Other things I tried after the "repair": - flash the stock BIOS again - Update the Nvidia Firmware Any other ideas? :D
@oOZellzimaOo
Жыл бұрын
Have you tried resetting it with a hammer?
@iLLusiveMan82
Жыл бұрын
Send it to krisFix...he's an expert at repairing GPUs and he's also located in Germany
@iliqiliev
Жыл бұрын
Cook it again in a real oven? 🤷♂
@HiltonHeslop
Жыл бұрын
Reflow the vram :) voltage is there perhaps a module shifted
@MrDeadrage
Жыл бұрын
Northridgefix youtube take a look
@MillaGamer
Жыл бұрын
I'm a board repairman here in Brazil, this defect is related to memories, communication failure, you can use the mats/mods software to discover the faulty memory channel and change the module, hugs from Brazil.
@der8auer-en
Жыл бұрын
Very interesting. Will look into that :) thanks for the comment!
@TheLikesofMee
Жыл бұрын
This gentleman from Brazil is correct.
@AlphaMachina
Жыл бұрын
Seems more likely that the BIOS is shot from the heat. I remember older GPU's not booting on their own, but showing up in the system on booting with onboard video when the BIOS memory had gone bad. This looks like a similar issue. Sounds like a vBIOS/eeprom failure. You could try reballing the eeprom chip, or replacing it all together.
@royboysoyboy
Жыл бұрын
This might help. Definitely try mats and mods on usb stick
@dreamcat4
Жыл бұрын
@@AlphaMachina vbios chip isnt near the location where heat was applied plus when its reflashed then what has been written to the bios memory can be checked / verified. wheras vram memory chip were very close by if not actually directly in line of fire. so the mats/mods suggestion makes more sense here.... however assuming this then you would still need to reball or reflow a gddr6x right next to the gpu core. and that task is pretty challenging. it isn't at all easy and should be handed to somebody else with better equipment and tools to stand any chance of a real success. of course relatively changing the spi chip is easier to eliminate as a potential cause. or it could be both. just not as much reason to suspect over an actual bad vram. for example when the card boots it probably reads the contents from the vbios and copies it over on to the vram. before going further... if 1 of the 12 vram chip is bad then that will consistently corrupt the contents of that space in the memory every 1/nth bit or byte could be a bad data. (assuming there is no ecc error correction)
@brothatwasepic
Жыл бұрын
Roman to the RMA person: I swear I only heated it up to 220 degrees Celsius Technician: WTF
@der8auer-en
Жыл бұрын
oopsie :D
@davidmoffitt981
Жыл бұрын
@@der8auer-en I wish YT allowed more than one like for comments such as this :D
@FacialVomitTurtleFights
Жыл бұрын
Well what did you think was going to happen lol geeezz
@brothatwasepic
Жыл бұрын
@@FacialVomitTurtleFights it's a common solder correcting technique it just sounds funny hehe
@ricdintino9502
Жыл бұрын
Thank you for sharing your failures as well as your successes. We learn from both.
@der8auer-en
Жыл бұрын
thanks for not killing me for it xD
@FoxvoxDK
Жыл бұрын
Let KrisFix-Germany have a look at it, he can re-solder and re-ball it for you, maybe for a collaboration or as a paid service, but he knows his stuff and I watch him regularly.
@owensatom
Жыл бұрын
A good collaboration video that would be
@Totto70770
Жыл бұрын
This
@rkan2
Жыл бұрын
Yeah, absolutely, collab with all the reballers.
@bubba966
Жыл бұрын
Yeah I'd send it his way if it was me. His work is very impressive on what he can fix.
@valshin
Жыл бұрын
Yeah, good idea
@Levent_Ergun
Жыл бұрын
If you cant read the cards details in GPUZ, that is usually a sign of EEPROM/vBIOS failure. I have had that happen to me and its usually either fixed flashing the bios few more times and or by replacing it the chip. All hope is not lost !
@adcommlaw
Жыл бұрын
is 50/50 with memory
@BrunoVilela1
Жыл бұрын
@Levent_Hates_This wiriting a new bios using a chip recorder tool instead of bios update software resurected a mobo in my case. definitely he should try replacing the chip or rewriting the bios again.
