Gigabyte RTX 3090 Graphics Card Repair - No Power and Not detected Part 2
Вставка
- Опубліковано 10 вер 2024
- ⭕ Ask questions and Engage in our Forum at northridgefix....
⭕ Need a repair? northridgefix.c...
⭕ Visit our ecommerce shop and buy all your tools at northridgefix.com
⭕ We are an Amtech Flux Distributor. Buy AMTECH NC-559-V2-TF Flux
northridgefix....
Microscope northridgefix....
UNI-T UTi260B thermal Camera
northridgefix....
Grinding pen
northridgefix....
Low Melt Solder
northridgefix....
NorthridgeFix T-shirts
northridgefix....
Weller Soldering station ebay.us/1mvgt5
=================================================== If you are local, drop in and say hello
NorthridgeFix
19365 Business Center Drive, Unit 7
Northridge, CA 91324
Phone: 818-668-7058
NorthridgeFix.com
Probably a dumb idea but does it need the heat sink on for that model? GPU shutting down from over heating?
yes but it wont stop the pc from posting.
I doubt that.
These cards are capable of running as low as 20W as I've seen in HiveOS when my 3080s sit idle. The card / GPU is not supposed to be heating up so intensely pre-post in the first place, but the thermal limits in place should compensate for no fans/cooling and simply downclock to a point where the GPU doesn't heat up as much as it did in the video.
I do think there may be some new components on the 3000 series for northwest to learn, but I have no doubt that with time, more of these repairs will be completed.
We did re-assemble the fans and heatsink but no change. The chirping sound is not normal. All video cards should be able to post without connected fans or heatsink. GPU heat is aggressive on this one and causing the entire board to heat up. This is abnormal based on a lot of working video cards I tested with a thermal cam.
So it might, but honestly the GPU should be able to POST and should really be running it's "2D" mode so to speak and draw much less power and generate WAY less heat than his gun was detecting. The fact that it went full throttle for thermals indicates the GPU fried and its basically just running a maximum current but no respect regarding how to use the current computationally speaking.
Most GPUs will get to the point of displaying the bios splash screen before shutting off, even without any cooling. Since the card didn't output anything, and was heating up, it must be an issue with some chip.
4:59 one of the low mosfets to the left side (ones that feed memory chips) isn't heating the same as the others, meaning there might not be voltage comming from it. Bottom low mosfet to the right isn't heating, probably means no tension comming to it. The RTX 30 series won't start unless all tensions are present.
At 5:19 you can see there are 2 mosfets on the left side thats not heating. The top one and the one at the 3rd from the bottom.
@northridgefix definitely worth further investigation
Thanks to your videos i found the courage to fix a macbook where the ram needed some reflowing. Even though I used the cheapest tools available since I most probably not going to do much more repairs other than this macbook the happiness when you haear the chime is priceless. Keep up the good work even if some issues cant be fixed with the limited time you have available. Just know that you inspire some viewers who try to gain skills in repairing their own stuff and maybe get them working again before they throw stuff like laptops into the trash.
To the comments that the lack of a pwm return signal from a fan may be causing the shutdown that likely isn't the case. GPUs don't have that sort of feature - otherwise mounting a water block would be more involved. Now, as to whether it is overheating or not. Yes, it will hit a thermal limit, but the card should aggressively throttle itself to under a dangerous temperature. However, the lack of any heatsink to at least disperse the heat would present a potential for runaway temps thus leading to thermal shutdown. I'd recommend mounting a temporary heatsink to at least draw some of the heat away. I don't think the gpu core is dead, a dead core wouldn't likely attempt to start at all and not produce any noticeable heat. That said, when you first looked with the thermal camera a couple of the mosfets did heat up significantly quicker than the others. Perhaps there is an additional short on those particular vcore lines.
I think you might be onto something. I like your thought process.
The world needs a "GPU Whisperer" right now!
I know you can do it, keep trying, cause you got more graphic cards than Amazon!
Loving the videos!
Props to you sir. Most electronic repair shops I know of won't even take graphic cards because the success rate is too low in repairing them. However by doing this you are learning more and more and I can see your success rate rising and rising. With the current GPU market, repairs are a big thing and I only see it growing.
How do you have any idea at all what his success rate is? lol. I love these braindead comments more than the videos.
As it should be when you invest that much money. I hate the current GPU market but its also making people much more thankful for what they have.
