Old Socket A motherboard won't quite boot, out of ideas, need some help.

[Nuke Dukem]

Hello, guys, I'm in need of some fresh ideas

The mobo (or rather, mofo) is my old Socket A Gigabyte GA-7DX rev 4.0 with my old Athlon XP 1700+ inside, I remember it was working fine some years ago when I put it in a box and stored it away. Last night I've been trying to get it to boot properly, but so far nothing has helped.

What it does:

• When trying to boot all fans start spinning, the DIMM LED lights up, the CPU heatsink heats up slightly (after a while), but the board can't actually POST, it just emits 1 long beep every 3 seconds;

What I've done so far:

• Did a bit of research online, found different answers/guesses as to what exactly the long beeps mean (mostly CPU faults); Manual says it's Award BIOS, sticker on the board says PhoenixBIOS D686 (same thing, I know), anno 1998;
  
• Inspected the board - it's as good as new - no bad capacitors, no physical damage, no nothing;
  
• Cleared CMOS - no change;
  
• Removed the battery, still no boot (battery is at 2.82V, don't know if it's enough but it sounds alright to me);
  
• Removed all memory - sound changes to 3 clicks per second;
• Moved RAM sticks around, single and dual - still long beeps;
• Tested RAM on another machine - RAM is fine;
• Played around with the DIP switch - no change;
• Put the graphics card in another motherboard - some artifacts, but at least it lets you know you've booted successfully;
  
• Tried both with and without a graphics card in the AGP slot - no change;
• Tested two different PSUs - no change;
• Removed the CPU heatsink, inspected the CPU - no damage, put on new thermal paste just in case, cleaned the socket and pins from dust - still no dice;
• The board is being tested while on cardboard surface, no way something is shorting.
• No CD/HDD/Floppy or anyhing else attached. Bare MoBo/CPU + RAM/VGA configurations being tested.

So now I summon you, self-proclaimed Gods of teh Interwebz, to help me beat some life into this fucker... or just announce the time of death and throw it in the bin.

I know what you're probably thinking - why even bother, it's slower than a phone, it's not worth it, just give up - yes, it would be the smart thing to do. But I'm not smart. I'm nostalgic and stubborn. I want to play some Max Payne 1&2, NFS Underground 1&2, Half-Life 1&2, GTA 3/VC/SA (and so on...) on my old hardware, complete with some CRT goodness. And judging by what's happened so far, my masochism marathon has already begun ;D

 

[Steevo]

Pull the CPU and clean pins and socket, pull the board and make sure nothing is bad on the backside of the board, attach nothing but CPU and cooler and see what you get.

 

[Nuke Dukem]

Pull the CPU and clean pins and socket, pull the board and make sure nothing is bad on the backside of the board, attach nothing but CPU and cooler and see what you get.

Already cleaned the CPU pins and socket, forgot to mention. Been testing the board on top of a cardboard box - no shorting possible. Nothing but CPU and RAM is on it.

Will edit the first post.

 

[dorsetknob]

don't suppose you have a spare CPU to try ?
Capacitors may Be kaput ( not swollen or leaking but Dried out from old age)

 

[itsakjt]

The way a video card is tested before POST can also be the problem. You checked the video card and it is giving artifacts. So something is definitely wrong with it and the motherboard BIOS is programmed such that it will check the graphics card "more thoroughly". I know a similar problem like yours. A friend with an artifacting HD 6850 GPU came to me. In my system, it would pass the POST, enter BIOS and even go to safe mode. In his system, it would not even pass the POST.
Since your system correctly gives the memory not found beeps without RAM, I suppose the CPU is ok.
I think POST as:
1. Check CPU/chipset
2. Check RAM
3. Check video card.

So I would suggest you to check using a different video card.

 

[DeathtoGnomes]

Gonna have to try another CPU

OR

You could try to contact Gigabyte and get the pin-outs for the CPU socket and check the voltages which would tell if the capacitors dried out as @dorsetknob thinks is possible.

The only other thing possible is a FUBAR BIOS chip.

 

[biffzinker]

Wouldn't hurt to replace the CMOS backup battery (CR2032.) The backup battery should be supplying 3.30V with power removed from the mobo so 2.82V is on the low side.

