The owner of this RTX 3080 asked me to mail it back to him with a letter stating it is unrepairable, for his insurance. As I offer a no fix no fee repair service there is no charge for this repair attempt.
Seems fair not to charge anything since you didn't attempt anything at all, the only thing i see is you making assumptions based on nothing. You've only did some measurements and came to the conclusion based upon those 'measurements' that is it the GPU itself... Allow me to introduce you this guy and watch him closely in real repairing!! th-cam.com/users/KrisFixGermanyvideos
Hello there sire. I have a similar issue with an RX 570, it has tolerable voltage and resistance readings, no shorts on the PCIe fingers, but denies to post or even beep. I have managed to make it do long and three short beep after shorting out two BIOS pins. Took some readings off of these 2 BIOSes and they vary, both seem to be shorted together. Im getting 3.3 and 1.8 reading on both when powered up and the others have 0 or close to volts. Just for the sake of confirmation, as Im no technician, both chips should have the same pinout and read alike in continuous mode, also not being shorted to each other on either of the pins except for common ground? Vmem controller and vcore get same reading of close to 0.9.
Oh just to mention - before I reasembled the card to make some more tests I spent 10 minutes cleaning the traces of heatsink compound from around the RAM and BIOS chip. It didn't make any difference.
Awesome Video as always it's unfortunate but its not always economically viable fix stuff. a RTX GPU chip is extremely hard to replace and Expensive to get as a verified working spare part . but that's what makes your channel great you show successes and failures and you can always learn something from both. Keep up the good work Cheers Richard
Awesome video? Are you nuts?!? In an awesome video there would be a GPU reball/reflow attempt because this type of card is prone to failure because of cold joints breaking after you fiddle with thermal pads and/or thermal compound. With all due respect to the creator of this video, but this repair attempt borders more towards amateurism rather than awesomeness...Besides that, my fingers itch when i see this kind of mishaps on multiple level...
@@sapperdeflap So i am guess from your comment you haven't watched the video! Richard did a good jobb finding the actual fault, a shorted PCIE lane number 15 i think , that means no amount of reballing would ever fix this if it was showing open i would agree with you. and furthermore this PCIE lane goes directly to the GPU so there are no components that could short it to ground on the way there? you must be one of those mechanics that just throw parts at a problem and hope they go away.
@@RetroUpgrade So the PCIe lane goes directly to the GPU you mean? Please elaborate yourself, because i'm still under the impression that all direct connections to the GPU go via the BGA underneath the GPU where those cold joints appear on some models. But i have to leave you with this knowledge because from my perspective you're not the right person to talk to about this matter... Have a good look again to the video, the old fart measures an OL which means Open Line, so no short on the PCIe Lane (see 26:20). BTW I'm one of those mechanics who always wear gloves, use ESD equipment, earth myself and use sharp probes so i don't scratch anything and trying to keep the devices as pristine as possible with a 95% success rate on all repairs, i watched two video's of this bloke with a 100% failure rate, but please do keep cheering him on in what kind of awesomeness this man delivers!
@@sapperdeflap You are right was think this was another video he did MB. it was open . but still if you go full tilt on "Professional measures" i wont say that's bad but it's actually personal preference , he has a ESD mat that is grounded and for 99% of the cases should be fine. and regarding the success rate XD i have 100% success rate atm but that doesn't mean i am the best at fixing video games it just means i got lucky !!. there are things that are unfixable and i some cases the client doesn't want to take a risky step like reballing specially like the one in question could send it back for insurance purposes. and bro please don't be so angry at the world life sucks already don't make it worse for yourself. offer a helpful comment instead, like what he did wrong and how to fix it !!
@@RetroUpgrade My helpful contribution/comment was to reflow/reball the GPU, i'm not going on tilt on anything, but if you call something awesome while it is absolutely not, i feel my OCD come up to put things in perspective and right context. The choices are limited for the RTX3080, it will either continue it's life as an expensive paperweight, or you should try the 'risky' step in reballing/reflowing and risk that the RTX3080 will end as the exact same expensive paperweight...?
thank you Richard for the video, you've inspired me to start learning about reparing graphics cards. Can I ask what was the adapter you plugged into the GPU.