@@th3batfink He is naïve and is not factoring in what 300-500k views accounts for ;)
It was not filmed by accident.
I learn a good bit from No Fix videos. Seeing how you go ahead and stop before being frustrated. I need to do that sometimes. Be less bullheaded and step aside for a minute to make sure I’m thinking clear
No fix = yes subscribe. You are a knowledgeable man. You have gained another follower. Respect, Sir.
5:04 On the center left had less heat and center right more heat, could be an issue?
On the left of the main GPU chip there is a cold spot between similar areas getting hot. Maybe some other component fried and that area does not get power
i spot the same thingy look like one of mosfets there is ded i think
Yeah I noticed that as well. One of those chips was not heating up like the others.
There's three cold power phases and that's normal because those three are for memory according to TechPowerUp review. That hot one is more interesting but maybe replaced chip just is different than others. That may be a problem if hot phase overheats or VRM controller notices the difference and shuts off with that sound.
@@birdman8649 i thought that was odd as well
Reflash the cards bios. Sometimes it gets corrupted by hardware failures.
That’s what I was thinking
Everytime you point with the hot air gun to tiny components I'm afraid for them to fly away haha ^^'. I was also optimistic in this repair, problem was spotted quickly and removing the sort looked promising. Keep it up master!
The middle inductor and controller gets a bit hotter than the rest. The lowest one remains stone cold.
I noticed that too
5:17 - On the left, in the array of components, one of them, the 3rd from bottom is strangely cold. Is that normal? Maybe a open mosfet?
5:03 Why there is one component colder than the rest (left side) and one is warmer (right side)? Could this be the cause?
I saw that too but im assuming that he knows enought to rule that out, cant say for certain though.
I went to a tech college for 2 years after leaving school. One half the building taught Electronics the other half was Computing - you couldn't mix you had to choose one - I choose computing. However, seeing your smiley face whenever you fix a problem makes me wish I choose electronics. I never do a smiley when I find / fix a bug .... mainly because I'm the one that created it.
I believe your assessment to be correct. When power supply MOSFETs fail in a shorted condition they will generally take out the component they were powering. The fuse was probably rated to prevent the trace from burning the board down, not to save the components, so that the board could get certified.
Hi Sir. I find your videos and their repairs excellent. I see it from Argentina, a city called Tigre. I wanted to ask you not to forget to activate the subtitles and it helps me a lot to understand some parts. Thank you very much and a hug from Argentina. Keep it up with the videos.
Try putting a Heatsink and a fan just to make sure that it is not shutting off because of thermal throttling and because it can't detect a fan. Maybe it is worth trying. :)
Did you replace the blown fuse at 6:15 of the first part of the repair?
He jumped it with a cable
Temporary soldered a wire
Hello sir , I know you don’t me but I’ve become a big fan of your work ! All jokes ro another time I’ve done some 3090 myself and I found it was a channel mosfet it could be any of the main power rail to the gpu! If like just test them before you give up ! Thank you 🙏 sir your stronger then this
What do you mean?
I have the same card with exact same blown vrm (same spot)
@@Dundunbra1997 this seems too be happening alot with these 3090s. Everywhere I watch 3090s are having issues in alot of ways. I could be seeing something that isn't that widespread but seems to be more than the other new cards.
Open mosfet, the cold one they mentioned could be messing with the signals, test with oscilloscope to see what you have.
That's got a FRIED GPU when those Vcore MOSFETs fail the GPU gets COOKED a good portion of the time going from my past experience.
sorry but i dont have the same ideia, +30 rx490, rx570, rx580 gtx 970, gtx980 all mosfets burns replace, some time replace CONTROL IC TO, and that done work again... in last 30 only 2 die after 2 day (in burning test, and bench), "core MOSFETs fail the GPU gets COOKED" some times it open to plus, but that i do find very ofen
Everyones talking about the 5:04 chip being cold on the left so hopefully he see's
One chip on the left (i think a memory one) didnt seem to generate any heat while all the others did when you turned on the card. Could it be that this chip is faulty? Hope you can figure this out
Yea, that's a possibility. But it could just be that it isnt needed, so it isnt currently working.
Maybe card detect no fan spinning and turn immediately off. Alex try connect graphics card fans ....
Yeah... Even I thought the same. He should try connecting all the fans. Bcz due to no fans connected the card turns it self off to avoid any thermal damage to GPU or components
someone get this comment higher. Newer cards do this and auto shut off to protect damaging component.
