Wow, that was me suggesting to put the black probe on the coil for the voltage injection method. I'm not really an electronics expert myself, I only do some basic tinkering, but when you were showing the method I was immediately thinking "why isn't this working, it should be possible", and then I remembered my school teacher: "The thinnest wire gets hot." Or in other words, the heat is dissipated at the component with highest resistance, which in the initial setup was the GPU itself, so it should work if the GPU is removed from the circuit. Thanks for the follow-up and not only confirming it works, but also explaining why it works. Edit: so to clarify, there were two reasons why it wasn't working initially. 1. The higher resistance of the GPU was limiting the current through the shorted MOSFET, and as you said you couldn't increase the voltage because you'd risk damaging the GPU. And 2. Because the GPU has a higher resistance than the shorted MOSFET, even if it was safe to inject 10V, you would only see the GPU getting hot and possibly misdiagnose a defective GPU.
Alvin853 You may not be an expert (who can truly say they are really, there is always more stuff to learn with electronics repair) but what you did here was apply some excellent logical thinking to the problem for exactly the right reasons. You were absolutely spot on and as soon as you said it I knew you were right. Hats off to you mate, you taught me a trick here. 😀And I suspect you taught it to a lot of other too. I think a big thumbs up for Alvin is well deserved guys. Well done, this suggestion was pure Gold! OK so I proved it but the credit is all Alvin's 👏👏👏 I pinned this comment so everyone knows who thought of it first😀
Click CC to turn on subtitles then click Settings -> Auto translate to set whatever language subtitles you like. That's how to do it on PC, not sure about mobile but it must be possible. It sometimes take an hour or two for subtitles to appear on a new video.
well done for this idea. I already do something similar, but I still use safe voltage (0.8V). Taking the GPU core itself out of the circuit allows more current to flow and makes the shorted mosfet hotter even at 0.8V, allowing me to easily find it on my thermal camera. With good thick PSU leads I end up with 5-6A at 0.8V which is always enough to find the short (sometimes even with the IPA method and no thermal cam) all the while keeping the GPU perfectly safe. Thanks for bringing attention to this method, and bravo for another great video
Just discovered you're channel and i love how you explain things, i have started electronics repair a year ago, and i love to learn from experienced people like you. Keep up a good work.
There's a way to check if gpus or other big chip are destroyed by internally burning up, which does not always work. If you put your the red lead of your multimeter on ground and the black lead on the chip power input (in this case the vrm output), and you put your meter in diode mode, you should see about 0.3v. This is because all chips have inverse polarized protection diodes on their power supply. These protection diode tend to blow up together with the rest of the chip. But as said, it doesn't always work. Sometimes the protection diodes are still ok.
That sounds interesting, like some other techniques such as looking for bulged capacitors, just because it *isn't* bulged doesn't always mean it is OK, but the technique is still a very useful one. Does this only work on Vcore for example on a GPU or does it also work on Vmemcontrol, PEX etc?
Isn't the additional benefit of this test that even if it isn't the Hi-Side MOSFET that it then gives you additional information that all High-side MOSFETs are okay and the next step is to look at capacitors and the Low-side MOSFETS. Really liked the Sunday video also, lots of good information in there that I'm going to take notes on in a little bit.
Absolutely true, this is essentially testing sections of the board discretely which is a standard practice in testing departments of these very gpus. I love that this channel actually knows electronics unlike most "can I fix it" trial & error channels.
i had never thought i would see this topic (written in this fashion) on "your" channel at least :) . ive seen this topic on so many electronics repair channels and every time i see it it hurts a little. i dont think one can inject voltage. current is what you are injecting. Voltage is just a potential difference which facilitates this current (flow of charge). irrespective of the voltage, the current is what makes the component hot ... or blows it up.
Thank you, very educational video, my big question is how to apply voltage if you dont have the 12v conector (medium or low tier graphics cards that get 12v directly from PCI-E or AGP slot)
So this works because by placing the black lead before the vcore it drives all the amps back to your power supply (ground) and not through the vcore? Whereas putting the red lead by vcore could push amps through the vcore to some other ground and destroy it.