Like others say - mining adapter or riser. It's very useful because: it makes it easy to physically work on a GPU it runs the GPU in PCIe x1 mode so there are only six 'data' connections between the PC and GPU it makes it easy to connect a current meter if you want to know how much power the GPU is drawing from the PCIe slot as well as the 6x/8x connectors and it's also cheap to buy. I show it's uses in detail on these videos: th-cam.com/video/-tfdjvfGQmo/w-d-xo.html th-cam.com/video/8Kv6A-nWjNM/w-d-xo.html th-cam.com/video/RMspWrf40iY/w-d-xo.html
Hi there, Did you try to plug out the hdmi, boot the pc with integrated graphics and then plug in the graphics card hdmi? I found out my mainboard bios has a priority for integrated or not integrated graphics. It the priority was set for video out to graphics card but the card was broken therefore I had no video out (also had a video out from the graphics card with I unplugged the 12V because a DRmos was bad). So I just plugged in the power cables into the graphicscard and the dvi into the mainboard. After the boot I put the hdmi into the graphicscard and I still had a picture (comming from integrated graphics). Greetings from Germany and thanks for the video
I have 2 1070 cards that power up all good, bios pex rst 27mghz vcore and fbvdd look good on oscilloscope. Yet neither one boots. Even mats won't run after monitor led lights up.
I'd be happy to look at them, no fix no fee. electronicanaria@outlook.com Every subscribers item I look is published on video so you get to see what work is undertaken. As you have a channel yourself maybe we could do some online collaboration (live stream etc)?
loving the channel and your contributions! Is there a source for the boardviewer schematics you recommend? specifically I am after the EVGA XC3 3080ti.
Maybe you tried this, what about changing the graphics card init order on your motherboard to built in first and then putting the 3080 on the x1 riser. Are you able to boot mods that way ? This exact setup is why I use a socket am3 with nvidia chipset video so that I can force the primary card init.
Well, I hoped I would never write this but my 3080 died tonight while playing warzone, lights went off in the house when it shorted, turned everything back on, pc didn't boot, went on to trouble shoot and found out that one of the three powers connectors is causing a short circuit, I didn't take it apart since it's still under warranty luckily and I'd rather not make it a 1500$ brick at this point but man am I curious if the entire GPU died or it's just a short on one of the power connectors. ( it's the third one from the left, every time I plug it in the PC shuts down and won't boot until I drain the power and plug the power back in, lights turn on, fans are spinning on the gpu with only two power connectors instead of three but obviously it won't work). Mine is the ROG Strix OC, bought and paid extra hoping I'd get the most reliable one and here I am, disappointed as hell not knowing what I'll get in return since mine was an amazing card with no coil whine :(
I'm sure that if you screw the cooler back, it pushes the gpu's balls together and if you take the cooler off, they will separate. Or the little golden pins inside the gpu are cracked.
how about this, sending cards back on warranty and still getting sent back to me faulty this has happen twice i brought a gigabyte rx6800xt ended up being faulty so they send me a refurbished 3080ti end up sending that back and they sent it back to me saying its fine to get no post / display whats so ever.
I'm sad to say, but it looks like a dead core. I would have watched some of techcemetery video on either YT or preferably Twitch. He has repaired alot of GPU's and he is an wizard when it comes to GPU repair
@Thomas Aslaksen You think I don't watch Tech Cemetery? He's very good at GPU repair, I've seen loads of his videos. He also recommends my channel (see the recommended links on TC as I recommend his) and a lot guys active on LER Discord are also on TC discord.
@@LearnElectronicsRepair Yes I think you watch TC, but just wanted to give him a shoutout. But I hope you will still make this good repair videos. Love your work and I have learned so much from you. Have a nice weekend and hope to see some more old tech in the future.
@@Thomas_Aslaksen Yeah TC is one of the good guys and is always welcome to be mentioned here. He is also a subscriber here and sometimes comments on my videos Tech Cemetery th-cam.com/video/RMspWrf40iY/w-d-xo.html One of the best dedicated GPU repair channels on YT 😁
Hey great vids, just wanted to say the low value resistors by the 8 way connectors are not acting as fuses, they are part of the current sensing circuit. (Shunt resistors)
@@kikihun9726 Yeah, the low resistance of these current sensors equates to a lot of Watts if you have a short - and they burn up just like fuses. However in future I'll start calling them by their correct name.
Hi, i got rx480 in same conditions, all voltajes, not video, comunication in pin 1 of memory, buy what about pin6clockf memory, im need here constant comunication???
Another possibility is if your test motherboard supports PCIe 4.0 the RTX GPU will not work in this riser unless the PCIe version is lowered to 3.0 in your Motherboard bios.