I would ask to put the card back together
Might be. I've seen videos of people saying without a heat sink on the gpu it will get too hot very fast, so I could see the manufacturer putting in a shut down feature if it does not detect fans. they need to put it back together to ship it back, so might aswell throw it back on the motherboard and see.. these new cards draw some big power and I could see a manufacturer putting a safe mode in place to stop damage.
Maybe alex should give this a try. „Auto turn off“ That could be the reason
@@herpderp7791 they will shut off but even on 30 series there is still enough time b4 they do to post over on tech graveyard he does this same thing all the time even on 30 series cards without any issues so odds of that being the problem are unlikely
Just guessing: check components under the thermal pad on the left of the GPU. It seems one of those transistors is not turning on...
Sometimes vrms could partially work, i. e. you might get the correct voltage and normal heat output under thermal cam, but the signal could be scrambled. The only way to test it is with oscilloscope. That would determine if any vrm is working properly or not.
Watching from philipines🙂 thank you for your tutorial. It can be lead me to become a pro.☺ love u so much. God bless you all the time.👍🏻
At @5:17 if you look on the left you'll see only one component not turning on. Can it be it? Ad can you look with thermal on the back side in case you see something unusual there
After watching your analysis and investigation. I think you made your conclusion.
i might not be correct , but @5:06 , there is a component which is heating more than others.. maybe replace the two components there? at the right side
You have a professional attitude, knowing when to stop due to your knowledge. I congratulate you on that as you not only know your limits but you also expand your knowledge and limits.
Thank you for these video's, really enjoy watching them.
I'm wondering why the customer didn't just RMA it through Gigabyte, its still under warranty. Are you going to re attach the original bad mosfet so that the customer can try to RMA it?
probably when he send claim RMA, they will return his money or replace with last gen series. That what happened in my country
@@akagaminolip wich sucks alot, in a normal situation I doubt anyone will try to repair when you could swapped it for the same card... but due to the shortage the best he could get is a money return... wich sucks since that card costs more now... or a replacement to something worse and some store credit....
I have a hunch there's still some liquid metal under the chip and it would take reballing to fix it.
What is the name of the power mosfet you replaced?
Thank you for the video again.
The crying emoji on the thumbnail told me what the outcome was going to be, no suspense in this video unlike the others :)
Knowing how expensive these rtx3090 are I realise someone somewhere has had a really bad day when this card popped.
Cheers.
Arctic Silver 5 is looking better than ever now!
Do those GPUs only need the additional PCI-E cables for the fans? You didn't seem to be giving it any power apart from the PCI-E slot. Still think the customer should've just RMA'd it, that's the whole point of a warranty
Maybe the 12V rail shorted to ground through the GPU, frying it.
Hey Alex, have always enjoyed your video.
Could be a BIOS issue. Try to switch to secondary BIOS if there is a switch on the board. If not, put the card into the system as secondary display, if that prevents PC from booting. Can always plug in after system post. Check if detected in Windows. Once booted, can try flashing new bios.
I looked at 4 times when you increase the Temperature boundary you can see it better... 5:27 Rightside 5th chip down is hotter then the others....Next left-side Column 1st chip not turned on and 6th from top..... Its not coming on......
Yes I noticed this
تحياتي الك من فلسطين
أنا بكتب هاد الكومنت بالعربي لأني تفاجأت هداك اليوم إنك عربي الأصل وبتحكي عربي علشان هيك حبيت اكتبه بالعربي احترام للغتنا والك
بصراحة أنا بتابعك وبتعلم كثير منك فتحياتي الك كمان مرة
On left, I see cold spot (3.rd from bottom) and on right side on the middle it gets more hot than the rest of them.
Oh boy I been waiting for this one!👍
I would check the power draw on the 12V GPU connector and compare it against a normal card's consumption profile during POST. If the card is drawing a way more power, then the GPU is shorted (yet the 0.7V-ish Vcore does not indicate that.. unless the rail supply can keep up with the "in-short" amps).
Hey Alex, when you press down on a chip to squeeze out the excess solder like in 1:45, how do you make sure that there isn't a short underneath (eg between the large pads or some of the pins)? Best regards, Jakob
Use flux
This is why you buy with warranty, £3k completely blown because the guy couldn't wait jesus...
Surely less than 12months old...should have been RMA'd no?
