Does it just not power at all right now? If it's not even powering up then it could be there's a short somewhere, or the Super IO has failed. Probably best to check the ATX connector for shorts between 5v > gnd and 12v > gnd, if it's less than 1 ohm then that'd be a short, 100ohms or 50 ohms is normal.

If there is a short then you could try pulling the motherboard out of the case and make sure that there are no standoffs in the case shorting a part of the mainboard that has no corresponding hole. A minimal setup out of the case is a good way to verify what's going on as well.

Do you have a POST code reader card or does the system make any beeps? It seems with no power that's not going to do much good yet? If it is powering and acts dead, wiggle / reseat the RAM, that tends to fix things often.

step one - disassemble

My only KT333 motherboard was an Epox that came with a very poor northbridge cooler - it was a passive chunk of aluminum, convex, and barely held in place with some thin and cheap thermal tape. At stock speed, it worked fine, and only (and very quickly) became an issue when I began to overclock the frontside bus. So I don't think that would have killed it.

The power supply that is in the photo is a low end one - I wouldn't trust it, and I hope you have a better one on hand. As has been mentioned, pulling it and testing it outside the case with a PCI video card and one stick of RAM would be the next step.

Thermalwrong wrote on 2023-09-10, 21:05:

Does it just not power at all right now? If it's not even powering up then it could be there's a short somewhere, or the Super IO has failed. Probably best to check the ATX connector for shorts between 5v > gnd and 12v > gnd, if it's less than 1 ohm then that'd be a short, 100ohms or 50 ohms is normal.

If there is a short then you could try pulling the motherboard out of the case and make sure that there are no standoffs in the case shorting a part of the mainboard that has no corresponding hole. A minimal setup out of the case is a good way to verify what's going on as well.

Do you have a POST code reader card or does the system make any beeps? It seems with no power that's not going to do much good yet? If it is powering and acts dead, wiggle / reseat the RAM, that tends to fix things often.

Thank you. My meter, though I'm not convinced it's the most accurate, indicates <1 ohm between 5v > gnd and 12 > gnd, so it reads like a short. However, this is the case with the board out of the case and isolated, as well. Testing the other pins more or less at random to check the meter does give readings usually in the 100s of ohms, albeit it only transiently.

It's as dead as can be when I try to power it up - no fans, no beeps, just as though it is off at the mains.

I've a bad feeling this is terminal for the board, or is there anything else that worth considering?

Repo Man11 wrote on 2023-09-10, 21:57:

My only KT333 motherboard was an Epox that came with a very poor northbridge cooler - it was a passive chunk of aluminum, convex, and barely held in place with some thin and cheap thermal tape. At stock speed, it worked fine, and only (and very quickly) became an issue when I began to overclock the frontside bus. So I don't think that would have killed it.

The power supply that is in the photo is a low end one - I wouldn't trust it, and I hope you have a better one on hand. As has been mentioned, pulling it and testing it outside the case with a PCI video card and one stick of RAM would be the next step.

Thanks. I'll dig out another PSU - I have tried a couple, but they are old and to be honest wouldn't be surprised if they are goosed, too. That's reassuring about the northbridge cooling, I did think the cooler doesn't look up to much and I think the same chipset on similar MSI boards only has passive cooling anyway. It's not like I've been overclocking the FSB.

strange_loop wrote on 2023-09-10, 21:59:

Thank you. My meter, though I'm not convinced it's the most accurate, indicates <1 ohm between 5v > gnd and 12 > gnd, so it rea […]
Show full quote

Thermalwrong wrote on 2023-09-10, 21:05:

Does it just not power at all right now? If it's not even powering up then it could be there's a short somewhere, or the Super IO has failed. Probably best to check the ATX connector for shorts between 5v > gnd and 12v > gnd, if it's less than 1 ohm then that'd be a short, 100ohms or 50 ohms is normal.

If there is a short then you could try pulling the motherboard out of the case and make sure that there are no standoffs in the case shorting a part of the mainboard that has no corresponding hole. A minimal setup out of the case is a good way to verify what's going on as well.

Do you have a POST code reader card or does the system make any beeps? It seems with no power that's not going to do much good yet? If it is powering and acts dead, wiggle / reseat the RAM, that tends to fix things often.

Thank you. My meter, though I'm not convinced it's the most accurate, indicates <1 ohm between 5v > gnd and 12 > gnd, so it reads like a short. However, this is the case with the board out of the case and isolated, as well. Testing the other pins more or less at random to check the meter does give readings usually in the 100s of ohms, albeit it only transiently.

It's as dead as can be when I try to power it up - no fans, no beeps, just as though it is off at the mains.

I've a bad feeling this is terminal for the board, or is there anything else that worth considering?