I didn't know that - thanks for the info. The test motherboard is a 4th gen with i3 or celeron CPU so I doubt that woould be the problem in this case, but it is certainly worth knowing. 🙂
Hey great video. You probably know this already but there's MATS floating around for the 3080. However most likely you'll have to use mods and get memory training errors, which seems common on the high end cards nowadays when there's cracked solder or ripped pads under the memory. I just fixed one like this the yesterday, and I've seen them a bunch of times
Yeah I have MATS for the 3080, but this one has an open circuit PCIe lane so I don't see how testing the memory, even if that is possible, would achieve anything
@@LearnElectronicsRepair yes you are right, but maybe there are more than one cracked balls, other wise how does the gpu know whether heat sink is bolted on or not, maybe the pressure of bolting squeezes the other cracked balls. Detecting plugged in is one thing, but detecting bolted on without contact switch or something like that is another thing
@@yogeshroy9289 There is certainly no problem with what you suggest, though I think if I am going to go down that route it would be better to try a reball than a reflow, if I can find a stencil. I'll ask the owner, Markus, see what he would like to do with it
Hello! Great video! How often would you say, that a relatively newer (for example less then 4-5 years old) video card is repairable? And how often would you say, that the problem on newer cards is something relatively easy to fix? I am a bit reluctant to try and repair a newer card as it seems that they have problems, which are harder to solve and sometimes even unrepairable. Pretty sure, that I would not be able to solve one of the trickier issues, so I am intrested in my chances of picking up a newer card, that I actually have a chance at repairing. Thank you for the answer in advance!
Gergő Lengyel Firstly I will say that newer GPUs are not inherently harder to repair than older ones. They all use pretty much the same circuit design and the methods I teach here using old cards are just as valid for repairing the latest ones. If you can fix a GTX 780 you have all the skills you need to fix a RTX 3080 Ok, now in my own honest opinion ... the answer to your question depends on A: where the GPU came from and B: the skill of the technician trying to fix it. If it is your own GPU that 'just failed' or one you acquire that has failed but no one has attempted to repair or generally 'f#ck with' them I would say that the chances of repairing a 'newer' card are quite high. There are exceptions though, certain models are known to have a very high failure rate in the GPU itself. If it is a graphihcs card that has 'done the rounds' especially on ebay spares or repair then in my experience you should expect to first have to repair the stupid problems such as very bad amateur soldering attempts, broken off components, etc, which probably occurred when someone else tried to fix it, and then once you get that sorted expect to have to then repair the original problem which was there in the first place. And other times with ebay GPUs the 'obvious problem' is that someone burnt it with a heat gun. Here are a couple of good examples of the two scenarios I just described, one I fixed though you can see how much effort was required to do it, and the other was DOA and uneconomical to repair. That's not the same thing as unrepairable, almost nothing falls into that category. GTX 980 ti from ebay spares or repair th-cam.com/video/ana2wbDsKkg/w-d-xo.html GTX 1080 ti subscribers card th-cam.com/video/RnXMMglJJD0/w-d-xo.html Unfortunately many of the cards I get sent by subscribers fall into the category of 'having done the rounds' and the success rate is not particularly good. I can usually diagnose them and fix the 'secondary damage' but they are not economical to repair regards the original problem. On the plus side however, I do offer a no fix no fee repair service, plus fixed price repairs, and every job gets published on video whether I can fix it or not, so that may be a good option 😉
It's a 'slang?' term for PCI Express The PEX supply is 0.9V-1.0V and powers the part of the GPU that communicates with the PCIe slot on the motherboard. This consists of two LVDS transmitters and receivers for each PCIe lane plus a reference clock signal. This video explains it really well th-cam.com/video/-tfdjvfGQmo/w-d-xo.html
You should give that Fluke a good clean. That bolster easily cleans up nicely and makes it visually more pleasing to work with. For the card probably bad memory like others have mentioned here. You earned a new sub here (since last week) while I have electronics experience since the '80's. Never to old to learn something and it seems we sometimes have a different approach at things which can be interesting.
@Luc Peeters Several comments are saying bad ram (admittedly I couldn't test it as the card disables my onboard video - I can't even get to the BIOS) but what I don't get is how can bad memory cause an obvious open circuit on PCIe lane 15?
@@LearnElectronicsRepair Very good question indeed. Sometimes you wish you had a schematic, pcb layout and a working card to compare on the bench. For an expensive card as this it is worth investing some time. Put it aside for a couple of hours and do something else that is not frustrating is also one of my strategies lately. I will watch the video again tonight to let it all sink in again.
@@Luke-san I actually do have a boardview for this card, I showed it in the video I thought you would have seen that. It's not gonna help in this case though, the open circuit PCIe lane goes directly to the GPU.
@@LearnElectronicsRepair Yup, that's why I said that I need to re-watch it, apologies for that! Sometimes I get interrupted every 1 or 2 minutes and miss a lot of important stuff. I should lock myself up and place my phone outside of my room. Also reason I slowed down on repairs since we all are contactable every single moment of the day and family life.
@@LearnElectronicsRepair Djeezzz, seems I missed like 80% the first time I watched it. Did you try to measure via Ohm's metering to see if there is no resistance at all? (as you measured in diode mode) Guess an internal issue in the GPU if it goes directly to the GPU. (could have been a nice warranty case). At first I was going to say a wrongly flashed bios or memory as that can also cause not booting at all I've seen but that can be ruled out now. Also I never knew that these diagrams were released by the manufacturer. Btw most excellent troubleshooting.
