help with a computer repair

[bug]

I'm attempting to fix someone's computer that won't turn on at all. I believe that the problem lies in either the pins for the power switch or the trace from the pins to the 20/24 pin socket.

I've tried to bridge the power switch pins without any success. However, bridging the green (#14?) and ground wire on the 20 pin connector WILL start the power supply.

First, does it sound like I am correct in my diagnosis? If so, what is the best option to replace a Super P8SGA motherboard? I need to get something that's compatible enough to avoid the "hey, this is a different motherboard!" BSOD in Win XP.

I'm not even sure which processor is in this thing yet -- I didn't want to remove the heatsink/fan without a good reason.

[zburns]

If you google "Super P8SGA motherboard" you should quickly find out that it is an ATX Form Factor motherboard. ATX is the specification standard that dictates "a standard format that mfgs use" to make this board.

Unfortunately, the PSU is turned on from the motherboard via the case switch; it is done this way because of other computer functions that automatically put the PSU in a "rest" mode but allow "instant" startup when called upon. Ie: coming out of "sleep" mode whether from "intentional sleep by the operator", or the OS putting the computer into "sleep" mode due to inactivity.

I recently read and, probably saved it, an article by Intel on ATX power supplies which was very good; I will look for it during the day and post a link to it. This article would give you the best explanation.

However, I can tell you this much. The mobo turns on the PSU via the +5 volt rail. Once the PSU gets this signal, (from my memory, also logic), the PSU locks itself on until it gets a "future" signal from the mobo to go to sleep mode or shut down.

So your problem lies either in the PSU internal "locking on circuit" or some malfunction within the mobo. You probably already eliminated the front case momentary switch. As long as you know it is not the front panel switch, it would be a PSU problem or a mobo problem. If I can find the article, you might be able to decide which component it is; however, in terms of fixing it, you would have to contact the PSU mfg for some help; you could fix or replace the PSU; fixing might be easy if you got co-operation from the PSU mfg. or even a computer repair shop. PSU circuits relatively easy compared to mobo.

If it is in the mobo, you might can get it fixed, but that could cost $ 100 or more, and it is an old computer in terms of technology, etc..

[bug]

I'm thinking that this is more of a "replace the part" kind of repair rather than something where it'd make sense to actually repair the component.

Since I was able to turn on the PSU by bridging the +5v rail (it would stay on until I cut the power or remove the bridge), I'm thinking that the power supply isn't at fault. Does that sound correct or is there still a possibility that it could be faulty?

The motherboard is definitely the ATX form factor -- I could tell right away by looking at it. http://www.supermicro.com/products/motherboard/P4/915/P8SGA.cfm

I see that it's the LGA775 socket, so I should be able to pick one of many from newegg. If I get a motherboard with the 915 (Grantsdale) chipset, should I be fine from the perspective of Win XP?

[bug]

Okay, it looks like ICH6 boards (like this one) are impossible to find now. However, ICH7 boards are available. If I install one of these, will Windows XP boot or will it bluescreen because of the newer chipset revision?

EDIT: I kludged together my own PWR_ON pins by sticking some paperclip pieces into the power pins on the 20 pin connector. I attached the power button lead to those pins and pressed the power button. IT WORKED! However, the computer turned off as soon as I released the button. Someone who knows more about electronics told me that it sounds like the wipers on the switch are bad as well. Does that sound right to anyone else? I'm going to scavenge a spare switch today and see if it will work.

If it does, it dropped a $60 repair (parts cost, at least) down to $0-2.

[zburns]

1. You unplugged the 20 cable and connector from the mobo, correct?

2. The 20 pin plug that was in the mobo is now in the "air" and you hooked two wires from the front panel switch to green # 14 and ground. When you press the front panel switch, the PSU starts up, but when you release the button, the PSU goes off. Is my statement correct?

3. Assuming my statement correct, you have done the same thing you did initially by " bridging the green (#14?) and ground wire on the 20 pin connector". In this case the PSU would start but when you removed the the "bridge connection", the PSU turned off.

4. If my above dialogue is correct, you only proved that the front panel switch works ok -- nothing wrong with it, provided it is a two pole switch only (has only two connections and therefore, only two wires hooked to it).

Again, assuming my comments correct above, you are missing the fact that the circuits you are dealing with are a little more complex than "the simple circuit of bridgeing # 14 and gnd in the 20 pin plug".