There are certainly [large] capacitors between the voltage planes and ground. The planes themselves will also have a capacitance. They'd have to be charged for a correct reading.

The way multimeters usually measure resistance is by applying a small known constant current, measuring the voltage, and calculating the resistance with Ohms law. If that current is consumed by charging the caps ... well, you catch my drift 😀

Assuming those caps were discharged when you begun measuring, I'd take quick readings from a meter from a meter I wasn't sure was accurate with a grain of salt. You could try to measure for a bit longer, and see if it stabilizes on a higher reading. Don't count in it though, that measuring current is very small.

There are also a lot of other components that potentially could interfer.

In short, I wouldn't necessarily interpret those readings as the board being shorted out and dead.

quote]

There are certainly [large] capacitors between the voltage planes and ground. The planes themselves will also have a capacitance. They'd have to be charged for a correct reading.

The way multimeters usually measure resistance is by applying a small known constant current, measuring the voltage, and calculating the resistance with Ohms law. If that current is consumed by charging the caps ... well, you catch my drift 😀

Assuming those caps were discharged when you begun measuring, I'd take quick readings from a meter from a meter I wasn't sure was accurate with a grain of salt. You could try to measure for a bit longer, and see if it stabilizes on a higher reading. Don't count in it though, that current is very small.

There are also a lot of other components that potentially could interfer.

In any case, I wouldn't necessarily interpret those readings as the board being shorted out and dead.
[/quote]

Ah, that makes sense of the resistance "peaks" I was noticing. It does make me wonder about the fact that I seemed to be getting no such transient recording when trying the 5 and 12v pins.

I tried with a known working PSU, still not a great one, but definitely a working one but still no response. I wonder if this one might be beyond my technical ability to resolve (if that weren't already obvious. . .)

strange_loop wrote on 2023-09-10, 21:59:

My meter, though I'm not convinced it's the most accurate, indicates <1 ohm between 5v > gnd and 12 > gnd, so it reads like a short. However, this is the case with the board out of the case and isolated, as well. Testing the other pins more or less at random to check the meter does give readings usually in the 100s of ohms, albeit it only transiently.

Hmm maybe your meter is bad ? With it out of the case: pull cpu, ram and just have bare board. If it still reads ~ 0 ohms from 5v and 12v to ground then there is a true board short. Could be just bad caps...

WAIT !! Added: are you sure you are measuring based on the Motherboard pinout and not the PSU pinout (if you google ATX pinout most are the PSU not the motherboard)
This is the motherboard side: http://www.interfacebus.com/ATX_Motherboard_M … tor_Pinout.html
Ignore all those others as most are PSU side and you will get bad readings !!
I have a KT266 Ultra Pro2 w/cpu and from +5v to ground starts at about 20 ohm and ramps to about 1k in a few seconds, with black lead to any ground and red lead in the +5v

Thanks, everyone. Sorry for the obvious question, just incase I'm doing something stupid. . . I'm measuring resistance across the pins, right? Or should I have the ground probe contacting some other ground e.g. the case?

And in terms of the scale on the multimeter, am I going to the correct range as in the photo?

I'm considering picking up a gigabyte board with an nforce chipset (I did have one back in the day but I don't want to just waste money and also give up on parts, and I also don't want to go too modern but I guess as long as it's Athlon XP I'm still good for Windows 98 era and can still put a voodoo 3 in it.

strange_loop wrote on 2023-09-11, 07:15:

Thanks, everyone. Sorry for the obvious question, just incase I'm doing something stupid. . . I'm measuring resistance across the pins, right? Or should I have the ground probe contacting some other ground e.g. the case?

And in terms of the scale on the multimeter, am I going to the correct range as in the photo?

I'm considering picking up a gigabyte board with an nforce chipset (I did have one back in the day but I don't want to just waste money and also give up on parts, and I also don't want to go too modern but I guess as long as it's Athlon XP I'm still good for Windows 98 era and can still put a voodoo 3 in it.

What you want to measure in any given time defines where you want to stick your probes. With your multimeter you measure a difference between two points and ground is the common reference point used to measure current, voltage or resistance in most cases. About ground, you certainly don’t need the case. Your MBs ground plane is connected to case and PSU, and PSU ground is finally connected to ground, that is earth through the socket.

You can find ground connections throughout your motherboard. For example, screw holes which are used to attach the board to the case and have those metal contacts, are also connected to the ground plane of the MB. And through there to the case and ultimately through the PSU socket to earth.

If you want to mesure resistance and find out if something is shorted to ground, you want use the lowest setting, because in case of a short, you will have a very low resistance. If I were you I’d get myself a multimeter with continuity test. It is handy for measuring shorts.

