Dead mobo?

cdabc123

Supreme [H]ardness
Joined
Jun 21, 2016
Messages
4,475
Last night my pc just shutoff completely, would not restart after that.

Specs:
Asus X99 deluxe motherboard
Es 10 core 10 thread cpu at 3.3ghz
Supermicro 560w server psu
4x8 ddr4
Custom water loop
980ti

So far ive tried:
Different psu (the psu will turn on for a half second then shut off like short circuit protection)

Removed board from case
removed all perpetual
Removed all but 1 ram stick
Reseated the CPU

There is no obvious damage on the board or socket/chip. Is it fair to think just the motherboard is dead? I use the box for esxi, gaming, and workstation stuff and its been pretty decent. Should I just buy a new x99 board or are there decent deals out there for upgrades to this setup?
 
Last edited:
only other thing to try that i dont see listed and really dont think will help in this case, is to remove the battery, clear the cmos good and put a new battery in. other than that i think youve covered everything and yup, shes dead jim. could try one of those $200 chinese board on newegg ;) or check fs/t maybe put up a wtb. otherwise, youre looking at board/chip/ram to upgrade...
 
Op made more clear for board.


only other thing to try that i dont see listed and really dont think will help in this case, is to remove the battery, clear the cmos good and put a new battery in. other than that i think youve covered everything and yup, shes dead jim. could try one of those $200 chinese board on newegg ;) or check fs/t maybe put up a wtb. otherwise, youre looking at board/chip/ram to upgrade...
Tried the cmos reset, no beans.

I do see the chinese boards on ebay as well as workstation boards for ~100 but I would really like to do this in the cheapest way possible. This rig was built out of spare parts.

I may try hitting up the fs fourm for either just a x99 board or a board plus ram.

What 8core+ combos are pretty cheap these days?
 
Last edited:
Check for a short on the EPS and ATX hot pins?

On the EPS connector, the 12V hot pins are the ones closest to the latch. You should have a few thousand ohms between them and ground. It's more complicated on the ATX connector, but I'd suspect the EPS input rail first anyway.

I've actually got a working Gigabyte X99 board lying around that I might consider trading for your dead one, if that's something you're interested in. A motherboard necromancy thread might be a good time.
 
Check for a short on the EPS and ATX hot pins?

On the EPS connector, the 12V hot pins are the ones closest to the latch. You should have a few thousand ohms between them and ground. It's more complicated on the ATX connector, but I'd suspect the EPS input rail first anyway.

I've actually got a working Gigabyte X99 board lying around that I might consider trading for your dead one, if that's something you're interested in. A motherboard necromancy thread might be a good time.
There is a short on the eps connector, I pulled the heatsink off the main power stage and did not see any damage.

Let me know what ya would say next steps are.

I may be able to throw in a few dead gpus (of varying degrees) if a trade is in order
 
Can you post a photo of that part of the board? Edit: The more detailed the better.
 
Phone camera is unfortunately not great,
Pn # ir3550-ir3553
20220127_154605_HDR.jpg20220127_154713.jpg


Capacitor in the middle is raised in comparison to the rest and the leads are not directly connected to the pads with solder, I just resoldered those two pins to ensure a connection.

Also worth noting the computer was not under any heavy load when it died
 
Phone camera is unfortunately not great,
Pn # ir3550-ir3553
View attachment 437447View attachment 437448


Capacitor in the middle is raised in comparison to the rest and the leads are not directly connected to the pads with solder, I just resoldered those two pins to ensure a connection.

Also worth noting the computer was not under any heavy load when it died
Which capacitor are you referring to? The little MLCC next to PQ1003 (which is probably actually refers to the power stage)?

Edit: When you say there's a short on the EPS pins, what is the actual resistance to ground you have there? Are we talking 0.0 ohms, or just really low, like 10 ohms or something?
 
Which capacitor are you referring to? The little MLCC next to PQ1003 (which is probably actually refers to the power stage)?

Edit: When you say there's a short on the EPS pins, what is the actual resistance to ground you have there? Are we talking 0.0 ohms, or just really low, like 10 ohms or something?
0 raising to 500ohms, my multimeter isn't very good so I'm unsure if the low resistance at the beginning is caused by that.

The mobo will let the psu turn on with the eps connector unplugged. (And showed lights, and 00 on the 7 segment display) the boot hangs of course.

The cap is the ones to the left of pq1003. The 5th from the top.
 
0 raising to 500ohms, my multimeter isn't very good so I'm unsure if the low resistance at the beginning is caused by that.
Does that change if you remove the CPU from the socket?