@@Tink-GB they don't honor warranty if your not the original owner... would of been better of putting £3k into some stocks or investment... better return.
@@kleanthisgroutides7100 why's it matter who owns it? If it's under warranty but they won't fix it then they're scammers imo
@@friendly0 so they are able to trace the cards and discourage selling probably.
@@kleanthisgroutides7100 I'm not even sure that's legal here, because selling things is legal, and it the card is only a month old and has 2 or 3 years warranty and they won't honour it... Well that's what a scammer would try do 😬
Hello. I think a component is missing under the capacitor at 2:45. I you look at the rest of the MOS there is an 1R8 Resistor, a capacitor and under it another component. The rest of the MOS have it but the one you use the which doesn't have it.
Check at 9:02 on part 1 video. It doesn't have any component. It's just empty. There is 1R8 resistor surrounded by 2 caps each. That's enough. 👍
@@ErickBuildsStuff Thank you
Part 3 waiting to be continued
Hi Alex, It would be great to see you do more graphics card repars. There are not many video's to find on this.
This is a dual bios board - there is a little micro switch at the top of the board, as one last ditch attempt it might be worth flicking the switch to the alternate position to engage the alternate bios (silent mode). Probably won't work as I believe the GPU has an internal short from the 12v shock it would have received on vcore when the mosfet shorted - but it's worth a shot before condemning the card to the trash.
I believe that you have been mistaken with the chips on 1:31 , the previous chip no was AL000U1N , the the new one you putted was AL000H1L .. maybe it's the cause of the problem !!
Big fan of you ..
Hoping of the best for you ..
some graphics card are disposable i think, but this man is great i love your work keep it up.
Good to watch, far too fiddly for my fat fingers, this fellow does a good job
I believe the GPU is faulty. I would ask big boss to buy you a BGA reworks station? Costs a lot but worth it I guess with the amount of graphics cards you have for repair.
My EVGA FTW3 ultra had a sudden death playing GTA5, after a year getting it. RMA secured me a new one. Works so far. These cards.. I hope they don’t push too hard for the 5nm cards.
One of power stages are off (near to video slots) and one ofpower stages at back side ar hotter than others! Why?!
The memory rail should be 1.35 volts. That's spec for DDR6X. You only got .9 volts. Card will shut down because the memory is not responding. Check the drmos, caps and resistors feeding the memory rail power.
I do not believe this is True, at least for some 3090 cards, because people are reporting ~0.85v for memory.
it would be interesting to know how many rtx 3000 series cards have passed through your shop since they were released
at 5:00 it seems one of the DrMOS chips remains off.
Honestly am new in this but I did not see you removing that loop wire you did
Yes I agree.
Hello
i have AMD rx580 sapphire 8gb work just fine but when i install the driver IT SHOW'S Exclamation mark (!)
it detected but it doesn't work it show's Exclamation mark on the driver (!)
i need to know what to do.
Thank for sharing all your work!
Mr. Alex, just ask Big Boss to assemble the GPU and try it out. Might not work, and two long beeps normally indicate RAM IC failure or hardware test failure... But still, worth the shot. Thank you for everything you teach us!
Look at 5:04 less heat on center left, is the ram chip?
@@rodolfogoncalveslicatti2433 Seems like it, maybe replacing the ram ic should do the trick. But it's kind of hard to tell. But you can see that some components do not heat the same as others. That leads me to think that there is more than one problem.
Most likely a problem witht the gpu chip itself when during its operation when the mosfet goes bad it short drain to source so it send a highvoltage full 12v on the gpu die instead of 3.3v or lower, because the mosfet is like a switch and in dcdc buck converter mode
So my RX 580 8gb sapphire nitro gives no display. Pc boots on(I can hear windows desktop etc), but no display, gpu rbg lights turn on but fans don't spin. So any idea what could be the issue? Thx for reading
Zero rpm mode, so fans probably wont spin. Do you see the bios at startup? If so then it is a driver issue on your windows.
@@JJ21lives4ever nope no display at all. And I think fans boot on PC boot and then stops, but in my Case they don't spin at all not even on boot
Try other ports. Usually far left is first output.
@@makaveli1337_ if there is no spin at boot, then it may not be getting supplemental power. Is the power connector properly seated?