The way the power on circuit for ATX Form Factor mobos and PSUs works is more complicated. Your "working front panel switch" (you proved it works), when pressed, activates a mobo circuit (in simple terms a transistor) that "bridges # 14 and gnd" -- I assume this is what happens in the actual mobo circuits. Other circuits on the mobo deal with "sleep and the software that initiates sleep" plus other conditions that "affect how the software" turns off the PSU and for what reason. Bottom line, the mobo has multiple ways for specific reasons to turn off "PART" of the PSU.

Even when the computer goes into sleep mode, it will come on "instantly" on command. This is because the PSU is never fully turned off, until the "OFF/ON" switch on the back of the case is turned off. Once the front panel switch is pushed and the computer comes alive, the PSU is never fully turned off. This is because the ATX spec requires "instant on" from that point on when coming out of "sleep mode, etc".

The + 12 volt rail which supplies the CPU and other circuits does go off, but one or both low voltage rails ( 5 and,or 3) stay on in order to allow the PSU to come back to "full operating" voltage without the "cold long PSU start up time" (caused primarily, by the time required to charge the PSU electrolytic capacitors and allow other circuit capacitors to be charged, "so that a more or less instant response" to the "full power on mode -- ie. coming out of sleep mode" takes place.

Yesterday, I tried to find a statement that said in effect that # 14 and gnd were "bridged by the mobo" for "Power On condition", and that this connection "had to remain" in order for the mobo to remain operational. The best I found was a statement "to the effect" that the ( I am paraphrasing) "power on signal" had to come on and, "as long as power on was "on", the PSU would be at full operating voltage. But I was not satisfied with the "wording" of what I was reading -- it, sort of, sounded conclusive -- it was the only statement I really found about "turning off", and I wanted more.

Simply saying, that the mobo/psu ON-OFF "how it works relationship" is a little complex -- but nothing as complex as CPU/RAM/etc.? high speed circuits.

Your PSU sounds ok but based on a simple test. Your front panel switch sounds like it is ok. You need to replace the mobo, at least temporarily, to simply prove whether or not the problem is the mobo. Your concern about ICH7 is exactly what? I think it was introduced in 2005, sort of Pentium 4 time frame, seems too early for "Vista" concern, unless Vista was originally planned for example, 2006 intro. Still XP should have been fully considered by Intel.

[bug]

You're right, the front panel button is okay. If I connect it to the 20pin connector (while it's still plugged into the 20/24pin socket) with two pins, the panel button would make the computer run until I let go of the button.

I put together a couple of the things that I tried and came up with a solution. I bridged the #14 and its ground with a staple (!) and then plugged the panel button cable into the proper place on the motherboard.

This will make the fans run constantly, but the power button will now properly turn the computer on and off. Since the computer is always running, there "shouldn't" be a problem with the fact that the fans are always running regardless of whether or not the computer is "turned on". The switch on the back of the power supply will kill power to the computer/fans, of course.

Is my fix electrically/electronically sound? It *works*, but that doesn't mean anything.


EDIT: My concern about going with ICH7 over ICH6 is that Windows XP likes to BSOD and refuse to start if you're using a different chipset than it was expecting. If you move your HD to another box (or use the wrong image when imaging computers), you'll often see this happen.

[zburns]

That is interesting. I will look some more at what you actually did electrically.
When you use the term # 14 pin bridged with a staple to ground, I assume you mean the #14 pin from the 20 pin plug, and you made the bridge on the top side of the cable plug with the cable plug, plugged into the mobo connector. Note 1--Please confirm this.

The fact that the front panel switch now works is what is interesting. I would assume that you have lost "all sleep modes". Or have you? Note--2.

Please confirm both of my notes: 1 & 2. Then I will look some more at how the mobo circuitry works in relation to power on, sleep modes, etc..

(To research what you have indicated is not a wasted exercise. It is an ATX board, what you have done applies to today's stuff, to the extent that the ATX spec has not been amended in this area, etc.)

[bug]

1. Yes, that's exactly how I did it -- stuck the staple in the top of the 20pin connector while it was plugged into the motherboard.

2. This is a business computer that's either used constantly or is off. Sleep modes aren't required in this situation. I'll try to test it today, though.