Also, it's still not clear to me which cap you're referring to. Can you circle it in the picture?
 
Ah my apologies those numbers were with the CPU out.

With the CPU in it reads ~2ohms continually.


This pic should make the cap more clear its the raised one.

20220127_165801_HDR.jpg
 
Abit unrelated but I figured I would post my temp "desktop" replacement its a Pico itx, 4 core Celeron at a blistering 1.83ghz. 2gb of ram and a msata ssd. Surprisingly tolerable under centos.

Should I water cool it? :p

Temp vs old.
20220127_172011_HDR.jpg460577_Snapchat-387347143.jpg
 
Last edited:
Ah my apologies those numbers were with the CPU out.

With the CPU in it reads ~2ohms continually.


This pic should make the cap more clear its the raised one.

View attachment 437478
If it's still soldered on there, it's probably just sloppy manufacturing.

Based on how the resistance changes when you remove the CPU, it sounds like you have a power stage where the high side FET failed, which is pretty common on hardware from this time period. You could try using voltage injection to figure out which one it is. Solder a wire to one of the coil negative terminals (the side farthest from the power stages) and then inject 12V at the EPS pins with a bench power supply with the CPU removed. The failed part will likely get hot enough to detect with isopropanol or freeze spray.

Edit: I guess a failed cap is still a possibility too, but you'd figure that out the same way. Inject 12V and see where the energy goes.
 
Actually go read the post. He stated "X99 deluxe motherboard" .... which is why I did not assume it was an Asus, and I asked.
I did. ASUS is the only company I know of that named things "X99 Deluxe" etc.
 
Edit: I guess a failed cap is still a possibility too, but you'd figure that out the same way. Inject 12V and see where the energy goes.
it does look like there is a ring of glue/schmoo left behind. their height is all over the place too. you are better at this though...

1643330660303.png
 
  • Like
Reactions: Azrak
like this
That cap off the board looks.... odd. There are marks on the board that make a circle outline like the cap used to be touching the board, but now isn't. I don't see those same marks around the other caps.
In fact, all of the caps have an odd look to their cans and I don't think it is a reflection. It looks like the cans have been subjected to metal erosion, like a gallium reaction.
 
One or two bad caps isn't likely to prevent the board from powering up.
 
One or two bad caps isn't likely to prevent the board from powering up.
It would if there's a big enough short through the cap to trip the power supply's over current protection. I've seen it happen with a single cracked 0803-size MLC cap, so it's not out of the question here.

As the others said, those caps do look a bit off, but he said he resoldered one, so I assume there'd be some flux residue or something. Wouldn't be hard to remove that sticky-outy one and check the resistance through it, though.
 
If it's still soldered on there, it's probably just sloppy manufacturing.

Based on how the resistance changes when you remove the CPU, it sounds like you have a power stage where the high side FET failed, which is pretty common on hardware from this time period. You could try using voltage injection to figure out which one it is. Solder a wire to one of the coil negative terminals (the side farthest from the power stages) and then inject 12V at the EPS pins with a bench power supply with the CPU removed. The failed part will likely get hot enough to detect with isopropanol or freeze spray.

Edit: I guess a failed cap is still a possibility too, but you'd figure that out the same way. Inject 12V and see where the energy goes.

Alright so I soldered a wire to the coil and ended up putting a regulated 10a through the board. With alcohol or my finger I didnt notice any part heating up.

Should I be trying something else for the voltage injection? Does it just need quite abit of power to notice anything.

Is it possible to inject a frequency and probe the fet side of the coils. With a scope to notice which one is dead?
 
it does look like there is a ring of glue/schmoo left behind. their height is all over the place too. you are better at this though...

View attachment 437489

Radial electrolytics are almost never glued, and when they are, it doesn't look like that.

The residue under the capacitor is most likely leaked electrolyte. Radial capacitors lift off the board like that when they fail internally and produce hydrogen gas that can push the cap off its leads if it can't vent anywhere else. Those capacitors don't have vents on the top, so the only place for the pressure to go is down out the rubber seal on the bottom, or to push the legs out of the body of the capacitor. Normally, capacitors around the CPU socket on higher end boards are polymer types, but they do make electrolytics without vents to make a capacitor look more expensive than it is.

If that capacitor failed, it could be shorted out, or it could have caused excessive current draw on the power regulation circuitry and killed one of them. I would suggest replacing all of the capacitors around the CPU to avoid chasing ghosts.
 
