EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: ThingWizard on April 03, 2021, 11:59:39 am
-
Hi!
All of a sudden this morning my computer did not wake from sleep, well it actually woke but there was no video. I tried the graphics card in another computer, same thing. No fan spin no video.
Ok, so I removed the heatsink from the graphics card and found a small black dot on the heat pad just over the vrms. Looks like something got hot.
I took my usb microscope out and saw this:
(https://i.ibb.co/KNrNwq6/WIN-20210403-13-28-23-Pro.jpg) (https://ibb.co/KNrNwq6)
That capacitor (named C178) looks a bit sad, like something may have sparked here. I tried cleaning it up and re-flow it but the problem still remains.
The problem area is located here (red Arrow):
(https://i.ibb.co/Wy7W4WP/here.png) (https://ibb.co/Wy7W4WP)
SO MY QUESTION: If the fan does not spin, I guess it is not getting power, but does the fact that the fan does not spin help me in troubleshooting, and if so in what way? - And, do we know where in the circuit that capacitor is. Are there schematics for these boards?
MY NEXT PLAN is to remove the heatsink and read the voltages coming out of the different rails in the VRM section to try to find the problem. A problem is I don't know what the voltages should be.
Any help really appreciated as graphics card prices are through the roof right now, and my bank account is not :P
Cheers!
-
my Zotac GTX 1060 6gb mini amp edition, the fans never spins until the card reach a defined temperature, and the speed will increase with the video load ??
the photo looks like a bad smt soldering and overheating and crust migration ??
-
That capacitor is shorted, cleaning or soldering will not help. Trying to power it may only cause more damage. It needs to be replaced or at least removed. Most likely fuse on that 12V power rail is blown as well. IIRC it's 0603 size with 4.7uF capacitance, should be rated for 16V at least. I think that part of VRM was powered from PCI-E slot with fuse located on the other side of PCB somewhere close to the slot on the end with video outputs. 1206 Size 10A fast action.
-
the photo looks like a bad smt soldering and overheating and crust migration ??
Thank you for taking time to respond, that certainly sounds like a possibility. Of course I don't know it that is the problem or not...
I cannot remember if the fan spins on idle or not, but in any case I have no video out.
[EDIT]No they apparently don't
"3D Active Fan
The semi-passive fans will remain off when the GPU is under a set loading or temperature for low power gaming. It allows gamers to enjoy gameplay in complete silence when the system is running light or idle."
https://www.gigabyte.com/fi/Graphics-Card/GV-N1070IXOC-8GD#kf (https://www.gigabyte.com/fi/Graphics-Card/GV-N1070IXOC-8GD#kf)
I'm not getting any warning beeps one usually gets with missing graphics card.
I'll report back when I have measured the voltage on the different "channels".
Also, do you recommend a "bake"? 200c for 8 minutes with a side of fries
-
I'll report back when I have measured the voltage on the different "channels".
Also, do you recommend a "bake"? 200c for 8 minutes with a side of fries
Replace the faulty capacitor and blown fuse. Not destroy extremely easily and cheaply repairable graphics card for no reason :palm: |O. That capacitor simply developed internal crack and has shorted.
-
the photo looks like a bad smt soldering and overheating and crust migration ??
The photo shows a capacitor which managed to fry for some time after shorting before fuse finally blew up.
-
That sounds credible! Thank you @wraper
I have no voltage at all from the L3 L5 L7 L9 L11 L12 coils(?)
EDIT: I found the fuse, F4 on the back side.
Yes it is blown! :) :-+
(https://i.ibb.co/R6QnJPV/WIN-20210404-20-20-17-Pro.jpg) (https://ibb.co/R6QnJPV)
Okay, so now I only need to know the value of the bad capacitor C178
-
If you want, I can mail you replacement parts for the cost of shipping and parts with no markup, since I have them. I will need to go to the post to ship some other things anyway. Should take a week or two to arrive.
-
That capacitor is shorted, cleaning or soldering will not help. Trying to power it may only cause more damage. It needs to be replaced or at least removed. Most likely fuse on that 12V power rail is blown as well. IIRC it's 0603 size with 4.7uF capacitance, should be rated for 16V at least. I think that part of VRM was powered from PCI-E slot with fuse located on the other side of PCB somewhere close to the slot on the end with video outputs. 1206 Size 10A fast action.
Oh thank you so much, I did not see this message before, must have scrolled right past it. Thank you for giving me the values for the cap and the fuse! This should be pretty straight forward then!
-
If you want, I can mail you replacement parts for the cost of shipping and parts with no markup, since I have them. I will need to go to the post to ship some other things anyway. Should take a week or two to arrive.