The rx 480 had a feature to save energy, fans only spin when needed. If u using video adaptor or dont have a sexond monitor to test, wait until windows start. A cheap Post Card Test can tell u something. Also try clean with isopropil
Wasn't there a component on the left side "not turning on" at all?! Would it make sense to check that?
Thanks for your videos, I learnt a lot about..
When you were looking under the thermal camera, I noticed at 5:10 that all the mosfets are warming up except for 3rd one up on the left hand side hot hot and then cold? is that mosfet even passing current?
This is obviously a very old video but I've been going back and watching through all of them. When he used the thermal camera on the GPU mounted in the motherboard, I noticed there was a spot that wasn't getting as hot as the rest of it on the left side. Looked like it could have been it a MOSFET. Could that have been the issue here that something was not getting power correctly?
Does anyone know how much the cost to fix gtx 3000 series ?
That title pic to gold 🤣 you really got the algorithm 😎
On the left side. When you turned the board on, all of the squarish components in a row got a heat signature except for one. What's that about?
your an artist man i learn a lot from you
Have you tried to use the Nvidia Modular diagnostic software (MODS) I see this program helped alot another channels to diagnose problems on the board, memory problem and i think gpu chip as well. Hope it helps
How much would a repair like this cost? Just curious.
Probably has a base rate based on product plus hourly labour rate. Then parts + complexity of issue on top. So hard to give a predetermined number
50 usd
He’s an extremely skilled technician with probably decades of experience. So he can rightfully charge a hefty price if he wants to. He does great work
I got quoted for £120 to repair blown capacitors on my 980ti lol
Is it possible that the drivers or smth flashed on the GPU is causing it to fuck up? Or is that only something that can happen once an OS is loaded?
under thermo cam you was focusing on what was getting hot but to the left of the cpu in the row of chips that was getting hot there is one that clear wasnt getting power 5.25
Question why do you hold your hot air gun at a slant angle instead of 90°. Also did you find people to help you in your work thanks in advance
What GPU brand fail the most or that you have worked the most in you shop? Thank you once again for the great videos and superb repairs!!
Yo guys connecting the fans does nothing. I mine and all my cards get 120MM fans, I basically remove the stock fans to preserve them. I've older and new cards, connecting the fans doesn't matter nor does the heatsink. If the card is protecting itself from damage, it'll just shut off and nothing else. It would POST without the heatsink and the fans.
Love and respect from LIBYA 😘😍😍😍😍
There is a 3rd rail called Vuncore that does memory controllers and caches also, check that the switching controller ic's are producing a pwr_good signal
yeah also seen one cold chip at the left side 👍😀👍
Is it possible to swap out the current GPU with another one from one of your spare cards?
when i growup i want to solder like you.
Did you guys noticed that at 5:26 the first "mosfets " on the Top left and the 6th are not heating is that normal ??
awsome work
It feels very sad when you remove those beautiful decorative solder bridges. 😭
Any chance it was the 3rd coil down on the right hand side of the Main chip it heated a slight bit when the thermal cam was on ? Unlikely I guess but worth a quick look
I always wonder why somebody doesn't just RMA such a card. It still has warranty. Now after this repair it does not.
At the 5 min mark, you can see 1 of the mosfets on the left size is cold. Maybe you are only looking for hot components but you should also look for cold components. There is
hi Alex this is samer from australia i watching your channel pretty much every day you are very professional technician and having a professional tools as well i wonder why you don't bring the Automatic BGA machine the WDS-620 and try to go for level up and repair the CPU'S and GPU's that is right it is bet expensive but it will do the job while you doing another device i hope i can see it in your workshop in the future and see videos for replacing GPU and CPU.. thanks regards
Awesome repair
You know.... I've heard the term wick off before, but you literally used a wick to get the extra solder out.... I feel dumb lol
Sorry, I may be wrong, but, seems like the bottom mosfet has heated a little more than the others before the gpu start to heat. Maybe it could be partially faulty?
i guess its the controller chip that sits on the corners of the die (at least at the gtx 1080) that failed. Is the glue around the die discolored on the corners?
I still don't understand why there was no RMA for such a new card.
In the previous video the customer said it was 2nd hand and no receipt. But that doesnt make sense either, since Gigabyte allowes transfer of warranty on resale. You just have to give them the serial number and they calculate the warranty from the date of manufacture.
No 30 series cards should need to be sent for 3rd party repair unless you modded them or otherwise did something that voids the warranty.
did you inspect the back of the board with the thermal camera?