@mjdevlog
Жыл бұрын
"Does it run on x16 now?" "Yes" "What did it cost?" "Everything else"
@der8auer-en
Жыл бұрын
🤣🤣🤣🤣
@MikeZak101
Жыл бұрын
Does
@mronion6417
Жыл бұрын
It’s really important to consider the history of this card. It’s very new (obviously), it has a PCIE problem, and it has been under LN2. The PCIE slot is essentially directly connected to the GPU chip, and since the capacitors tested fine, the original problem was almost certainly with the GPU core itself. Since the heat was applied to the GPU core, and the board was covered with what looks like thermal insulation tape, it’s unlikely that another component received too much heat. My guess would be that the bumps connecting the chip and the substrate have failed/become misaligned. Buildzoid explains this really well in his video about chip degradation. It’s important to note that the PCIE controller and the memory controller (which showed up as not detected in the GPU-Z report) are both on the edge of the GPU die, and therefore are most likely to be affected by thermal expansion. Furthermore, LN2 means that the card has been under a much larger change in temperature than normal. The GPU-Z report of x16 functionality being restored is further evidence for this theory (assuming it is accurate). As for a solution, fixing this issue is difficult and not always permanent. The key idea is to get the die and the substrate to a similar temperature, so that the bumps connecting the die and the substrate realign. Most fixes of this type put the GPU in an oven for about 10 minutes, which is plenty of time for the die and substrate to reach equilibrium. I suspect the heat gun heated the die to a much higher temperature than the substrate, exacerbating the issue. You could try the oven method or turning up the temperature of the preheater to a similar level as the heat gun. Alternatively, since you have liquid nitrogen, you could try to cool down both the die and substrate. It is possible that these strategies won’t fix the issue. However, they shouldn’t cause any additional damage if another component turns out to be faulty. I wish you the best of luck with this repair, and I hope to see you break some world records with this GPU!
@der8auer-en
Жыл бұрын
I thought the GPU would not move as there is some glue on the edges. Possible that the GPU got too hot though. The heatgun uses 350°C temperature. My current plan is to send it to KrisFix :) Already talked to him
@samuellourenco1050
Жыл бұрын
Probably, the heat gun finished it of by the means of too much heat, thermal expansion and thermal contraction. And something tells me that the BGA balls got damaged too by this process.
@dungufed349
Жыл бұрын
@@der8auer-en Best decision you could've made. I hope he does a video about this aswell!
@badvoiceallnoise
Жыл бұрын
@@der8auer-en solid plan
@northwestrepair
Жыл бұрын
When i saw that blue heatgun, i knew exactly what kind of repair this is going to be. Should have sent it to me instead.
@der8auer-en
Жыл бұрын
sorry that you had to suffer watching this video xD What would your call be? GPU dead or chances for survival?
@northwestrepair
Жыл бұрын
Power monitor chip. Memory test. Signals check.
@JulianUccetta
Жыл бұрын
In my very limited experience fixing GPUs, I've only run into something like this once, and it turned out to be a bad EEPROM/BIOS chip, though this could also indicate faulty memory *or* a faulty memory controller. There is a possibility the GPU heated up to a point where something in the memory controller just went awry. Best of luck!
@FR4M3Sharma
Жыл бұрын
Should've asked your neighbor KrisFix Germany or Tony at NorthWest Repair. Those guys are masters of this craft.
@balika011
Жыл бұрын
You forgot a lot of steps. First you have to remove the edge bonding. Second you need to heat it up a lot slower. Most likely you ripped traces by heating up too fast with the gun. There is a good reason why rework machines and thermal profiles exists. After you are done you need to remove the flux. Flux is corrosive and conductive. (There are "no clean" fluxes. At such a low voltage that a gpu runs at they are also must be cleaned.)
@arch1107
Жыл бұрын
i don't think that he ripped any trace, but being a so big bga chip, it is always a possibility, he did used a very powerful heat source but he did it at like half the heat to melt the solder, the only way to confirm this is remove the bga to reball
@glenndoiron9317
Жыл бұрын
No-clean fluxes are not corrosive or conductive if properly used. You absolutely don't need to remove them, unless (a) there were contaminants on the board to begin with, or (b) they were not heated up to activation temperature.
@ThePaFcio
Жыл бұрын
If you'd like to go with reballing, you can try pre-formed solder balls that require stencil to be put in place. Stencil probably could be ordered from pcb manufacturer without any issues if you'll have the layout file.