That would be great! I'll send you a direct message for details
-
I got the parts today and will replace them. Again thanz a million Wraper - During this time I also bought and Andonstar AD207 microscope that was very helpful de-soldering the blown components - looking forward to trying it on for size :)
Nice to have the forum back from the damp bilge of the server wreck.
-
Okiedokie, well. this was... interesting.
I replaced the components and tried the card out. Immediately fiery death of the neighbouring capacitor C177. What's the value for that one do you know Wraper?
Could it have been shorted too (Didnt think to test) or are we talking crazy voltages here because of some other reason?
Strangely enough the Fuse seems to be okay.
(https://i.ibb.co/xgvgMfC/IMG180104-041016-000048-F.jpg) (https://ibb.co/xgvgMfC) (https://i.ibb.co/wcPyXyg/IMG180104-042250-000049-F.jpg) (https://ibb.co/wcPyXyg) (https://i.ibb.co/2W77Qpr/IMG180104-043042-000050-F.jpg) (https://ibb.co/2W77Qpr)
-
All 4 capacitors are the same. Most likely it already had a hidden crack. And soldering nearby pushed it a bit further. Well, you still have 2 fuses and 2 caps left.
-
Excellent, thanx!
Yes - Luckily you sent me 3 caps, I'll try and replace that one then too with another one of them.
I checked the other two ones, they do not seem to be shortened.
-
Measure if there is no short across those capacitors when you replace a bad one. There should be something going on with those capacitors as I've seen that cap cracking on another such card. Maybe it's a mechanical stress because card is deformed in that area when cooler is mounted. They are located right next to a screw hole after all.
-
The rest checked ok, replaced the burnt one. Did a quick test, no fire and smoke and the gpu warms up. I just had it running a few seconds without heatsink, but so far so good - it seems to be alive. I think I'll be ok to runt the card for a few seconds more without heatsink to see if I get any video...
[EDIT]
Good news, I got the bios post screen on my monitor - The card seems to be alive again!
Thank you so much Wraper, I could not have done this without your help, awesome man!
-
Just an update here, I got the heatsink back on and the card installed in my computer. Right now I'm using it, and it has been going for a few hours without any problems.
What a great feeling! I will update in a couple of days when I know if it seems stable or not.
I'm thinking what might have caused this problem in the first place. Also, how can a crack in a capacitor cause it to short? I could better understand if a crack caused it to go open circuit. I tried googling SMD ceramic capacitor fail modes but I didn't really find anything good.
-
I'm thinking what might have caused this problem in the first place.
Maybe it's a mechanical stress because card is deformed in that area when cooler is mounted. They are located right next to a screw hole after all.
Also, how can a crack in a capacitor cause it to short? I could better understand if a crack caused it to go open circuit. I tried googling SMD ceramic capacitor fail modes but I didn't really find anything good.
MLCC contains many thin layers which often short together when crack appears.
(https://product.tdk.com/de/system/files/sog_mlcc02_003_en.png)
-
Dave Jones has done a couple videos on this topic. Check out there one where he goes in depth on the subject: https://www.youtube.com/watch?v=QgKY5QWehME (https://www.youtube.com/watch?v=QgKY5QWehME) (EEVblog #1037)
Cheers
-Tim
-
Thank you to everyone who contributed to this thread. I have almost exactly the same issue. My Gigabyte GTX1070 visibly blew capacitor C176.
I also checked and Fuse F4 looks to have blown.
So now due to everyone's comments I am feeling positive about replacing these two components and getting my GPU back working.
I am going to research the parts and how to solder SMD components and give it a try.
Thanks everyone.
-
Can you give me value of these five capacitors? My capacitors blown up and i must replace them all.
-
Interesting read! I'll take some notes on this in case my own gpu decides to fail.
But sadly.... you have also deprived a poor scalper of his oh so hard earned profit... :'(
Pretty soon we'll be fixing 3DFX Voodoo cards on this forum... |O
-
Hi all. Looks like this is a fairly common failure, thoughts on below pics? I'm guessing too far gone to try a repair but thought was worth asking...
-
As long as PCB is not shorted, it's repairable. And even if layers are shorted, it should be possible to remove a bit of PCB material and fix that. Then badge some capacitors on top and it should be fine.
-
Wow, what a common issue and man did you caps blow but yeah... solder them off, hit it with some Isopropyl Alcohol and it will look better.
@Wraper, if I did not replace the caps, would it still work? Are the caps necessary or just to filter power supply etc?
-
Wow, what a common issue and man did you caps blow but yeah... solder them off, hit it with some Isopropyl Alcohol and it will look better.