Hello Richard. I have a gtx 960 which I've purchased as not working and I've no idea where to begin with it. If I was to send it to you, would you be able to maybe take a look and fix it? I can explain the issue in an email maybe? Is it possible to make contact privately to explain the issue in more detail as it may be beyond repair or not feasible to fix. Thanks very much. Graham
Is there any business that repairs cards that have been damaged? I have a 10gb msi 3080 that is dead upon arrival in a pc I purchased. If I could get it repaired I would cry happy
I wonder if the “patch” he did with lane 1 of his GPU would work on lane 16 of this one. Basically use a potentiometer to find out what resistance is needed to be addded on the broken pcie lane and then solder equivalent resistor. See these videos. I have not seen this elsewhere. th-cam.com/video/ln2Xh-_ckUU/w-d-xo.html
I am using cheap microscope China made from a dashcam. With large led light ring it looks very good. Better then his too for what ever reason. I think I paid less then 100 for it but I had a articulating the magnifying lamp which I used as a base to hold my scope. Maybe I will show it on my channel one day. Too busy trying to fix cards lol. Honestly, I don't think scope quality is that important. You can always bring the card closer to see better if needed for sure. My cheap scope can do it, any scope can do it.
First of all richard wasn't using the microscope at the start just the overhead camera zoomed in , secondly a highly zoomed in microscope image would have given a lot of ppl motion sickness,i am talking from experience making videos with a microscope myself. i have been to his shop and actually tried out the microscope and its a top notch one. and the image it produces is of the highest quality. unfortunately youtube algorithm doesn't play nice with high detail images. hope this clears up why it look "BAD" Btw its not a personal attack , just educating the masses on that youtube compression sucks 😁😆
Hello ive just come across your channel ive a faulty plait 3080 non lhr i would be willing to donate are pay if you could help me with this it turns on but no video output ive been trying to get it fixed for along time nice videos very well detailed
for the next time you can buy a programmer for the chip so you dont need to put it into a computer of let it run but than ur can just take the chip off and flash it whit that programmer and put it back.
@@LearnElectronicsRepair Have ppl even watched all the way through the video XD, You clearly show that PCIE lane 15 is shorted and that lane goes directly to de gpu. how could the possibly be a ram issue ? can someone explain why you are all saying ram ?
@@RetroUpgrade PCIE lane 15 wasn't shorted, it was an open line. Perhaps bad solder ball joint to GPU? or in the worst case a broken trace, where you can't even fix by reballing
@@konstantinosdimitroulas1950 if it boots you can use MATS/MODS (Don't know about 3xxx series GPUs) but this card didn't even boot. GPUs with mem corruption can sometimes boot but has errors like error 43 and no picture.
@@muneebhamdi7605 I'll go along with internal damage to the GPU chip, as that has clearly happened with the PCIe lanes, but bad mem as many are commenting, it doesn't make sense considering the evidence.
@@LearnElectronicsRepair yes I agree the memory controller is part of the gpu the 30 series seem to have this issue it seems as my 3090 practically has the same issue
The owner of this RTX 3080 asked me to mail it back to him with a letter stating it is unrepairable, for his insurance. As I offer a no fix no fee repair service there is no charge for this repair attempt.
Seems fair not to charge anything since you didn't attempt anything at all, the only thing i see is you making assumptions based on nothing. You've only did some measurements and came to the conclusion based upon those 'measurements' that is it the GPU itself... Allow me to introduce you this guy and watch him closely in real repairing!! th-cam.com/users/KrisFixGermanyvideos
Hello there sire. I have a similar issue with an RX 570, it has tolerable voltage and resistance readings, no shorts on the PCIe fingers, but denies to post or even beep. I have managed to make it do long and three short beep after shorting out two BIOS pins. Took some readings off of these 2 BIOSes and they vary, both seem to be shorted together. Im getting 3.3 and 1.8 reading on both when powered up and the others have 0 or close to volts. Just for the sake of confirmation, as Im no technician, both chips should have the same pinout and read alike in continuous mode, also not being shorted to each other on either of the pins except for common ground?
Vmem controller and vcore get same reading of close to 0.9.
Oh just to mention - before I reasembled the card to make some more tests I spent 10 minutes cleaning the traces of heatsink compound from around the RAM and BIOS chip. It didn't make any difference.
Awesome Video as always
it's unfortunate but its not always economically viable fix stuff. a RTX GPU chip is extremely hard to replace and Expensive to get as a verified working spare part .
but that's what makes your channel great you show successes and failures and you can always learn something from both.