 

[natr0n]

It's either the ram or gpu not seated correctly.

wiggle them maybe

 

[TheoneandonlyMrK]

The way a video card is tested before POST can also be the problem. You checked the video card and it is giving artifacts. So something is definitely wrong with it and the motherboard BIOS is programmed such that it will check the graphics card "more thoroughly". I know a similar problem like yours. A friend with an artifacting HD 6850 GPU came to me. In my system, it would pass the POST, enter BIOS and even go to safe mode. In his system, it would not even pass the POST.
Since your system correctly gives the memory not found beeps without RAM, I suppose the CPU is ok.
I think POST as:
1. Check CPU/chipset
2. Check RAM
3. Check video card.

So I would suggest you to check using a different video card.

Exactly this one long beep is normally no GPU detected in that era board.

 

[Derek12]

As said try another video card
Clean the AGP slot on motherboard too and check if there is damage on its pins.
Check nearby mosfets/tracks/SMD components

 

[TheoneandonlyMrK]

As said try another video card
Clean the AGP/PCI slot on motherboard too and check if there is damage on its pins.
Check nearby mosfets/tracks/SMD components

If it needs power then make sure it's plugged in right too.

 

[fritoking]

Of you need a new board I may be able to help

 

[Nuke Dukem]

don't suppose you have a spare CPU to try ?
Capacitors may Be kaput ( not swollen or leaking but Dried out from old age)

No spare CPU, I suppose I could find something for around $2-3 on my local second-hand market. That, or find someone down for tests. Now, about the capacitors - yeah, I'm aware they might sometimes appear ok but actually have gone rogue, but between the amount of time I spent soldering a couple of capacitors on some board last time and the fact that I count no less than 41 of these bastards on this board, the hairs on my lower back begin to stand up.

The way a video card is tested before POST can also be the problem. You checked the video card and it is giving artifacts. So something is definitely wrong with it and the motherboard BIOS is programmed such that it will check the graphics card "more thoroughly". I know a similar problem like yours. A friend with an artifacting HD 6850 GPU came to me. In my system, it would pass the POST, enter BIOS and even go to safe mode. In his system, it would not even pass the POST.
Since your system correctly gives the memory not found beeps without RAM, I suppose the CPU is ok.
I think POST as:
1. Check CPU/chipset
2. Check RAM
3. Check video card.

So I would suggest you to check using a different video card.

I don't know, it could be that. Same as with the CPU, this is something I might find cheap or get someone willing to lend stuff for testing. Doesn't sound likely, but at this point, who knows.

Gonna have to try another CPU

OR

You could try to contact Gigabyte and get the pin-outs for the CPU socket and check the voltages which would tell if the capacitors dried out as @dorsetknob thinks is possible.

The only other thing possible is a FUBAR BIOS chip.

Hmm, as in run the board with no CPU and stick the probes down the socket holes? I like it. Sounds elegant.

Wouldn't hurt to replace the CMOS backup battery (CR2032.) The backup battery should be supplying 3.30V with power removed from the mobo so 2.82V is on the low side.

One of my newer boards will act as a donor. Tomorrow.

It's either the ram or gpu not seated correctly.

wiggle them maybe

It's often the case that something as silly as this is the culprit. Knowing that, I must have changed the DIMMs over 50 times during testing yesterday and some more today, I reckon this is as good a wiggle as they're going to get. I also manually held the graphics card at different angles just to try that. No luck so far.

Exactly this one long beep is normally no GPU detected in that era board.

Now this sounds nice. Concrete and nice.

As said try another video card
Clean the AGP slot on motherboard too and check if there is damage on its pins.
Check nearby mosfets/tracks/SMD components

I looked at the slot yesterday, everything in there is peachy as far as I'm concerned. Clean and straight. Blew some air in just in case.

If it needs power then make sure it's plugged in right too.

It's a GeForce 2 Ti, the cooler on it is the size of the chipset cooler on the motherboard. Simply adorable. No additional power needed.

Of you need a new board I may be able to help

You have one of the same model? I appreciate your willingness, however shipping costs from Ohio to Bulgaria may be a bit of a bitch

I like where this is going. Can't guarantee I'm gonna find another GPU for testing first thing tomorrow, but I'm definitely gonna look into that. Thanks for the brainstorming so far, guys. I will write back.

