Recently a friend of mine asked me to help troubleshoot his home computer. He was having Windows boot errors and a few other associated problems. When I first booted it up I realized that it would not actually POST until the switch on the power supply was flipped off and on. I also discovered that every time the machine restarts, the same process has to be used again (flipping the PSU switch) before the machine will POST and move on to Windows. Now, this reminded me of a problem I had in the past with a faulty power supply, so I made a hypothesis that the PSU was the problem here.

However, I used a PSU tester AND later replaced the unit with a brand new equivalent and the same thing happens. What else could be causing the problem? Windows was reformatted and after the switch is flipped once, the computer boots normally and will stay in functional state until the machine is restarted. Could it be something deeper, such as the motherboard or RAM?

FYI, I replaced the IDE cable connecting the only hard drive in the unit, which was reformatted and works fine in another machine. I also replaced the video card with an equivalent model with no change to the boot problem.

Thanks for any help you can provide.

One thing to check is the momentary contact start switch on the front of the case. Make sure it isn't stuck. Did this computer function properly at one time?

Though it didn't function without the problem occurring while I worked on it personally, he claims that it functioned properly before the Windows problems starting occurring. He honestly didn't know much about the boot problem.

The original problem, and why he asked me for assistance, was because Windows was failing to load properly and was giving a STOP message. Due to a lack of my own time to work on the machine, and with his approval, I reformatted the hard drive and reinstalled Windows 2000 Pro on it. This boot problem has happened both before and after the format and clean reinstall.

By momentary contact switch, you are referring to the power button and the mechanism behind it? I've looked at the power button, and the mechanism seems fine. The system just never POSTs until the switch is reset. And I can attest to the fact that it is always after one switch reset of the PSU--never more, never less.

Yes, that is the switch I meant. I have seen that switch stuck in. Something is nagging at me about the way ATX PSUs work. There is always supposed to be power to the MB. Try uplugging from the wall and operating the PSU switch and the power button on the case. Try reseating the PSU to MB connector and check the case to MB wiring. This may be a defective MB. I hope someone else will come along with better advice.

I ran memtest86 on the machine today, and it found an error. Could defective RAM be the cause of such a strange problem? I am going to test a different stick of RAM later and see if anything changes.

What you're not saying is how long this has been going on? Did the user replace ram? Or any other hardware? Make any bios changes? Could be very simple. Maybe you are running async with ram/CPU .. EG: Barton 3200 and PC2100 ram running @ DDR266. Depends on the motherboard and CPU as to what the problem may be with that. You may only need to clear CMOS and reset optimized defaults, or if the user replaced some ram or added a stick that is not identical to the other it 'can' cause issues unless you manaully set timings sometimes. For instance if he were running an XP2000 palamino and added a stick of DDR3200 with an existing stick of PC2100 this would cause an issue with possible reboots and such. You would need to manually set timings (if the mobo allows) to PC2100 speeds for both sticks at the relative defult speeds of the 'slower stick' between the two. Meaning (example) if he added a very fast stick of Kingston hyperX that ran by default @ cas 2 6-2-2 1T and the other stick was rated @ DDR266 cas 2.5 8-4-4 2T you will indeed have MAJOR issued until you manually sync both sticks in the bios. This may not be the uussue at all though.

You gave very little info to go on on you're post No mobo type/speed/chipset. No ram speed/type cas- ratings. No CPU type/speed.

Also could be a stick of mem that is going bad. You would need to use memtest86 to test the modules. Could be and IDE/SATA drive that has a badl cable or needs to be reseated. If it's a self built unit vcore may need to be raised .. the board might be undervolting. Also if it's a self buit unit you may have and errant standoff (mobo standoff) under the board causing a short. You see it can be so many things but it's hard to narrow it down unless you know some specifics about the system and if the user made changes .. and it it ever ran correctly at any point.

Biostar NF3250GB
6800GT BFG
Seagate 200 gig
LanBoy/550W Antec
Audigy 2
XP Professional