Keep up the good work
Cheers Richard
Awesome video? Are you nuts?!? In an awesome video there would be a GPU reball/reflow attempt because this type of card is prone to failure because of cold joints breaking after you fiddle with thermal pads and/or thermal compound. With all due respect to the creator of this video, but this repair attempt borders more towards amateurism rather than awesomeness...Besides that, my fingers itch when i see this kind of mishaps on multiple level...
@@sapperdeflap So i am guess from your comment you haven't watched the video! Richard did a good jobb finding the actual fault, a shorted PCIE lane number 15 i think , that means no amount of reballing would ever fix this if it was showing open i would agree with you.
and furthermore this PCIE lane goes directly to the GPU so there are no components that could short it to ground on the way there?
you must be one of those mechanics that just throw parts at a problem and hope they go away.
@@RetroUpgrade So the PCIe lane goes directly to the GPU you mean? Please elaborate yourself, because i'm still under the impression that all direct connections to the GPU go via the BGA underneath the GPU where those cold joints appear on some models. But i have to leave you with this knowledge because from my perspective you're not the right person to talk to about this matter... Have a good look again to the video, the old fart measures an OL which means Open Line, so no short on the PCIe Lane (see 26:20). BTW I'm one of those mechanics who always wear gloves, use ESD equipment, earth myself and use sharp probes so i don't scratch anything and trying to keep the devices as pristine as possible with a 95% success rate on all repairs, i watched two video's of this bloke with a 100% failure rate, but please do keep cheering him on in what kind of awesomeness this man delivers!
@@sapperdeflap You are right was think this was another video he did MB.
it was open .
but still if you go full tilt on "Professional measures" i wont say that's bad but it's actually personal preference , he has a ESD mat that is grounded and for 99% of the cases should be fine.
and regarding the success rate XD i have 100% success rate atm but that doesn't mean i am the best at fixing video games it just means i got lucky !!.
there are things that are unfixable and i some cases the client doesn't want to take a risky step like reballing specially like the one in question could send it back for insurance purposes.
and bro please don't be so angry at the world life sucks already don't make it worse for yourself.
offer a helpful comment instead, like what he did wrong and how to fix it !!
@@RetroUpgrade My helpful contribution/comment was to reflow/reball the GPU, i'm not going on tilt on anything, but if you call something awesome while it is absolutely not, i feel my OCD come up to put things in perspective and right context. The choices are limited for the RTX3080, it will either continue it's life as an expensive paperweight, or you should try the 'risky' step in reballing/reflowing and risk that the RTX3080 will end as the exact same expensive paperweight...?
Couldn't we try GPU reflow or reball? I'm thinking open connection might be caused by bad solder connection beneath the gpu chip.
thank you Richard for the video, you've inspired me to start learning about reparing graphics cards. Can I ask what was the adapter you plugged into the GPU.
Its a simple mining adapter pcie to usb mining
It's called "riser"
Like others say - mining adapter or riser. It's very useful because:
it makes it easy to physically work on a GPU
it runs the GPU in PCIe x1 mode so there are only six 'data' connections between the PC and GPU
it makes it easy to connect a current meter if you want to know how much power the GPU is drawing from the PCIe slot as well as the 6x/8x connectors
and it's also cheap to buy.
I show it's uses in detail on these videos:
th-cam.com/video/-tfdjvfGQmo/w-d-xo.html
th-cam.com/video/8Kv6A-nWjNM/w-d-xo.html
th-cam.com/video/RMspWrf40iY/w-d-xo.html
What about trying a GPU reflow?
Hi there,
Did you try to plug out the hdmi, boot the pc with integrated graphics and then plug in the graphics card hdmi?
I found out my mainboard bios has a priority for integrated or not integrated graphics. It the priority was set for video out to graphics card but the card was broken therefore I had no video out (also had a video out from the graphics card with I unplugged the 12V because a DRmos was bad). So I just plugged in the power cables into the graphicscard and the dvi into the mainboard. After the boot I put the hdmi into the graphicscard and I still had a picture (comming from integrated graphics).
Greetings from Germany and thanks for the video
I have 2 1070 cards that power up all good, bios pex rst 27mghz vcore and fbvdd look good on oscilloscope.
Yet neither one boots.
Even mats won't run after monitor led lights up.
I'd be happy to look at them, no fix no fee. electronicanaria@outlook.com Every subscribers item I look is published on video so you get to see what work is undertaken. As you have a channel yourself maybe we could do some online collaboration (live stream etc)?
@@LearnElectronicsRepair shipping cost makes it not worth it.
If it was 3070, then yes.
Thank you for trying to help.
loving the channel and your contributions! Is there a source for the boardviewer schematics you recommend? specifically I am after the EVGA XC3 3080ti.
Lift the core and look for ripped pads. Also check resistances on that lane between the pin and the core. Ask #NorthWestRepair.