@SoranoGuardias
Жыл бұрын
I know professionals that HATE that method. It's easier to tape the BGA to the stencil and smear solder paste on and then hit the BGA with a heatgun to form the balls.
@therealnoodledog6660
Жыл бұрын
I've actually fixed an AMD card with that method. I used the oven method. You did a couple of things wrong according to my experience: 1) you didn't use an oven, you should have used an electric oven and not an hot air gun with hot pad 2) you allowed direct contact of the card on a heated surface, when in the oven i used 4 aluminum papper balls to rise the card out of the floor of the oven 3) you should have distributed the heat evenly, i think the yellow paper cover to protect the GPU surrounds you used is actually what killed it, because it didn't allowed the components to expand and contract in sync and that caused tears on components
@bojanmilovanovic747
Жыл бұрын
This... Never use kapton tape ... its streching under heat and it can displace some of the components with it...
@gsuberland
Жыл бұрын
Those caps at the back aren't for decoupling, they're AC termination capacitors. With decoupling you're typically putting the capacitor between a voltage rail and 0V to provide a low impedance path for high frequency currents. PEX uses AC coupled differential pairs. The capacitors are in series with the signals, and are used for DC blocking.
@downwiththesneaker9933
Жыл бұрын
So something similar happened to me on a 3070. GPUZ showed 0MB of VRAM so I flashed the vbios with nv flash and that did nothing. So I unsoldered the vbios chip and put it in my usb EEPROM Flash and compared the vbios I flashed to it and they were wrong. So nv flash did not work or did the flash wrong. So I used the program for the usb flasher and flashed the vbios on the card soldered it to the 3070 and BOOM it worked. Maybe try doing the same thing?
@Cristi-co8vr
Жыл бұрын
The heatgun you used is not for electronics, is for soldering plumming copper pipes :D The air power of that hotair gun can displace some components and even solder. The electronics hotair gun has temperature, air control and different nozle sizes for each IC you use it on. You can draw conclusions now.
@jonahjameson272
Жыл бұрын
He had no idea what he was doing and brokeeee it
@valshin
Жыл бұрын
You better need infrared repair station for such level of repair. It can follow special thermal profile with phases like flux activation, solder melting, etc. Of course, it's possible with the heat gun, but the skill level required in this case is very high + the heat gun itself should be dedicated for electronics repair with temperature control.
@TerraWare
Жыл бұрын
Send it to KrisFix-Germany, he's a master at fixing gpu's and is in Germany. Perhaps it may have been a better idea to have sent it to him before you cooked it but if anyone can fix it it's him. Would make for an awesome video too imo.
@TheBrambolinie
Жыл бұрын
Electrical Engineering student and former System Analysis engineer here: leadfree solder usually melts around 213 degrees celsius IIRC. It might take a while to reach that temperature under the BGA as there might be a solid copper plane below it (multiple internal PCB layers). This means longer baking time is needed. In our factory, this takes multiple minutes (5 to 10) for a single PCB. With the kapton tape you might have displaced a smaller component due to forces of the kapton tape displacing the component while the solder was fluid (smaller thermal mass of other components). Baking definetely can work, but i suggest you do it in an oven where it is fully enclosed without any foreign objects on it such as tape. You can retry it this way I would say as it is dead anyways. Thermal stress is not good for PCB's. Im not sure if the initial error was related to the die however...
@ZylonFPV
Жыл бұрын
I like how you include terrible failures on this channel. Far too much of KZitem/social media is things being or at least appearing perfect. Keep failing hard! And winning too 🎉
@wazaagbreak-head6039
Жыл бұрын
Content is content, why waste the money
@killer01ws6
Жыл бұрын
I always enjoy your experiments and ideas you come up with to make our PCs stronger, more reliable.. As with many scientific endeavors we do not always get positive results, but thank you for sharing the bad with the good.
@ThePopolou
Жыл бұрын
Take a look at @northwestrepair And a few of his videos. Very competent repairer, perhaps the best amongst a few on YT. Uses Mats diagnostic on boot up that should identify the fault area (since you appear to have a partial start of the card) and has a link to all his software and guides - he's a legend. Seems like the GPU started but the memory didn't.
@nexxusty
Жыл бұрын
Best on YT actually. By far.