 

[fritoking]

Ha! Sorry , didn't realize you were in Bulgaria... I actually have a couple of socket A boards here with cpus

 

[qubit]

Looks like you've been pretty thorough with the troubleshooting. Only thing I can suggest is to replace that cmos battery with a fresh one and especially the graphics card with a properly working one. If that doesn't work, then it's probably toast.

Besides dry capacitors, it's quite possible that something got corroded right inside where you can't see which is causing this problem, or a component has aged and gone out of spec.

Tagging @eidairaman1 for more brainstorming action.

 

[Jetster]

No dip switches?

 

[Bones]

With some boards the repeated long beep first described in your post is a RAM error alarm that's typical with many of them. I have one that's very similar if not the same related to the board model.
For testing set the dipswitches to have it run at 100MHz.

Try "Rocking" the CPU in the socket - That means moving the CPU locking lever up and down several times to help clean off the contact points in the socket and the pins of the CPU. Do this for at least a minute before setting the cooler back on to test, same thing with the RAM and GPUs too.

Since the board had sat up for so long the most probrable thing is simple corrosion somewhere, that's why I suggest rocking the CPU and repeatedly moving the RAM sticks up and down, not a side-to-side wiggle - That just makes the contact points in the slots loosen up and can cause a bad contact spot(s). Always do it the way the sticks go in and out and it's the same with the GPU.
Check and be double-sure your CMOS jumper is in the correct position.
Also look into the ATX PSU header to be sure nothing looks out of place/dirty there too, if it has the extra 4 pin header for the CPU you'd need to do the same with it.

As suggested check the caps themselves, ANY caps with a bulge are bad, even if the bulge is barely noticeable.
If it still doesn't work after all that, if you feel like tinkering with it you could reflow the legs of the caps with a soldering pen/station if you have one to use. Believe it or not I fixed one of my "Dead" AN7's by simply removing and reinstalling the caps, none were replaced at all and it's doing fine, was throwing the dreaded 90 code before I did that to it.

Good luck!

 

[jaggerwild]

Find someone with the same board, ask them to hot swap your Bios chip in to reflash it(cheaper then buying a new one) Ive done it a few times.

 

[Nuke Dukem]

Ha! Sorry , didn't realize you were in Bulgaria... I actually have a couple of socket A boards here with cpus

I will be sure to snag them if I ever drop by

Looks like you've been pretty thorough with the troubleshooting. Only thing I can suggest is to replace that cmos battery with a fresh one and especially the graphics card with a properly working one. If that doesn't work, then it's probably toast.

Besides dry capacitors, it's quite possible that something got corroded right inside where you can't see which is causing this problem, or a component has aged and gone out of spec.

I'll get to the battery and graphics card in a bit. I also had the thought of something quietly rotting on the inside, although it has been stored like medicine - in a cool and dry cabinet.

No dip switches?

I did mention that in the first post, maybe you missed it. Yes, there's a DIP switch. I did some fiddling. Some more today. No change.

With some boards the repeated long beep first described in your post is a RAM error alarm that's typical with many of them. I have one that's very similar if not the same related to the board model.
For testing set the dipswitches to have it run at 100MHz.

Try "Rocking" the CPU in the socket - That means moving the CPU locking lever up and down several times to help clean off the contact points in the socket and the pins of the CPU. Do this for at least a minute before setting the cooler back on to test, same thing with the RAM and GPUs too.

Since the board had sat up for so long the most probrable thing is simple corrosion somewhere, that's why I suggest rocking the CPU and repeatedly moving the RAM sticks up and down, not a side-to-side wiggle - That just makes the contact points in the slots loosen up and can cause a bad contact spot(s). Always do it the way the sticks go in and out and it's the same with the GPU.
Check and be double-sure your CMOS jumper is in the correct position.
Also look into the ATX PSU header to be sure nothing looks out of place/dirty there too, if it has the extra 4 pin header for the CPU you'd need to do the same with it.

As suggested check the caps themselves, ANY caps with a bulge are bad, even if the bulge is barely noticeable.
If it still doesn't work after all that, if you feel like tinkering with it you could reflow the legs of the caps with a soldering pen/station if you have one to use. Believe it or not I fixed one of my "Dead" AN7's by simply removing and reinstalling the caps, none were replaced at all and it's doing fine, was throwing the dreaded 90 code before I did that to it.

Good luck!