Maybe you tried this, what about changing the graphics card init order on your motherboard to built in first and then putting the 3080 on the x1 riser. Are you able to boot mods that way ? This exact setup is why I use a socket am3 with nvidia chipset video so that I can force the primary card init.
@zeferrum Thanks for the suggestion - I'll give it a go
Do you have a schematic diagram for galax gtx 1650 super ex
Well, I hoped I would never write this but my 3080 died tonight while playing warzone, lights went off in the house when it shorted, turned everything back on, pc didn't boot, went on to trouble shoot and found out that one of the three powers connectors is causing a short circuit, I didn't take it apart since it's still under warranty luckily and I'd rather not make it a 1500$ brick at this point but man am I curious if the entire GPU died or it's just a short on one of the power connectors. ( it's the third one from the left, every time I plug it in the PC shuts down and won't boot until I drain the power and plug the power back in, lights turn on, fans are spinning on the gpu with only two power connectors instead of three but obviously it won't work). Mine is the ROG Strix OC, bought and paid extra hoping I'd get the most reliable one and here I am, disappointed as hell not knowing what I'll get in return since mine was an amazing card with no coil whine :(
Thank you
Keep up the good work
Cheers
I'm sure that if you screw the cooler back, it pushes the gpu's balls together and if you take the cooler off, they will separate.
Or the little golden pins inside the gpu are cracked.
thats not how solder works
hello where i can find the diagram of any gpu ? i want to learn also how to repair the gpu. thank you in advance
From were I can take a diagram for Assus strix 1080ti?
how about this, sending cards back on warranty and still getting sent back to me faulty this has happen twice i brought a gigabyte rx6800xt ended up being faulty so they send me a refurbished 3080ti end up sending that back and they sent it back to me saying its fine to get no post / display whats so ever.
I'm sad to say, but it looks like a dead core. I would have watched some of techcemetery video on either YT or preferably Twitch. He has repaired alot of GPU's and he is an wizard when it comes to GPU repair
@Thomas Aslaksen You think I don't watch Tech Cemetery? He's very good at GPU repair, I've seen loads of his videos. He also recommends my channel (see the recommended links on TC as I recommend his) and a lot guys active on LER Discord are also on TC discord.
@@LearnElectronicsRepair Yes I think you watch TC, but just wanted to give him a shoutout. But I hope you will still make this good repair videos. Love your work and I have learned so much from you. Have a nice weekend and hope to see some more old tech in the future.
@@Thomas_Aslaksen Yeah TC is one of the good guys and is always welcome to be mentioned here. He is also a subscriber here and sometimes comments on my videos
Tech Cemetery
th-cam.com/video/RMspWrf40iY/w-d-xo.html
One of the best dedicated GPU repair channels on YT 😁
Hey great vids, just wanted to say the low value resistors by the 8 way connectors are not acting as fuses, they are part of the current sensing circuit. (Shunt resistors)
But they can burn too💁♂️
@@kikihun9726 Yeah, the low resistance of these current sensors equates to a lot of Watts if you have a short - and they burn up just like fuses. However in future I'll start calling them by their correct name.
My 3080ti has the same problem, have you figured out a fix sense this video was pulished?
Hi, i got rx480 in same conditions, all voltajes, not video, comunication in pin 1 of memory, buy what about pin6clockf memory, im need here constant comunication???
Another possibility is if your test motherboard supports PCIe 4.0 the RTX GPU will not work in this riser unless the PCIe version is lowered to 3.0 in your Motherboard bios.
I didn't know that - thanks for the info. The test motherboard is a 4th gen with i3 or celeron CPU so I doubt that woould be the problem in this case, but it is certainly worth knowing. 🙂
Hey great video. You probably know this already but there's MATS floating around for the 3080. However most likely you'll have to use mods and get memory training errors, which seems common on the high end cards nowadays when there's cracked solder or ripped pads under the memory. I just fixed one like this the yesterday, and I've seen them a bunch of times
Yeah I have MATS for the 3080, but this one has an open circuit PCIe lane so I don't see how testing the memory, even if that is possible, would achieve anything
Maybe a cracked solder ball under the gpu, can give a try at reflow
Yes that I could give a try - but if that was the case really it should have worked in PCIe x1 mode,
@@LearnElectronicsRepair yes you are right, but maybe there are more than one cracked balls, other wise how does the gpu know whether heat sink is bolted on or not, maybe the pressure of bolting squeezes the other cracked balls. Detecting plugged in is one thing, but detecting bolted on without contact switch or something like that is another thing
@@yogeshroy9289 There is certainly no problem with what you suggest, though I think if I am going to go down that route it would be better to try a reball than a reflow, if I can find a stencil. I'll ask the owner, Markus, see what he would like to do with it
Hello! Great video!