By the way, short doesn’t mean necessarily that something is shorted to ground. It is again all about reference points and short is possible with any two points which should have different voltages. Ground, which is zero voltages, shouldn’t be connected to any voltages. You could also have a short between +5V and +12V without either one being shorted to ground (this is generally speaking and not related to the case at hand).

CharlieFoxtrot wrote on 2023-09-11, 07:44:

What you want to measure in any given time defines where you want to stick your probes. With your multimeter you measure a diffe […]
Show full quote

strange_loop wrote on 2023-09-11, 07:15:

Thanks, everyone. Sorry for the obvious question, just incase I'm doing something stupid. . . I'm measuring resistance across the pins, right? Or should I have the ground probe contacting some other ground e.g. the case?

And in terms of the scale on the multimeter, am I going to the correct range as in the photo?

I'm considering picking up a gigabyte board with an nforce chipset (I did have one back in the day but I don't want to just waste money and also give up on parts, and I also don't want to go too modern but I guess as long as it's Athlon XP I'm still good for Windows 98 era and can still put a voodoo 3 in it.

What you want to measure in any given time defines where you want to stick your probes. With your multimeter you measure a difference between two points and ground is the common reference point used to measure current, voltage or resistance in most cases. About ground, you certainly don’t need the case. Your MBs ground plane is connected to case and PSU, and PSU ground is finally connected to ground, that is earth through the socket.

You can find ground connections throughout your motherboard. For example, screw holes which are used to attach the board to the case and have those metal contacts, are also connected to the ground plane of the MB. And through there to the case and ultimately through the PSU socket to earth.

If you want to mesure resistance and find out if something is shorted to ground, you want use the lowest setting, because in case of a short, you will have a very low resistance. If I were you I’d get myself a multimeter with continuity test. It is handy for measuring shorts.

By the way, short doesn’t mean necessarily that something is shorted to ground. It is again all about reference points and short is possible with any two points which should have different voltages. Ground, which is zero voltages, shouldn’t be connected to any voltages. You could also have a short between +5V and +12V without either one being shorted to ground (this is generally speaking and not related to the case at hand).

Thank you. With this is mind, am I able to measure anything useful at all in this specific case with the motherboard isolated from any external grounds, i.e. no case, no PSU, sitting on a non-conductive surface? I wasn't sure if it was helpful at this stage to be measuring continuities/discontinuities of internal circuits

If measuring with the PSU attached to get a ground through the mains, I guess I can just jam the probe in at the back of the atx plug whilst the plug is in the MB socket?

strange_loop wrote on 2023-09-11, 08:06:

Thank you. With this is mind, am I able to measure anything useful at all in this specific case with the motherboard isolated from any external grounds, i.e. no case, no PSU, sitting on a non-conductive surface? I wasn't sure if it was helpful at this stage to be measuring continuities/discontinuities of internal circuits

If measuring with the PSU attached to get a ground through the mains, I guess I can just jam the probe in at the back of the atx plug whilst the plug is in the MB socket?

If you want to measure shorts on motherboard against ground, for example, you don’t need to connect anything to it. You have the ground reference on the board. Just slap the board on your desk and start measurung and stick the other probe to some of the countless ground points on the mb. If you want to find out that your system is all the way properly grounded, then sure, you want to measure from the mainboard ground all the way to the PSU socket, but this is hardly something that is at all relevant here.

You don’t also need to see that much trouble taking care that your board is not affected by the surface you are doing the measurements, unless you have metal table or some other metal crap on the surface. You’ll be fine if the surface isn’t conductive stuff so it can’t short pins under the MB for example. If you have conductive surface, it would be a similar situation if you’d screw MB to case without stands and thus shorting pretty much all solder points under the mb more or less together and to mb ground. And that is not generally a good thing 😅

Edit: like I previously said, measuring with multimeter is always relative: you have a reference point and you measure the other point against it. Multimeter has no clue if that reference is ground or whatever. It is 100% your choice based on what you want to measure. If you have a live circuit and you put other probe to 5V and other to 12V and measure voltage, multimeter should show approximately 7V, because that is the potential difference between those two points.

strange_loop wrote on 2023-09-10, 20:45:

Well, since I'm waiting for the IDE CF adapter for the compaq, and since I'm also starting to get into the swing of things (it's […]
Show full quote

Well, since I'm waiting for the IDE CF adapter for the compaq, and since I'm also starting to get into the swing of things (it's sort of starting to come back to me, now), I thought I would turn my attention to my late Windows 98 machine which will probably be the one that sees the most use, realistically.

