external hdd.

[eagl]

Look at the capacitors on the mobo.  If any of them are bulged or have any crusty stuff on them, they're blown and the mobo is probably wrecked unless you can solder new caps on. 

[Bizman]

I know how to tell a failing cap by its looks and I can also replace them. I'm no electrician, though, so I can't tell if there's a broken fet or something like that. I also lack the equipment to measure the capacitors that aren't bulged, and for what I've been told it would be quite an effort to remove each single cap for measuring. There's about 35 of them in this case, so if it were a critical business computer I might even replace each of them. OTOH a business rig should have a working backup to reduce the downtime.