How often would you say, that a relatively newer (for example less then 4-5 years old) video card is repairable? And how often would you say, that the problem on newer cards is something relatively easy to fix? I am a bit reluctant to try and repair a newer card as it seems that they have problems, which are harder to solve and sometimes even unrepairable. Pretty sure, that I would not be able to solve one of the trickier issues, so I am intrested in my chances of picking up a newer card, that I actually have a chance at repairing. Thank you for the answer in advance!
Gergő Lengyel
Firstly I will say that newer GPUs are not inherently harder to repair than older ones. They all use pretty much the same circuit design and the methods I teach here using old cards are just as valid for repairing the latest ones. If you can fix a GTX 780 you have all the skills you need to fix a RTX 3080
Ok, now in my own honest opinion
... the answer to your question depends on A: where the GPU came from and B: the skill of the technician trying to fix it.
If it is your own GPU that 'just failed' or one you acquire that has failed but no one has attempted to repair or generally 'f#ck with' them I would say that the chances of repairing a 'newer' card are quite high. There are exceptions though, certain models are known to have a very high failure rate in the GPU itself.
If it is a graphihcs card that has 'done the rounds' especially on ebay spares or repair then in my experience you should expect to first have to repair the stupid problems such as very bad amateur soldering attempts, broken off components, etc, which probably occurred when someone else tried to fix it, and then once you get that sorted expect to have to then repair the original problem which was there in the first place.
And other times with ebay GPUs the 'obvious problem' is that someone burnt it with a heat gun.
Here are a couple of good examples of the two scenarios I just described, one I fixed though you can see how much effort was required to do it, and the other was DOA and uneconomical to repair. That's not the same thing as unrepairable, almost nothing falls into that category.
GTX 980 ti from ebay spares or repair
th-cam.com/video/ana2wbDsKkg/w-d-xo.html
GTX 1080 ti subscribers card
th-cam.com/video/RnXMMglJJD0/w-d-xo.html
Unfortunately many of the cards I get sent by subscribers fall into the category of 'having done the rounds' and the success rate is not particularly good. I can usually diagnose them and fix the 'secondary damage' but they are not economical to repair regards the original problem.
On the plus side however, I do offer a no fix no fee repair service, plus fixed price repairs, and every job gets published on video whether I can fix it or not, so that may be a good option 😉
@@LearnElectronicsRepair Thank you very much! It was very heplful!
May I ask what is PEX?
It's a 'slang?' term for PCI Express
The PEX supply is 0.9V-1.0V and powers the part of the GPU that communicates with the PCIe slot on the motherboard. This consists of two LVDS transmitters and receivers for each PCIe lane plus a reference clock signal.
This video explains it really well
th-cam.com/video/-tfdjvfGQmo/w-d-xo.html
@@LearnElectronicsRepair Thank you Sir.
You should give that Fluke a good clean. That bolster easily cleans up nicely and makes it visually more pleasing to work with.
For the card probably bad memory like others have mentioned here. You earned a new sub here (since last week) while I have electronics experience since the '80's. Never to old to learn something and it seems we sometimes have a different approach at things which can be interesting.
@Luc Peeters Several comments are saying bad ram (admittedly I couldn't test it as the card disables my onboard video - I can't even get to the BIOS) but what I don't get is how can bad memory cause an obvious open circuit on PCIe lane 15?
@@LearnElectronicsRepair Very good question indeed. Sometimes you wish you had a schematic, pcb layout and a working card to compare on the bench. For an expensive card as this it is worth investing some time. Put it aside for a couple of hours and do something else that is not frustrating is also one of my strategies lately. I will watch the video again tonight to let it all sink in again.
@@Luke-san I actually do have a boardview for this card, I showed it in the video I thought you would have seen that. It's not gonna help in this case though, the open circuit PCIe lane goes directly to the GPU.
@@LearnElectronicsRepair Yup, that's why I said that I need to re-watch it, apologies for that!
Sometimes I get interrupted every 1 or 2 minutes and miss a lot of important stuff. I should lock myself up and place my phone outside of my room. Also reason I slowed down on repairs since we all are contactable every single moment of the day and family life.
@@LearnElectronicsRepair Djeezzz, seems I missed like 80% the first time I watched it. Did you try to measure via Ohm's metering to see if there is no resistance at all? (as you measured in diode mode) Guess an internal issue in the GPU if it goes directly to the GPU. (could have been a nice warranty case). At first I was going to say a wrongly flashed bios or memory as that can also cause not booting at all I've seen but that can be ruled out now. Also I never knew that these diagrams were released by the manufacturer. Btw most excellent troubleshooting.