@Wraper, if I did not replace the caps, would it still work? Are the caps necessary or just to filter power supply etc?
Those caps are for decoupling RAM VRM IIRC. Some caps are needed there because they are separated from the rest of power rail by ferrite beads. If there are no caps at all, VRM will not work properly.
-
Edit - does seem like there's a short so guess that is my main problem.
Thanks for the replies and help! A few more questions if I may.
Had a go at cleaning up as below, though may have gone too far, look okay still to try and put some caps back?
How do I check the F4 on the back (time to get a multimeter I guess) or should I just assume to replace anyway?
Also noticed that LB 20 just above the screw hole is looking pretty gunky, could that be a problem as well?
Just to add it looks like the vrm thermal pad was on a couple of the caps that went so I wonder if overheating contributes to the issue.
-
Yes you will need a multimeter to test if the Fuse F4 is open or short.
You should also test the pads on the board where the Capacitors came from and make there are no shorts between the ends off where the capacitors were.
-
I deleted all 5 of the capacitors there (~C174).
The smaller one was accidental when the iron swiped it. I had been through a few attempts with new caps on but others were short blowing the fuse a few times, so I just tried it without those 5 caps and the thing worked (obviously with a new fuse in it).
I am using it for just constant calculation load rather than games etc, but it has been stable for ~12 hours of run time so far.
-
Late reply here. I also want to thank everyone for awesome help and great explanations here. My card is still running fine, and has been now for about 3 months daily use. :-+ :-+ :-+
-
so I just tried it without those 5 caps and the thing worked (obviously with a new fuse in it).
I am using it for just constant calculation load rather than games etc, but it has been stable for ~12 hours of run time so far.
That's really interesting, over my paygrade to understand how this thing can work without the caps :scared: :wtf: :popcorn:
Congrats on getting it going!
-
My card is back up and running now too, thanks for all the info and tips!
I replaced the fuse but left the caps off as wasn't too confident on putting them back given the state of the pads.
I did get another spark and it looked like maybe C630 had gone, didn't measure short though and seems to be working so will just cross fingers for now.
I did seem to be measuring a short on the original blown cap pads but think may just be the limits of my meter as understand resistances can be very low on some parts of the GPU.
-
I did seem to be measuring a short on the original blown cap pads but think may just be the limits of my meter as understand resistances can be very low on some parts of the GPU.
There should be no low resistance and it's most likely the reason why you had that spark. The only place with really low resistance is GPU VCORE.
-
Hi guys, i got exactly same issue on my gigabyte 1070 mini
F4 fuse open, C175, C176 and C177 capacitors seems to be burnt. Would be really helpfull if someone could give me values or indications to get a replacement for the capacitors and fuse.
Any help would be really appreciated
Ill wait your answer
-
I went with wrappers' suggestion higher up in the thread.
4.7uF 16v
Since these are decoupling caps I ended up desoldering all the burn ones, replacing the fuse that is also suggested earlier on in the thread and that card has been mining Ethereum for about 5 months straight (24/7) no problems now.
I would suggest ordering more fuses that you think as you might blow a few whilst testing what works.
-
Thanks for your fast reply, I'll try to see if i can get capacitors with those values, in the shops that i ask near my city they have ones for different values in 0603 dimensions. After I replace them I'll let you know how it goes,
Great help, thanks guys.
-
...Most likely fuse on that 12V power rail is blown as well. IIRC it's 0603 size with 4.7uF capacitance, should be rated for 16V at least. I think that part of VRM was powered from PCI-E slot with fuse located on the other side of PCB somewhere close to the slot on the end with video outputs. 1206 Size 10A fast action.
Wraper's exact statement quoted which I followed.
I had to order online from Element14 or similar to get the parts. I actually found the fuse the hardest part to source. Remember it has to be "fast action".
Just a thought from my experience, if you have not soldered to a GPU before you might want to follow my suggestion of deleting the capacitors and trying a new fuse first. I say this because the ground plane on the GPU is a large piece of copper which I had a lot of trouble heating up. This meant I had to purchase a hot air gun to then allow my 60W soldering iron to have a chance of bringing the ground pad up to soldering temp.
-
Hi Wrapper
I just found this posting and have a full size 1070 and have a similar issue with the blown fuse on the 12v PCI-e socket.
So some questions ... have you or anyone else had this issue with the full size card ? I looked for cooked or shorted caps and so far have found none.
Does the full card have the same cap cracking issues ?
I have so far determined two areas on the card that use that 12v PCIe power rail ... 1st is the power feed for the Logo and Fan Stop LED's on top side of card.
2nd is it seems to also feed power to the video memory. I know that the memory doesn't use 12vdc but about 1.5v ... and so far have not been able to find the controller for the two mosfets that feed mem power so as to regulate the 1.5v.
