For repair, please contact me by following the link in the channel ABOUT page. Buy me a candy at paypal.me/tonynameless Tools, schematics, boardview files etc are available here drive.google.c...
@@northwestrepairthere is no way to know, only difference is amount of liquid inside, who knows if a lack of liquid or bad vacuum. cooler master does these vapor chamber. but are not for sale to public i think
@@SkateClipsAndTips Referent AsRock 7900XTX - max delta 15 C - purchased 13.12.2022 - i think the problem was only with "original" AMD cards, other referent vendor cards are ok
The solution is to contact AMD and have the card replaced. It's a known issue acknowledged by AMD and they should replace the card. The less ideal solution is to install the gpu vertically. Moving the hot card to a vertical position won't work, the card needs to be on a vertical mount position to start with. Der8auer tested this in his channel; Vertical = no issues, moved to horizontal while running = temp rises to 110c, moved back to vertical = temperature stuck to 110c. The issue is insufficient liquid inside the thermal chamber and happened to a small batch (according to AMD, hard to tell how big that small batch is) of manufactured rx 7900xtx reference cards.
There was an issue in the factory where too little liquid was injected in the first few batches, which was fixed after customers and reviewers found their cards overheating. The amount of liquid injected was inconsistent, so some affected cards may work in some conditions/orientations, while others won't. There really isn't anything you can do to fix that, since, AFAIK, the pressure inside is below atmospheric, so, unless you have special equipment, you wouldn't be able to reseal it. The only choices here are to RMA the card, to change the cooler to an aftermarket solution, ir to live with faulty cooling and underclock the GPU until that cooler can handle the heat
I have a 6900XT with the issue. Fixing it is possible. Open the vapor chamber port, attach copper line, inject more liquid, pull vacuum, pinch chamber port closed along copper line, solder pinch point so it holds vacuum good. Done. The vacuum can cost a bit from amazon/ebay "for just one job" but I have built them using $20 fish tank air bubblers, though i've never tested what the actual vacuum capability is.
If this is the case AMD should just go back to a conventional design. stick with the shroud thou. What was under the shroud is what put me off buying a reference card in the first place.
@@keibohow69 First of all, AMD aren't the only ones using vapir chambers, since they are great (when they work properly). To begin with, the factory that produced them isn't even theirs, it's a supplier used by others too. There's no reason not to use them if they believe they are a better solution, just need to be better with the QA next time
ONE way but only if the costumer cannot try RMA and so on... DIY heatpipe technique basically you find the vaporchamber seal open it up to fill mode liquid then heat the chamber with the heat gun until liquid vapors out fill it more then use solder to plug the hole. The +100C chamber vapor out will decrease the internal air pressure.
Awesome that you got it fixed, and glad it wasn't something wrong with the chiplet design since Nvidia is also moving in the same direction (alongside already being used for CPUs from Intel and AMD). Unfortunate that all the board views were incorrect, and even after getting it fixed, the vapor chamber was faulty. Believe a lot of the early reference cards had that issue. I want AMD to compete in the GPU market so that not all roads lead to Nvidia. Hopefully the customer is able to get the vapor chamber replaced/fixed. Always enjoy watching your repair videos 👍
Red Devil XTX here. I stuck a Bykski waterblock on it and it's all OK fingers crossed (screw the warranty). The amount of heat coming off the radiator is astonishing. This does show the setup is moving heat but I'm amazed.
AMD always change their layout with every generation, as I remember from previous videos, Happy it did not go to waist. I was bored so went back to your oldest videos and checked if there's any that I didn't watch and liked. Your videos are therapeutic tbh with the type of music and vibes you make in them!
The hotspot problem was probably due to broken vapor chamber, plus you might need to use PTM 7950, to compensate for a potentially uneven die or contact due to bad mounting. It wouldnt be of much help if the cooler itself is the main culprit though. I repasted My Hellhound XTX with PTM 7950 and my hotspot never gets higher than
I hear some release day 7900xtx had underfilled vapor chambers. On my stock cooler hot spot was 10c cooler if GPU is oriented vertically I put mine on a waterblock with PTM 7950 and works with no issues
Maybe get a cooler for cheap from someone who installed water cooling block on the card and see if another cooler fixes the issue. I also remember something about early models being under-filled. Not sure if drilling and filling would help, those chambers are filled under vacuum or certain pressure, so that liquid evaporates at lover temperature.
Early 7900XTXes have had issues with that vapor chamber, cooler simply needs replacing, there is now way around it, unfortunately :( Good job on the repair and not giving up no it!