There is a better way of injecting the full 12 volts needed. Put the 12 volts needed by the circuit at the normal entry for the circuit. But instead of straight in add a series current limiting resistor in line with your power supply that you computed would limit current to a safe level. In that way you can run an entire device on its regular supply voltage without damaging the circuit. If you have a thermal imaging camera it will trace out the entire line to the shorted component. You'll be able to track any short within a few minutes.
I know it is off topic, but you seem to know everything I want to understand and I thought about this. When you damage a pad or have had a pad damaged, using replacement pads, how do you glue them down! I have put them in place and put uv protection under and around the pads and activated them with the light, everything is hard but the pad is not glued down! Any help? What is the order that things happen on the board? The power goes to the bios to the gpu to the bios to read and then to the gpu then the Controller chip then the pwm chip to the mosfet? I am so confused and no one seems to say it directly to make it what order it happens and what is being sent when it is and why. I hope you understand my question. Thanks for the videos and your knowledge. Have a nice day.
No worries, I'll answer best I can 1. I've never actually used replacement pads so I can't answer that directly. However when I have for example had to replace tiny SMD components where one of more pads have come off, I usually take a little spot of epoxy resin, apply it to the PCB with a needle, then glue the capacitor or resistor or whatever down. Leave it overnight then I can solder thin wires to replace the broken tracks/pads. maybe that will help 2. Funny that 'no one seems to say it directly to make it what order it happens', hmmm maybe you just didn't see this video 😉 ua-cam.com/video/SmVdl8kew_4/v-deo.html Especially Chapter 6 and Chapter 8 which I believe answer your question directly but I highly recommend you watch all of it, in a few sessions as it is quite long. Since I made this video about 6 months I learned more stuff myself and probably should, no call that *I will* make a follow up at some time.
@@LearnElectronicsRepair thanks for taking the time to answer. I must have missed the videos you mentioned. Thanks for the link. I will start watching these videos in the morning. Thanks again for your help. To you and your family, stay safe and well.
I'm not having much luck with voltage injection. My GPU is a Radeon R9 280, if I inject voltage directly into the 12v rail, no current is drawn. Only if I inject voltage directly to vcore (gnd to main, positive on one of the electrolytic caps on vcore cap) it draws 1amp. This is with 1.1v injected, i'm not comfortable going above that. I probed the board with my hand, nothing gets hot, only warm ish component is the GPU but thermal camera doesn't read more than 32C. I do have several mosfets shorted on vcore, but they don't get hot not even luke warm. Any suggestions?
Hey man, I have a dell rtx 3080 non lhr that crashes under load. I suspect it may be a faulty memory bank issue. Would you be interested in repairing it for me?
Sure - emai me electronicanaria@outlook.com I offer a no fix no fee repair service and every subscribers GPU will feature on video whether I can repair it or not, but I can pretty much guarantee I can tell you, even if I can't fix it, what is wrong with it and why it can't be (economically) repaired
Yesterday I was very confused when you were inserting 12 v between power and ground, knowing high side mosfet is broken. If the gpu was not already broken it is now, because 12 v went directly from a bench power supply to the gpu ... I was wondering why you didn't use the method you showed today. But let's be reasonable, there is a 99.9% chance the gpu is gone anyway
@@kriswillems5661 I never inserted 12V between power and ground on this video or the previous 'yesterday' one. I only inserted 0.8V. Check it again, the link is here. ua-cam.com/video/8BlKgTydPk8/v-deo.html I did use both the $1 short finder and the VC480C+ to trace the short (both of which actually inject around 4.5V) but because of the current limiters on these devices they are safe on Vcore and other low voltage power rails - and I proved that on this video ua-cam.com/video/XEbWyQgX_Mc/v-deo.html
@@kriswillems5661 Hmm you should have watched the video. He didn't inject 12V into the GPU 😂. Anyways you're right a broken high side probably means a toast gpu core.