My blown fuse has a 10 written in a green square and looks to be plain white ceramic body.
This doesn't seem to make sense for me as the 12v PCIe socket is only rated for 75 watts. So that 10 amp fuse would allow far more than it's rated 75W ... like about 120W.
I would be putting a 6A which would be 70W
Does this make sense to you ?
Since all I have bee able to find is this blown fuse ... I'm wondering what could have caused this ... bad mem chip ? Have checked for dead shorts on bypass caps for mem and all seems good there.
Any ideas ?
Plus one more question ... regarding the power for the GPU ... I seem to be getting a dead short across the bypass caps for the GPU or so it seems.
I have read that the GPU has a very low resistance across it's power pins ... so would this cause what looks like a short to ground from Vcore ?
I read 0.5 ohms from any of the large inductors to 8 pin power connector GND ... seems to me to be very low ... like way too low.
I have 4 other dead cards and they all exhibit the same readings from any inductor to power GND.
I have to assume this is normal ?
Pics included
-
Update to last post .... I missed one of the 12vdc feeds from the PCIe slot ... seems one of the mosfet phases is connected to the slot power.
It's the the first set up from the bottom above the fuse . Possibly a bad fet blew the fuse ...
-
Capacitor cracking issue seems to be typical for particular 1070 mini card because they are located very close to a screw hole. On other cards capacitors sometimes fail too but mosfet failure is way more common.
-
Might be a good idea to replace them with soft termination caps (AEC-Q200 qualified).
-
Hi All,
I have another update and some questions
In my 1st post I stated that it looks like there is short from Vcore to GND or in other words a short across the GPU power pins.
I went ahead and removed what I'm pretty sure are the low side fets ... 12 in all. They all test good on the bench.
The six highside fets test good still on the board ... and there still seems to be a short ... only thing in my mind would be a shorted cap(electro) or the GPU is shorted.
I think wrapper has said that GPU fails are fairly rare ... I included a pic of the meter reading from the drain on the second low side fet to the source of that same fet.
As one can see it looks pretty shorted to me ... less than 0.5 ohms. So we are not seeing a reasonably low ohm reading from the GPU.
To recap ... the fuse F4 blew and it feeds 12v from the card slot which feeds the LED's, Memory and the first fet's drain at the bottom of the high side column of fet's.
The one closest to the fuse and card slot connector. I'm not sure if it is a phase for the overall VRM or not.
The source for this fet is connected to all the other fets so we can probably assume it is a phase as it's source is connected to Vcore.
So any ideas would be welcome at this point.
Do I dare put the fets back on and refuse and try the card out ?
Also the issue of the fuse rating ... I still think 10A is far too much ... I'd go with no more than 6.25 MAYBE 7 tops.
-
I still think 10A is far too much ... I'd go with no more than 6.25 MAYBE 7 tops.
Old thread, however I feel I should comment about this. By standard it's allowed to pull 75W from PCI-E slot, some cards even pull somewhat more. That means 6.3A at 12V. So no, you shouldn't use fuse rated for lower current. Especially considered it's a miniaturized fuse.
-
I still think 10A is far too much ... I'd go with no more than 6.25 MAYBE 7 tops.
Old thread, however I feel I should comment about this. By standard it's allowed to pull 75W from PCI-E slot, some cards even pull somewhat more. That means 6.3A at 12V. So no, you shouldn't use fuse rated for lower current. Especially considered it's a miniaturized fuse.
Just for completeness: The wikipedia article about PCI Express (http://"https://en.wikipedia.org/wiki/PCI_Express#Slot_power") indicates that 75W is total power from the slot, including 3.3V: All PCI express cards may consume up to 3 A at +3.3 V (9.9 W). (...) A full-sized x16 graphics card may draw up to 5.5 A at +12 V (66 W) and 75 W combined after initialization and software configuration as a high-power device.
-
Hi wraper,
Thanks ... I'm not sure what ended up happening with this card. I think I did get it up and running but unfortunately I think I left it with a slightly under rated fuse as you point out ... For 84W the fuse should be 7 Amp. But thanks for chiming in at such a late date.
Journey
-
Hi wraper,
Thanks ... I'm not sure what ended up happening with this card. I think I did get it up and running but unfortunately I think I left it with a slightly under rated fuse as you point out ... For 84W the fuse should be 7 Amp. But thanks for chiming in at such a late date.
Journey
Fuses usually survive up to about 1.5x of rated current in long term, however they may eventually fail, especially if they experience long loading sessions. It depends a lot on ambient temperature and how heat is transfered away from it.