Some of the first run of 7900 XTXs with vapor chambers had a manufacturing defect relating to not enough fluid in the vapor chamber. The problem would typically only show up if the graphics card was run vertically rather than horizontally (or was it the other way around). It got past QA because they weren't testing them in both orientations, I think. This should be a warranty claim for sure.
Tony, great repair work.. at last its alive.. My friend had similar 1st gen cards, which were on fire like this one.. getting hot very fast and crashes. He had to modify and add a liquid cooling solution, after which the card is functioning perfectly with no more issues. But you know display card liquid cooling is not a cheap solution and needs adequate space in the PC case, so may not be compatible for everyone. 👍👍
try each orientation when its cold. changing the orientation under load will only make it hotter. it wont cool down no matter which way you orient it once the vapor chamber fails because the water cant condense once its that warm i heard on a derbauer vid i think
Solution for vapor chamber is to replace it or fix it by adding more liquid. There is a port that has been pinched already, you can open it and fill through there then attach a copper line to it to pull a vacuum through. Pinch the line while under vacuum then solder so it holds, fixed.
my first 7900XT from XFX had a bad vapor chamber too so it isnt just AMD reference cards. Really liked the card though, 2900-3100mhz stable for the first few months then the cooler shit the bed.
I have the XFX 7900xtx and thought I had the same exact issue. But when I opened it up, I noticed that all of the thermal paste pumped out of the right side of the GPU core, near the memory chips. Turns out the thermal pads on the right side memory chips were too thick, causing the vapor chamber to sit uneven on the gpu core. I ended up using PTM7950 on the core, then used thermal putty on all of the memory chips. Which squishes much better. On testing, my hot spot temps went down from 85C to 72C max, under full load. On normal gameplay, I don't peak above 63C.
@@kulilin3104 I repasted and repadded it while i was talking with XFX, made damn sure i got the right thickness pads too. I've done it with multiple other GPUs with no issue so it wasnt an issue on my end, just a bad cooler.
That was factory defect on early models. If memory serves, vapor chamber had too little liquid. Depending on deficiency, you could get it working, if oriented like on motherboard, but I think it wasn't 100% guarantee. Luckily AMD did admit the issue and will RMA it for fixed model, if memory serves. Can't remember other ways to tell which cards are than throwing one in and have this issue or whether there is serial check tool anywhere.
I don’t suggest this but If you are crazy enough you could drill a hole in the vapor chamber and add distilled water and solder it shut after because the chamber and could be solder shut. The issue is AMD didn’t add enough water to the vapor chamber.
with amd gpus is even with my 6650 xt I have issues with hotspot and it having issues with like 40c deltas my solution was using PTM7950 with it. It improved temps by over 20c on hotspot alone and i now chill at around 75-80c hotspot with 65c core
I have an early 7900 XTX reference card and I apparently got lucky as mine doesn't have the overheating issues that many have. I have run various tests on it and it gets warm but nothing like I have seen here.
hint that might work with 7900xtx chamber... let it cool down to the room temp then do the 90 degree turns with the gpu while doing stres test.(before it reaches high temperatures and see if there is any differences in temps) what i think is going on there....once the gpu gets to higher temp the "water" liquid inside evaporates and there is no cycle in it to cool down the gpu core. basicaly once it evaporates no matter what u do while testing it. it wont do any difference, it has to cool down untill the vapors inside the chamber becomes again liquid.(that's my theory, idk ) its like once it evaporates...there is no liquid that comes back to the core location to cool down and then again evaporates from the heat. either there is too little liquid inside the chamber or the design bad optimized to help liquid inside that make the cycle properly (like a closed loop circuit think of a aio cooler but with steam/vapor) i wish i could do some tests on that gpu..with different coolers air cooled and watercooled...im interested with the results..
Hotspot changing all the time is due to there being multiple hotspot sensors around the core and it´s always reporting the hottest one. Meanwhile, the "core" sensor is probably in the galaxy far, far away, somewhere on the edge of the chip. Vapor chamber was an early issue, i´m sort of surprised the customer did not RMA the card - but maybe he couldn´t or just didn´t care. AMD reference cooler designs in general are quite poor - but the same can be said for some board partners solutions, especially when they reuse heatsink from different model and don´t bother adjusting the cooling for the card (look at Asus Dual 6700XT - uses Nvidia 3060ti heatsink and does not bother cooling down the VRMs on one side).