I wonder if you should cut the small segments into small independent video for those with small attention span. For example. Using esr meter to find shorts on gpu vrm (maybe even use that super cheap esr meter kits)
zeferrum Tell you what, I'll put chapters in video the description. Honestly though anyone with a 'small attention span' really is not suited to electronics repair work. It often takes long periods of concentration and the ability to put many facts together to come up with a plausible diagnosis.
@@LearnElectronicsRepair I am only after getting you views so that you get rewarded for all the time you put in. The videos are priceless to me and I watch from end to end. Usually a few times.
@@zeferrum No worries mate, thatnks for your support. I have been putting chapters in some longer videos recently. I'll try a bit of an experiment putting chapters on some videos and not on similar videos to see if it increases the view hours 🙂
@@therealb888 OK I wil lstart doing that - and also see if it increases the watch hours as it does add more work to video editing. I have dome it already on some videos as you know. I think it also depends on the topic. Some videos lend themselves to a structured chapter by chapter viewing. Others quite simply you have to watch the whole thing to understand the fault diagnosis technique invoilved 😉
Wow, that was me suggesting to put the black probe on the coil for the voltage injection method. I'm not really an electronics expert myself, I only do some basic tinkering, but when you were showing the method I was immediately thinking "why isn't this working, it should be possible", and then I remembered my school teacher: "The thinnest wire gets hot." Or in other words, the heat is dissipated at the component with highest resistance, which in the initial setup was the GPU itself, so it should work if the GPU is removed from the circuit. Thanks for the follow-up and not only confirming it works, but also explaining why it works.
Edit: so to clarify, there were two reasons why it wasn't working initially.
1. The higher resistance of the GPU was limiting the current through the shorted MOSFET, and as you said you couldn't increase the voltage because you'd risk damaging the GPU.
And 2. Because the GPU has a higher resistance than the shorted MOSFET, even if it was safe to inject 10V, you would only see the GPU getting hot and possibly misdiagnose a defective GPU.
Very brillant idea.Great explanation
Many many thanks.
Alvin853
You may not be an expert (who can truly say they are really, there is always more stuff to learn with electronics repair) but what you did here was apply some excellent logical thinking to the problem for exactly the right reasons. You were absolutely spot on and as soon as you said it I knew you were right. Hats off to you mate, you taught me a trick here. 😀And I suspect you taught it to a lot of other too.
I think a big thumbs up for Alvin is well deserved guys. Well done, this suggestion was pure Gold! OK so I proved it but the credit is all Alvin's 👏👏👏
I pinned this comment so everyone knows who thought of it first😀
Thanks for suggesting and explaining!
@@zeferrum nice but isnt modern driver that has both mosfets inside hi and low side, how to be safe not to blow anything
That's actually really smart of you if you aren't an expert or experienced in electronics.
A friend showed me your channel and my gosh is there a lot of content on here! Keep it up my man your knowledge and explanations are fantastic.
Thank you. And your friend!
Amazing channel , im not native on English, but here i can understand methods much easier! Regards from Brasil.
Click CC to turn on subtitles then click Settings -> Auto translate to set whatever language subtitles you like. That's how to do it on PC, not sure about mobile but it must be possible.
It sometimes take an hour or two for subtitles to appear on a new video.
@@LearnElectronicsRepair thank you Richard! The way you explain make it simple , step by step i use CC when i have doubt only 😊
well done for this idea. I already do something similar, but I still use safe voltage (0.8V). Taking the GPU core itself out of the circuit allows more current to flow and makes the shorted mosfet hotter even at 0.8V, allowing me to easily find it on my thermal camera. With good thick PSU leads I end up with 5-6A at 0.8V which is always enough to find the short (sometimes even with the IPA method and no thermal cam) all the while keeping the GPU perfectly safe. Thanks for bringing attention to this method, and bravo for another great video
Just discovered you're channel and i love how you explain things, i have started electronics repair a year ago, and i love to learn from experienced people like you. Keep up a good work.
Excellent idea! Never thought about this method! 👍
I'm Happy to show you😉
Hi loved your concluding comment That was a Short 1 for me.funny. thanks for your expertise, we all love to learn. Regards Stephen in Australia.