Hello Richard. I have a gtx 960 which I've purchased as not working and I've no idea where to begin with it. If I was to send it to you, would you be able to maybe take a look and fix it? I can explain the issue in an email maybe? Is it possible to make contact privately to explain the issue in more detail as it may be beyond repair or not feasible to fix. Thanks very much. Graham
Sure. I offer a no fix no fee service and fixed price repairs. email me electronicananaria@outlook.com
Is there any business that repairs cards that have been damaged? I have a 10gb msi 3080 that is dead upon arrival in a pc I purchased. If I could get it repaired I would cry happy
I wonder if the “patch” he did with lane 1 of his GPU would work on lane 16 of this one. Basically use a potentiometer to find out what resistance is needed to be addded on the broken pcie lane and then solder equivalent resistor. See these videos. I have not seen this elsewhere. th-cam.com/video/ln2Xh-_ckUU/w-d-xo.html
I'll have a look at that, thanks
I have this same issue sometimes on a gtx 980 4gb ssc acx2.0
Heya not everything is repairable but you show whatn is wrong
Faultfinding sequence is fair. But please get as decent microscope. I could not see the smds clearly.
Really you don't think my Amscope clone Trinocular microscope is a good one?
www.aliexpress.com/item/1005001954165293.html
@@LearnElectronicsRepair I imagine its video compression that he is talking about. Thank you Richard for your work.
I am using cheap microscope China made from a dashcam.
With large led light ring it looks very good.
Better then his too for what ever reason.
I think I paid less then 100 for it but I had a articulating the magnifying lamp which I used as a base to hold my scope.
Maybe I will show it on my channel one day. Too busy trying to fix cards lol.
Honestly, I don't think scope quality is that important. You can always bring the card closer to see better if needed for sure.
My cheap scope can do it, any scope can do it.
your bonkers mate, try cleaning your screen
First of all richard wasn't using the microscope at the start just the overhead camera zoomed in , secondly a highly zoomed in microscope image would have given a lot of ppl motion sickness,i am talking from experience making videos with a microscope myself.
i have been to his shop and actually tried out the microscope and its a top notch one. and the image it produces is of the highest quality. unfortunately youtube algorithm doesn't play nice with high detail images.
hope this clears up why it look "BAD"
Btw its not a personal attack , just educating the masses on that youtube compression sucks 😁😆
Ahhh bummer. I hope Marcus didn’t pay too much for it.
No fix no fee.
I notice that you have a lot of diagram
Hello ive just come across your channel ive a faulty plait 3080 non lhr i would be willing to donate are pay if you could help me with this it turns on but no video output ive been trying to get it fixed for along time nice videos very well detailed
email me please electronicanaria@outlook.com
for the next time you can buy a programmer for the chip so you dont need to put it into a computer of let it run but than ur can just take the chip off and flash it whit that programmer and put it back.
GPU processor chip faulty, the best way is replace new GPU processor chip.
@Leo Ming I certainly have the equipment to do that but where would you find a replacement part?
@@LearnElectronicsRepair May I know, what is your GPU processor chip model number code?
@@leoming5648 I'll have to look on Monday when I am at the workshop. You can email me electronicanaria@outlook.com
@@LearnElectronicsRepair do u found the chip model number code?
@@leoming5648 The owner decided he doesn't want to pay for an expensive repair and asked me to return the graphics card to him
Give it 5 years and they'll be plenty of doner cards around
mostly bad ram. its too new to have solder balls problems. ram goes to hell on MANY new gpus
any idea how to check?
Several have suggested that, but how could bad ram cause an obvious open circuit on PCIe lane 15 - which I clearly found
@@LearnElectronicsRepair Have ppl even watched all the way through the video XD, You clearly show that PCIE lane 15 is shorted and that lane goes directly to de gpu. how could the possibly be a ram issue ? can someone explain why you are all saying ram ?
@@RetroUpgrade PCIE lane 15 wasn't shorted, it was an open line. Perhaps bad solder ball joint to GPU? or in the worst case a broken trace, where you can't even fix by reballing
@@konstantinosdimitroulas1950 if it boots you can use MATS/MODS (Don't know about 3xxx series GPUs) but this card didn't even boot. GPUs with mem corruption can sometimes boot but has errors like error 43 and no picture.
bad mem chip for sure
How is bad mem gonna cause an open circuit on one of the PCIe lanes?
@@LearnElectronicsRepair most likely bad memory controller on the gpu chip itself I have a 3090 exact same symptoms
@@muneebhamdi7605 I'll go along with internal damage to the GPU chip, as that has clearly happened with the PCIe lanes, but bad mem as many are commenting, it doesn't make sense considering the evidence.
@@LearnElectronicsRepair yes I agree the memory controller is part of the gpu the 30 series seem to have this issue it seems as my 3090 practically has the same issue