@mbwatson1000
Жыл бұрын
Another vote for northwest repair. His work is exceptional. You are highly competent in your field, so I'm sure you'd respect another specialist who is just as competent in his.
@pctotil
Жыл бұрын
I would sugest also the guy at Northridgefix, his videos are great , take a look
@montywhite3646
Жыл бұрын
Capacitors are like little batteries that store a static charge. Many of the components on a graphics card are static sensitive. Putting a previously active card onto a conductive metal surface is like randomly shuffling your feet on a carpet, building up a static charge in your body and randomly discharging it on your PCB for fun. They come in an anti-static bag for a reason.
@Alexandres668
Жыл бұрын
Have you tried to read the bios chip ? It could be that you killed the actual eprom with the heat
@jrok96
Жыл бұрын
your response reminds me of a cheech and chong record i once had . when cheechs car wouldnt start and chong says "have you checked the air in the tires lately".
@BobHannent
Жыл бұрын
A GPU will appear in Device Manager but not boot if the BIOS is dead, this is a classic problem with vintage GPUs where the BIOS flash memory has gone bad over time.
@ruxandy
Жыл бұрын
Roman pinned a comment after uploading the video, but it seems that nobody bothers with reading those anymore.
@BobHannent
Жыл бұрын
@@ruxandy except it wasn't there when we posted, it must have been pinned just after
@cldpt
Жыл бұрын
@@ruxandy ackchyually... even after a "successful BIOS" under Windows, the BIOS could very well still be corrupt. But I myself am more inclined for a dead GPU core.
@loonie5468
Жыл бұрын
"I hope we just didn't kill it." I remember thinking a similar thought after taking my faulty RX 580 8GB out of the 220 degrees heated up oven. The result was the same.
@diego1694
Жыл бұрын
Seems like the obvious options are that you either fried part of the chip or you broke the solder. In that case, the only hope would be to have it professionally reballed.
@arch1107
Жыл бұрын
well, i agree with you on the solder part, but i think the problem was there before the attempt, in my opinion this could be just a bad gpu and asus should replace it now, it might be other components, we need someone with more equipment and more knowledge to confirm the problem
@victorx4648
Жыл бұрын
You need to remove the GPU chip from the card. Clean up all the soldering pads. Inspect the pads and traces. Re-ball the GPU chip. Solder the GPU back again, but not with a heat gun, but with the proper soldering station with the upper heat generator, bottom heating plate, the stands that let you keep the card even (so it won't warp), and temperature monitoring. Also, you need a microscope or a magnification glass to see when the soldering balls are melted and if the GPU chip sat down evenly.
@vitor900000
Жыл бұрын
You most likely caused another cold solder that wasn't there before. The card can probably be repaired with a proper reballing.
@markissboi3583
Жыл бұрын
Been waiting for a weekly post thought you mitve sailed away for xmas island holidays dont matter roman good or bad always informative .
@w04h
Жыл бұрын
227 degrees is a typical temperature for lead free solder but I don't think these kinds of chips use that. I think they use indium (or bismuth?) pastes with about 180 degrees if not lower melting point so for that 220 would really be high. I found an article of someone using 200 successfully with an oven. Ill post as a reply because of youtube.
@zmuk3727
Жыл бұрын
it's possible, bismuth based lead free alloys range anywhere from 138c to 183c
@primus711
Жыл бұрын
I seriously doubt they use that as its more expensive consumer grade stuff is always whats cheaper
@ksks6802
Жыл бұрын
DO NOT use an oven you are currently using for food prep. Alloys and pcb components atomize and float around in your oven forever.
@Th3Fly1ngCow
Жыл бұрын
I love this type of content Always so informational
@panscrank
Жыл бұрын
And hilarious
@morpheus9137
Жыл бұрын
You learnt how to break a GPU. Next week, the hammer...
@urikaufman4258
Жыл бұрын
It could be a millon things. For starters, you assumed the issue was related to the GPU soldering (which is the most difficult problem to solve even with proper reballing tools) when the PCI-E x8/x4 problem could be caused by other faulty components. The temperature applied would be enough to liquify unleaded solder with flux so probably the new problem is actually related to BGA balls not making full contact with the PCB or GPU, but bear in mind that even with that massive amount of Kapton tape (which can actually lead to other problems) every single component of the board got heated (from the PCB heater and also through thermal transfer via ground plane for example) so pretty much there could be misplaced components, disconnected pins or shortcircuits in any component of the board, especially those closer to the GPU like RAM. Probably check BGA balls on the memory chips arround the GPU and next time attempt less invasive means to solve the issue before resulting to heating a GPU with a heat gun (actually, never use a heat gun with BGA, particularly with bigger chips like a GPU). In any case you should use a rework station with temperature control but even then, a rework station is not meant to be used with big BGA chips.