It needs a new cooling block. You can't fix the vapor chamber because there is negative pressure inside(vacuum) vapor chamber and you will need special equipment to add coolant.
I had the same problem with my reference 7900xtx i had to return the card after i got my money back i bought a sapphire pulse model. It was 10% cheaper too.
I had similar problems with the reference model of 5700xt with vapor chamber and it went worst over time until the point of thermal shut down the solution that work after try to tighten screws and deform and supplement retention bracket was put thermal pad behind the backplate around the GPU core and in behind the mem.
easy way to fix bad temps is to send back for refund/replacement or replace that cooler with different one from ebay, some of the 1st batch cards had the problem
personally on my 5700xt msi evoke before UC/UV I also have hot spot temperatures up to 110C and according to AMD this is normal. after lowering the voltage and frequency I am around 90/95C with the ventilation going from 100% to 45/60% without almost any loss of performance.. thank you amd.. what I find strange here is that the card consumes very little, the temperature problems that I had were with a consumption of 200/220W and with only two fans on the GPU plus I never had a delta so high it's crazy all these problems with amd cards (my settings are 1v vs 1.2v and 1900MHz vs ~2150MHz) I would therefore tend to recommend a UC/UV but if the vapor chambers are defective the card still risks heating up too much at times..
Feel like the issue i was having with the ebay 2080ti Founders was a bad vapor chamber maybe. not sure if they have one but I bought some nzxt bracket for it and its direct die cooled now. kept the old back plate zip tied on its jank.
I've pulled down the supply voltage to a memory chip to match fault artifacts, maybe you can do the same to show a memory fault in the test, ie, you now know which chip is producing the fault. I vaguely remember it was something about not enough distilled water in it. Gamers Nexus has a cooler manufacturing vid somewhere that might explain the water and if there is a vacuum.
Vapor chamber works on the same principles as the heatpipes do. It is literary a chamber with some liquid, something like wick and vacuum. There are supports that prevents it from crashing. In the Vacuum, water boils at way lower temperature. It evaporates and the steam is going away from the hot spot. It reaches the cold part and condensates. As water it flows back to the hot spot creating the circulation. Fact that the water is changing phases is big cooling effect. Plus this circulation is drawing the heat away. There is a capillary effect as well and the wick is helping it out. Steam travels true the middle around the wick and water travels on the walls. So those two do not collide and are not stopping each other. Solid peace of copper might look like better solution but it is way less effective than heat pipes or vapor chamber. It might also look like vapor chamber is a saving of money due the way less copper used but the machining makes it costing more. When heat pipes or vapor chambers loose vacuum they become useless. All you have then is small amount of copper. Efficiency tanks wit to little or even to much water in the system as well. First thing to check is soldering joint on the filling port. If it is compromised there is your answer.
DIY HEATPIPE construction videos could help a solution to refill the vaporChamber. Basically fill half volume heat up until boil, add more liquid but stay lower than 60% then solder to seal the port/plug. The actual vapor inside was flowing out so the chamber if correct soldered will have lower internal pressure when cold. Of course in the factory they use vacuum, so they do not need to heat it up each chamber for a long time and so on...
just a pro tip. to fix the hotspot on this card you have to and is literally the only way to fix this hotspot on this card is using a kryosheet and thermal puddy for the pads. 10 delta after that always after long gaming 15 max. I have had these cards come in probably 50 times in the last 6 months. this is literally the only thing i have ever done to get this issue fixed on this card. this issue will still happen with the hotspot on this card even in a waterblock due to the die being not even whatsoever.
There is no way to repair the defect vapor chamber, it is under a certain atmospheric pressure and they injected too little water into them and drilling a hole into them will break the low pressure environment they operate at. they are built defect and have to be replaced and there is no other way to fix that issue
I dont quite have that high of temps on the hot spot but its pretty close and im running water cooling. card will max out at 60 degrees and hot spot will be 90 doesnt really go any hotter than that though. I figure just run it and if it dies then it dies.
@@sg1trogdor the pad Linus tech tips store sell the legit stuff there are other sources but you can't always trust that they are actually from Honeywell they don't sell to the public it's a pad that when it heats up changes phase to a paste Linus tech tips did a great video on it and you will never have to repaste for the life of the card it doesn't dry out or degrade
@@worldwar208 Your awesome got some and put it on yesterday and my hotspot temps dropped about 15C. now im hovering around 55 core and 70-75C on the hotspot
The hotspot problem is only gonna get worse even if vapor chamber is fine, PTM7950 is the perfect solution if you want repaste and never repaste again for entire card lifetime. for me i went from 92c hotspot on my liquid devil 7900 XTX to 60-70c hotspot min max at full load with PTM7950 MX4 which i attempted 2 times pumped out in 1-2 weeks went from 75c hotspot back to 92c hotspot, these chips output to much power for MX4 to be effective long term.