Hehe this time I made joke without even realizing it 😉
There's a way to check if gpus or other big chip are destroyed by internally burning up, which does not always work. If you put your the red lead of your multimeter on ground and the black lead on the chip power input (in this case the vrm output), and you put your meter in diode mode, you should see about 0.3v. This is because all chips have inverse polarized protection diodes on their power supply. These protection diode tend to blow up together with the rest of the chip. But as said, it doesn't always work. Sometimes the protection diodes are still ok.
That sounds interesting, like some other techniques such as looking for bulged capacitors, just because it *isn't* bulged doesn't always mean it is OK, but the technique is still a very useful one. Does this only work on Vcore for example on a GPU or does it also work on Vmemcontrol, PEX etc?
Isn't the additional benefit of this test that even if it isn't the Hi-Side MOSFET that it then gives you additional information that all High-side MOSFETs are okay and the next step is to look at capacitors and the Low-side MOSFETS. Really liked the Sunday video also, lots of good information in there that I'm going to take notes on in a little bit.
Absolutely true, this is essentially testing sections of the board discretely which is a standard practice in testing departments of these very gpus.
I love that this channel actually knows electronics unlike most "can I fix it" trial & error channels.
i had never thought i would see this topic (written in this fashion) on "your" channel at least :) .
ive seen this topic on so many electronics repair channels and every time i see it it hurts a little.
i dont think one can inject voltage. current is what you are injecting. Voltage is just a potential difference which facilitates this current (flow of charge). irrespective of the voltage, the current is what makes the component hot ... or blows it up.
Thank you, very educational video, my big question is how to apply voltage if you dont have the 12v conector (medium or low tier graphics cards that get 12v directly from PCI-E or AGP slot)
Brilliant... Thanks 🙏👍 for sharing ... Greatings from Portugal 🇵🇹 🌟
You're welcome
Thx for a very good video with great information.
very informative. thank you
another great video thanks again for your knowledge
Excellent information!
A thermal camera is very usefull!
You can also measure the voltage across each coil, to detect shorted current.
So this works because by placing the black lead before the vcore it drives all the amps back to your power supply (ground) and not through the vcore? Whereas putting the red lead by vcore could push amps through the vcore to some other ground and destroy it.
Current will flow only when we have shorted mosfet? On not-faulty VRM current will by 0A?
Is this injection method still save in case of short circuit is on low side MOSFET ?
There is a better way of injecting the full 12 volts needed. Put the 12 volts needed by the circuit at the normal entry for the circuit. But instead of straight in add a series current limiting resistor in line with your power supply that you computed would limit current to a safe level. In that way you can run an entire device on its regular supply voltage without damaging the circuit. If you have a thermal imaging camera it will trace out the entire line to the shorted component. You'll be able to track any short within a few minutes.
I know it is off topic, but you seem to know everything I want to understand and I thought about this. When you damage a pad or have had a pad damaged, using replacement pads, how do you glue them down! I have put them in place and put uv protection under and around the pads and activated them with the light, everything is hard but the pad is not glued down! Any help? What is the order that things happen on the board? The power goes to the bios to the gpu to the bios to read and then to the gpu then the Controller chip then the pwm chip to the mosfet? I am so confused and no one seems to say it directly to make it what order it happens and what is being sent when it is and why. I hope you understand my question. Thanks for the videos and your knowledge. Have a nice day.
No worries, I'll answer best I can
1. I've never actually used replacement pads so I can't answer that directly. However when I have for example had to replace tiny SMD components where one of more pads have come off, I usually take a little spot of epoxy resin, apply it to the PCB with a needle, then glue the capacitor or resistor or whatever down. Leave it overnight then I can solder thin wires to replace the broken tracks/pads. maybe that will help
2. Funny that 'no one seems to say it directly to make it what order it happens', hmmm maybe you just didn't see this video 😉
ua-cam.com/video/SmVdl8kew_4/v-deo.html
Especially Chapter 6 and Chapter 8 which I believe answer your question directly but I highly recommend you watch all of it, in a few sessions as it is quite long. Since I made this video about 6 months I learned more stuff myself and probably should, no call that *I will* make a follow up at some time.