Dip switches were tested at multiple positions, 100 MHz was covered - no change. I rocked the socket like a hurricane (360+ times - yes, I counted them), both with the CPU in and out - no change. RAM and AGP slots are positively clean - if there ever was anything in the way at the start, all the friction from installing and removing and blowing would have surely removed it by now. I only wiggled the AGP card slightly, I don't intend to bend the contact points. The CMOS Reset jumper is in its correct position. The ATX header is clean, plus the board lacks a 4-pin CPU header altogether. All caps appear perfect. I might resolder ot replace them, but only as a last resort, cause it would take me a lot of time. Thanks for listing all the stuff

Find someone with the same board, ask them to hot swap your Bios chip in to reflash it(cheaper then buying a new one) Ive done it a few times.

As adventurous as that sounds, it would be easier to just find someone who does BIOS programming and flash the chip there. I can't figure out how I'd remove the chip. Only some fancy IR station comes to mind.

So, folks, today the board saw the following:

• Replaced the battery with the next best one I could find, running at 2.98V - that changed nothing;
• Installed another graphics card (one that I was told was 99% guaranteed OK) - no go; Same guy is going to lend one that is 100% in the following days;
• No CPU in the socket = fans spin, no clicks or beeps from the speaker;
• Some more RAM swaps, switches and substitutions were performed, the only times the dreaded beeps were gone were when a DIMM was not properly seated (apparently), but no good boot to speak of;
• DIP switch got its fair share of further tickling, nothing new;

What I'm contemplating right now - finding out how to test the voltage on the CPU/RAM/AGP slots. I think it's the most empirical and accurate thing to do now.

Thanks for your support so far. Keeps me going!

 

[Bones]

The BIOS chip to your board is removable (Looks like it is from what I dug up) so getting it out shoudn't be a huge issue.
Instead of finding someone to program the chip you can get one of these http://www.ebay.com/itm/NANO-USB-Pr...593344?hash=item3f2b8ce400:g:B6QAAMXQrNtR0sQZ and do it yourself easily.
I have one and it's a life-saver, easy to use and the software it need is in the listing too, just have to scroll further down for the links. Extremely useful item and once you learn how to use it you won't regret getting one.
As it is in the board make sure the little dot on the beveled end is aligned with the arrow in the socket and flash away - No floppy to deal with and no need to worry about switches in the flashing routine including what's used to erase a chip, the software does it all.

 

[Melvis]

Is the motherboard grounded? like in a case?

 

[Derek12]

[...]

• Replaced the battery with the next best one I could find, running at 2.98V - that changed nothing;
• Installed another graphics card (one that I was told was 99% guaranteed OK) - no go; Same guy is going to lend one that is 100% in the following days;
• No CPU in the socket = fans spin, no clicks or beeps from the speaker;
• Some more RAM swaps, switches and substitutions were performed, the only times the dreaded beeps were gone were when a DIMM was not properly seated (apparently), but no good boot to speak of;
• DIP switch got its fair share of further tickling, nothing new;

What I'm contemplating right now - finding out how to test the voltage on the CPU/RAM/AGP slots. I think it's the most empirical and accurate thing to do now.

Thanks for your support so far. Keeps me going!

So you don't get any beeps without RAM? or the pattern changes?
I think the CPU and therefore, its power circuits, is working, hence the beeps, I think it could be...

Broken trace in the middle layers of the PCB (not fixable), probably going into the AGP slot
Bad soldier somewhere in the motherboard (unless it's a thru-hole component or non-BGA, it's not fixable
Bad chipset (non fixable unless you can resolder or replace it)
Some bad transistor/MOSFET/capacitor (fixable).

 

[DeathtoGnomes]

So you don't get any beeps without RAM? or the pattern changes?
I think the CPU and therefore, its power circuits, is working, hence the beeps, I think it could be...

Broken trace in the middle layers of the PCB (not fixable), probably going into the AGP slot
Bad soldier somewhere in the motherboard (unless it's a thru-hole component or non-BGA, it's not fixable
Bad chipset (non fixable unless you can resolder or replace it)
Some bad transistor/MOSFET/capacitor (fixable).

if it is bad solder joint, someone around TPU has baking experience.
(I forgot who.)

 

[Derek12]

if it is bad solder joint, someone around TPU has baking experience.
(I forgot who.)

True, I forgot about baking/reballing them

 

[biffzinker]

someone around TPU has baking experience.
(I forgot who.)

@Mussels?