I was lucky enough to find a machine with the kind of case I'm after, an Athlon XP, Radeon 7500 64mb, 512mb RAM, and what I think is a good motherboard from the time (MSI KT3). I made my windows 98 boot floppy and burned the CD, put a hard drive in and - amazingly, had a fresh copy of Windows on in an hour or so and all working really nicely. Then I decided to start putting some of the bits in that I've saved over the years, starting with my audigy platinum (the breakout box looking great in a beige case). Card went in, system booted, installing drivers there was a error that caused a restart then a failed boot (I can't remember the message now). Powered down, did a quick visual check and then pushed the power button. . . then nothing. Completely dead. I tried a couple of other PSUs, nothing with them, either.

Looking closely, I eventually saw that one of the northbridge fan wires was broken. Could this have cooked the northbridge and killed the motherboard?

Really disappointed with this one. I couldn't believe my luck with how well it was going and it looked to be my perfect 1999-2001 machine, all ready for a voodoo. I hope it's salvageable, doubt I'll easily find exactly the same board again.

Thanks to the group, sorry I'm inundating with topics/questions!

These motherboards have capacitors that are basically capacitor plaguebearers, I would be AMAZED if it actually booted without a recap.

FWIW I have multiple socket 462 CPUs here from kt266/kt3 mobos, and most of the time it's been southbridge failure (can't load BIOS) ~ if you've got nothing in the board but CPU and at powerup don't hear beeps complaining about no ramfield, it's a sign the southbridge is cactus.

Thanks again, I'll do some more checks later.

My multimeter usually is on a metal surface and belongs in the garage, hence the dirt and my being used to working on conductive surfaces! I'll need to pick up a new one for cleaner/more precise work.

Any opinions on a GIGABYTE GA-ZN400 Pro 2, nForce board, do they have any particular reputation for longevity?

strange_loop wrote on 2023-09-11, 09:35:

Thanks again, I'll do some more checks later.

My multimeter usually is on a metal surface and belongs in the garage, hence the dirt and my being used to working on conductive surfaces! I'll need to pick up a new one for cleaner/more precise work.

Any opinions on a GIGABYTE GA-ZN400 Pro 2, nForce board, do they have any particular reputation for longevity?

It doesn’t matter if your multimeter is normally stored on a metal table, drawer etc.! This isn’t a clean room type of stuff, my friend 🙂Just don’t make measurements on the MB on a conductive surface or on the surface which has lots of metallic, conductive crap on it.

But if you are still going to buy a new multimeter (which I recommend), get one which has at least a continuity test.

strange_loop wrote on 2023-09-11, 09:35:

Thanks again, I'll do some more checks later.

My multimeter usually is on a metal surface and belongs in the garage, hence the dirt and my being used to working on conductive surfaces! I'll need to pick up a new one for cleaner/more precise work.

Any opinions on a GIGABYTE GA-ZN400 Pro 2, nForce board, do they have any particular reputation for longevity?

I've never had that particular motherboard, but my experience with Gigabyte motherboards has been good, and NF2 is good, though I understand that Via is better for Win98 compatibility. It will almost certainly need the capacitors replaced.

I had an MSI K7N2 from when I upgraded my former employers' personal computer to a Core 2, and it seemed fine. But it eventually began to POST erratically, until It finally died. My understanding on looking around was that it was most likely one of the casualties of the switch to lead free solder, combined with heat and poor case cooling.

[/quote]

I've never had that particular motherboard, but my experience with Gigabyte motherboards has been good, and NF2 is good, though I understand that Via is better for Win98 compatibility. It will almost certainly need the capacitors replaced.

I had an MSI K7N2 from when I upgraded my former employers' personal computer to a Core 2, and it seemed fine. But it eventually began to POST erratically, until It finally died. My understanding on looking around was that it was most likely one of the casualties of the switch to lead free solder, combined with heat and poor case cooling.
[/quote]

Ah, yeah, looks like the nForce is a bit too modern to go with the Voodoo 3 route. I have always had good experience with gigabyte boards, too.

Is it almost invariably true that simply with their age now 20-25 year old boards will need capacitors replaced?

strange_loop wrote on 2023-09-11, 13:10:

Is it almost invariably true that simply with their age now 20-25 year old boards will need capacitors replaced?

There was a time frame where capacitors tended to be less good then before and after...
But not only the factor quality is important.... Was it used a lot? Did the system run them hot? ...
All this together formed a time period where caps tended to explode a lot... Especially in hot systems you use every day.... like PCs, Monitors, Test Equipment and TVs.

I have at least 10 Mainboards in my repair pile with exploded/leaked/shorted capacitors.... and not enough time 😁
But i also have pefectly working 286 386 and 486 systems made way before the boards that have failed (Pentium 3,4 time frame).
Another but: The earlyer systems had different problems like Vartas 😉