@@Justgreen89925 Yep it was in my case 1-2 weeks from 75c hotspot to 92c hotspot 2 times on MX4 then i went PTM7950 which is now on day 248 or so no change.
I've notice that my RX7900XTX from Asus with original cooler was reaching up to 88-90 degrees. After moving to EKWB Vector 2 WB I'm not seeing more than 75-78 degrees.
Most 7900XTX have super high hotspot, mine is often at more than +20C delta (Asus model), even with a working cooler the contact with the big chip and its chiplets is bad I think +50C delta is really crazy tho, and indeed the cooling/vapor chamber is failing
Its either the vapour chamber issue, which is the most egregious of the issues AMD has been plagued by in this new gen of cards, or reference design coolers in general being poorly done for the calibre of card, such a shame because the 6000 series was good in this regard and consumed not so much more power. Besides, some of the lower end cards with reference model coolers get so fucking toasty that they greatly benefit from extra thermal mass and pads on the back of the GPU/backplate should be advised, and that doesn't seem to be done by AMD, causes way more headaches/worries just to save a pitance on final costs that board partners would gladly soak just to deliver an actual good product.
Funny thing if the card is outside warranty, AMD should least replace the broken vapor chamber, i am curious if AMD even gonna do this, it would be great injustice if they did not replace cooler, i personally only watercool tho but thats not something avg user wants to do.
I have a laptop with a 6800m in it, hoping we can figure out how to do a memtest on this GPU. It randomly won't initialize drivers and lock up on windows start and give an adrenaline error. tried all new drivers and firmware updates, not going down the hardware diagnosis. When it doesn't give the error it works and benchmarks fine for indefinte use. 6/10 times it does freeze up and sometimes black screens on windows bootup. It will give an adrenaline error and let me submit a crash report. Only fails sometimes on initializing drivers. Benchmarks/load test seems fine afterward if it loads.
Doesn't look good, Hot spot is too high temp. These thermal paste is really bad but I got one good about thermal pad lowest is 50-60C. That's why we recommend to modify my graphic card and their graphic card to change thermal paste to one good pad. I have a RX 7900 XT I learned how to modify my graphic card but bit damaged however it is fine for me.
the vapor chamber on the cooler has failed. just had the exact same problem on a 2080 mine was a blower style card so i have them laying all around on donors
It's way better to tell the client his/her card can't be fixed and a few days later you call again and say it is fixed than many of those repair shops that damage the card and after they try to "repair" it, it's definitely dead. It's way better.
Is there a reason you don't use an ESD wristband or gloves when handling GPU's? The reason I ask is because I work with circuit board production, and where I work, anyone who touches PCB, especially those who repair them, must wear both. I don't personally repair nor regularly touch any PCB, so I'm just asking out of curiosity
Might be a stupid question but couldn't you drill 2 tiny holes either end of the vapour chamber and fill it with liquid metal then solder up the holes?
usually a working vaporchamber is equivalent to +1000W/m.k, LMs do not reach 150W/m.k; So its probably better to fill it with some ml deionized water (it needs volume to go vapor flow, so too much water is also bad)
hello tony , nice video, i have msi b450m mortar max , when cpu in socket CPU VDDP not turning on , i testing with R5 1400 , can you help me please or at last , give me link to someone great on repairing motherboards , thank you
meh, Vapor Chambers work fine on the other like 95% of GPU's ever made with them - just a particular design issue. not a reason to make Coolers worse and remove Vapor Chambers, we just gotta not make mistakes when making things. Heatpipes and Pumps and whatever else have had mistakes with them more than never in the history of Electronics too, shrug.
I dont understand why people are buying these ref models bcs they cost the same like much betters cooled aib models like Power Color Hellhound, that has like 20c less on hotspot and its silent..
After a whole day of research, adding fluid to vapor chamber without specialized equipment is impossible since it needs a negative pressure as well as liquid, AmD has admitted the failure and was replacing cards with new ones. Other option is different cooler only reference models have this problem. Some xtx cards seem like they dont throttle and overheat and crash, maybe these overheat seals and allow pressure to escape. So 1000$ new gpu had non fixable error which took multiple people to point out for them to fix the issue, before amd said pound sand its your fault