@@LearnElectronicsRepair thanks for taking the time to answer. I must have missed the videos you mentioned. Thanks for the link. I will start watching these videos in the morning. Thanks again for your help. To you and your family, stay safe and well.
Amazing!!
I'm not having much luck with voltage injection. My GPU is a Radeon R9 280, if I inject voltage directly into the 12v rail, no current is drawn. Only if I inject voltage directly to vcore (gnd to main, positive on one of the electrolytic caps on vcore cap) it draws 1amp. This is with 1.1v injected, i'm not comfortable going above that. I probed the board with my hand, nothing gets hot, only warm ish component is the GPU but thermal camera doesn't read more than 32C. I do have several mosfets shorted on vcore, but they don't get hot not even luke warm. Any suggestions?
Hey man, I have a dell rtx 3080 non lhr that crashes under load. I suspect it may be a faulty memory bank issue. Would you be interested in repairing it for me?
Sure - emai me electronicanaria@outlook.com I offer a no fix no fee repair service and every subscribers GPU will feature on video whether I can repair it or not, but I can pretty much guarantee I can tell you, even if I can't fix it, what is wrong with it and why it can't be (economically) repaired
Heya yesthat is a better point to put the input voltische well done @alvin853. and great video again
great!
But if the high side drain mosfet capacitor short and not the mosfet???? they burn the cpu cause the mosfets block the corrent?????
Genius....
Yesterday I was very confused when you were inserting 12 v between power and ground, knowing high side mosfet is broken. If the gpu was not already broken it is now, because 12 v went directly from a bench power supply to the gpu ... I was wondering why you didn't use the method you showed today. But let's be reasonable, there is a 99.9% chance the gpu is gone anyway
This board was not fixable to start with. A broken high side mosfet almost always mean you're done - no fix possible.
@@kriswillems5661 I never inserted 12V between power and ground on this video or the previous 'yesterday' one. I only inserted 0.8V. Check it again, the link is here.
ua-cam.com/video/8BlKgTydPk8/v-deo.html
I did use both the $1 short finder and the VC480C+ to trace the short (both of which actually inject around 4.5V) but because of the current limiters on these devices they are safe on Vcore and other low voltage power rails - and I proved that on this video
ua-cam.com/video/XEbWyQgX_Mc/v-deo.html
@@kriswillems5661 Hmm you should have watched the video. He didn't inject 12V into the GPU 😂. Anyways you're right a broken high side probably means a toast gpu core.
Sorry, my bad.
@@kriswillems5661 np
I wonder if you should cut the small segments into small independent video for those with small attention span. For example. Using esr meter to find shorts on gpu vrm (maybe even use that super cheap esr meter kits)
zeferrum Tell you what, I'll put chapters in video the description. Honestly though anyone with a 'small attention span' really is not suited to electronics repair work. It often takes long periods of concentration and the ability to put many facts together to come up with a plausible diagnosis.
@@LearnElectronicsRepair I am only after getting you views so that you get rewarded for all the time you put in. The videos are priceless to me and I watch from end to end. Usually a few times.
@@zeferrum No worries mate, thatnks for your support. I have been putting chapters in some longer videos recently. I'll try a bit of an experiment putting chapters on some videos and not on similar videos to see if it increases the view hours 🙂
@@LearnElectronicsRepair Please do put chapters in your videos, I do watch the full video but structuring it with chapters helps to learn better.
@@therealb888 OK I wil lstart doing that - and also see if it increases the watch hours as it does add more work to video editing. I have dome it already on some videos as you know. I think it also depends on the topic. Some videos lend themselves to a structured chapter by chapter viewing. Others quite simply you have to watch the whole thing to understand the fault diagnosis technique invoilved 😉
12:40 - I _would_ be able to see the results of your finger test if the that coil got much hotter than it is!
Brillant!!!!
Thanks
1660 super repair guide plz
Cool..
❤️❤️❤️❤️
Voltage isn't "injected", it's applied.
@G Absolutely correct - voltage can only be applied. Current can be injected 🙂