Radial electrolytics are almost never glued, and when they are, it doesn't look like that.

The residue under the capacitor is most likely leaked electrolyte. Radial capacitors lift off the board like that when they fail internally and produce hydrogen gas that can push the cap off its leads if it can't vent anywhere else. Those capacitors don't have vents on the top, so the only place for the pressure to go is down out the rubber seal on the bottom, or to push the legs out of the body of the capacitor. Normally, capacitors around the CPU socket on higher end boards are polymer types, but they do make electrolytics without vents to make a capacitor look more expensive than it is.

If that capacitor failed, it could be shorted out, or it could have caused excessive current draw on the power regulation circuitry and killed one of them. I would suggest replacing all of the capacitors around the CPU to avoid chasing ghosts.
The cap isnt swollen on the bottom. it looks completley normal. however the legs wernt fully soldered to the board which may have been causing a issue with loading of the fets. Either way, although ive been meaning to buy a hot air rework station I have yet to so I wouldnt be able to touch much of the power stage on this board~.

~the quest for a new mobo begins~
 
Indeed, would have given you a great deal. They are expensive on ebay.
Lol I've noticed. Sadly no hits on my wtb post. I may have to bite on the add posted for a 3900x combo. Wouldn't be a bad upgrade just way out of my budget
 
Lol I've noticed. Sadly no hits on my wtb post. I may have to bite on the add posted for a 3900x combo. Wouldn't be a bad upgrade just way out of my budget
I seem to have an Asus X99-A exhibiting the same behaviour whereby the mobo runs if the EPS x8 +12V connector is disconnected but will only power up for 1sec if left attached.What was the outcome in the end for you ? You fixed it?
 
I seem to have an Asus X99-A exhibiting the same behaviour whereby the mobo runs if the EPS x8 +12V connector is disconnected but will only power up for 1sec if left attached.What was the outcome in the end for you ? You fixed it?
I traded some other hardware for a 3900x combo and have been generally enjoying it (except the x99 system was abit more friendly under esxi)

The limit of my trouble shooting came down to assuming one of the power stages was dead but stoping there as I dont have a hot air rework station to do much if I found out which.
 
I traded some other hardware for a 3900x combo and have been generally enjoying it (except the x99 system was abit more friendly under esxi)

The limit of my trouble shooting came down to assuming one of the power stages was dead but stoping there as I dont have a hot air rework station to do much if I found out which.
Did some "interesting" investigations on my "recently dead" ASUS X99-A motherboard, ref. the 1-2sec power cycling problem described above:

It appears that the culprit is actually the circuit built around the < 3V3 downto 1V8 regulator > chipset (U501), (measured to provide +1V8 rail to the X99 chipset ???) -

*Without* the 8-pins E-ATX +12V being present:
(U501) reaches crazy temps of +120C on my motherboard after a few minutes once the +3V3 rail is made available as part of the standard ATX-24pins connector-
At that point the motherboard will not restart but the regulator is well beyond its stress point...

*With* the E-ATX +12V (x8 connector) connected , what happens is that there is what seems to be a temperature sensing that gets enabled (versus without the EATX +12V present) -
When U501 reaches +85C, the motherboard reboots ...In doing so U501 cools down a few degrees until it become loaded one more and reaches +85C once more and the cycle repeats.

So the following above explains why the mobo cycles with the EATX connector fitted and why it does not without.
See link for mp4 video ( video showing U501 temp triggers power cycling) showing the motherboard being maintained ON (look at 7-segment at bottom showing "00") while either the +12V E-ATX is not connected or if connected while the temp is maintained <+85/86C approx.


Before getting to that point I had also check the x8 switchers off the E-ATX connector (underneath heatsink) and all are sound and switching correctly to +1.7V
Power draw is about 0.7A@12V for all those 8 switchers going concurrently (no CPU in the socket) - No short circuits etc...

So it seems that U501, X99 chipset regulator circuit is somehow at fault here.
I did not get to the point of finding out if possibly the X99 chipset itself is faulty (unlikely ?) or if there is some bad/dry joints somewhere.
No short circuits on the output of U501 caps that I could see.
I decided to bite the bullet and buy a second hand replacement of the same X99-A (not that cheap) as I need to bring what is my main computer back to life.

Until then I will attempt to revive this "dead" one...
 

Attachments

  • HAFS6039.JPG
    HAFS6039.JPG
    1 MB · Views: 0
  • DURW4881.JPG
    DURW4881.JPG
    1.3 MB · Views: 0
  • ECAZ5754.JPG
    ECAZ5754.JPG
    1.5 MB · Views: 0
  • YFLJ9633.JPG
    YFLJ9633.JPG
    1 MB · Views: 0
  • MLLN9058.JPG
    MLLN9058.JPG
    1,008.3 KB · Views: 0
  • IMG_20220907_193627.jpg
    IMG_20220907_193627.jpg
    785.5 KB · Views: 0
  • IMG_20220907_163507.jpg
    IMG_20220907_163507.jpg
    530 KB · Views: 1
Last edited:
Did some "interesting" investigations on my "recently dead" ASUS X99-A motherboard, ref. the 1-2sec power cycling problem described above:

It appears that the culprit is actually the circuit built around the < 3V3 downto 1V8 regulator > chipset (U501), (measured to provide +1V8 rail to the X99 chipset ???) -

*Without* the 8-pins E-ATX +12V being present:
(U501) reaches crazy temps of +120C on my motherboard after a few minutes once the +3V3 rail is made available as part of the standard ATX-24pins connector-
At that point the motherboard will not restart but the regulator is well beyond its stress point...

*With* the E-ATX +12V (x8 connector) connected , what happens is that there is what seems to be a temperature sensing that gets enabled (versus without the EATX +12V present) -
When U501 reaches +85C, the motherboard reboots ...In doing so U501 cools down a few degrees until it become loaded one more and reaches +85C once more and the cycle repeats.

So the following above explains why the mobo cycles with the EATX connector fitted and why it does not without.
See link for mp4 video ( video showing U501 temp triggers power cycling) showing the motherboard being maintained ON (look at 7-segment at bottom showing "00") while either the +12V E-ATX is not connected or if connected while the temp is maintained <+85/86C approx.


Before getting to that point I had also check the x8 switchers off the E-ATX connector (underneath heatsink) and all are sound and switching correctly to +1.7V
Power draw is about 0.7A@12V for all those 8 switchers going concurrently (no CPU in the socket) - No short circuits etc...

So it seems that U501, X99 chipset regulator circuit is somehow at fault here.
I did not get to the point of finding out if possibly the X99 chipset itself is faulty (unlikely ?) or if there is some bad/dry joints somewhere.
No short circuits on the output of U501 caps that I could see.
I decided to bite the bullet and buy a second hand replacement of the same X99-A (not that cheap) as I need to bring what is my main computer back to life.

Until then I will attempt to revive this "dead" one...
Nice would be interested to see if you go much farther with the trouble shooting.

Ya unfortunately thats what pushed me to the am4 setup. X99 really hasnt come down to a reasonable price.
 
Found the .fz BoardView file with details of <U501> (RT8065ZQW) and the components around it.
https://drive.google.com/drive/folders/15N_Q919Iew9r8Cs6J0x_Qy_QPLnOkcE7?usp=sharing
Hopefully will be able to make sense of the information to carry the repair...

And I did fortunately.
It appears that U501 is actually a DC/DC switcher that would normally regulate down to +1V05 for the MEmory (1.05ME reference in the boardview file)
No switching going on at the <LX> pin so I have ,what simply seems to be, a dead IC <U501> = RT8065ZQW
Verified levels against boardview file and put together a quick schematic with position more or less matching component placement to quickly verify the findings.

All other inputs appear correct (VIN=5V , EN=3V3 , current Feedback loop values match the current incorrect reading on the output etc....)

Digikey has it for £0.81 : Digikey RT8065ZQW
Ordered it and should be able to have it swapped next week without too much trouble - That would be a dirt cheap fix.
Might have to return the board I ordered yesterday if all goes well.

UPDATE: Got U501 replaced and got the correct voltage of < +1V05> out of it (now switching correctly ): the motherboard no longer power cycles like it used to ,so it has fixed this part.
However, the BIOS is not starting and the motherboard stays on Q-code "00" which tells me the BIOS is not running...At that point, not seeing any other "hot" spots under the thermal camera I called it a day as I think it might have suffered some damages somewhere else of the board, during the time <+1V8> was coming out of <U501> instead of the expected <+1V05>
My guess is possibly damages to the X99 chipset.
I migrated CPU+DDR to a second hand <X99-A II> and in that way regained access to my main computer usage.

It is a shame really..but hey. I tried. IT could have been just that switcher to replace.
 

Attachments

  • 1V05_MEMORY.PNG
    1V05_MEMORY.PNG
    1.5 MB · Views: 0
Last edited:
Back
Top