Жыл бұрын
Well trying to actually melt the solder and not making it worse on a big BGA chip like that will require proper equipment and those are not cheap or small devices. They will use infrared heating, both sides to get the chip to around 220C for lead free from what I have seen if you want to remove it, takes a while, then reball it with leaded solder and re-solder it at about 200C using a profile that will heat up the card over say 7 minutes or so (guesstimating). Big chips like that will most likely require fine calibration of heating zones to make sure the PCB dont warp, warping of the PCB or chip is an issue, I know I done that XD So you would need donor cards to develop a profile for your rework machine to even start doing the actual cards you want to fix. I knew what you where trying and what would happen. It requires expertis and good equipment to do this, there are plenty of videos on YT showing pros doing it. I can only do simple chips on older cards due to my low skill level and less then ideal equipment, tough mine is better then what you used and still the best I can do is 20 year old GPU's with few big balls. Ram chips you can do by hand. I probably cant link videos here so wont but if its possible I can dig up some that shows how its done. But you seem to have the resources to invest in the stuff needed, after that its time and practice so I dont see why you could not learn to "reball" and rework cards if you really wanted too.
@dreamcat4
Жыл бұрын
so much agree with what you said here. it seems in current situation (for this one) the better shot would be to leave alone the gpu core and just try to reball only nearby affected gddr6 chip. is more realistic goal. assuming that is what he messed up given those specific indicators
@DjPaulWoods
Жыл бұрын
Right tool for the job would help. Use that heat gun for paint stripping and get an AGP data line tester for diagnosis
@Nevakonaza.
Жыл бұрын
Love these kind of vids, Would be amazing to see this beauty come back to life even if its in collaboration with someone else who has a bit more knowledge on memory issues,GPU reballing etc.. Its just a shame that those kind of services are so expensive now :( i have a faulty Vega 64 but the repair cost would be more than the card is worth, Yet its still a perfectly powerful card for even todays games. :(
@BillyC500
Жыл бұрын
I am only just now finding the EN channel. I had thought the English videos just stopped (just imagining the extra work to publish content in multiple languages) . Happy to stumble on the channel looking for direct die content!
@der8auer-en
Жыл бұрын
welcome to the other side :D
@MoDzAudio
Жыл бұрын
I'm so curious as to why you went right into trying to reflow the die and not checking other points of interest first. To me it was obvious that the die itself wouldn't be the issue but memory or a cap somewhere around the die would be to blame. I'm just curious why you decided on that first.
@morpheus9137
Жыл бұрын
Ovens and paint stripping guns is the non tech youtube 'fix' for broken gpus, and yes it never works.
@AI_MicroSolder
Жыл бұрын
I had the same problem. What fixed it was taking the m.2 990 pro 1 tb ssd out of the number one slot. It will share the pcie x16 with the graphics card and number one m.2 slot. Once I moved the ssd to the next slot down. My strix oc 4090 will run in x16 pcie
@st3althyone
Жыл бұрын
Goddamn, it’s a shame this happened, Roman. Yet those are the risks of experimenting with all these high-end GPUs. Always send it to a repair shop when in doubt. Especially with these $1K+ GPUs, it’s better to spend a little bit of money than to end up with a paperweight. I do component-level repair on motherboards, and it’s incredible the shit I see when people send in their stuff for repair after they’ve mangled the shit out of it. It seems you had some thermal shock, and the GPU needs to be reballed. It’s a simple enough process, but only when you have the right equipment, which isn’t cheap. So don’t despair, my friend; I’m sure there are some competent shops in Deutschland that will be able to and willing to set you right. Good luck, and stay safe.
@ChaosHusky
Жыл бұрын
Re-balling isn't always the magic fix though if the failure is underneath the die and on the carrier substrate or other failure. Certainly doesn't help PS3s and Xbox 360s like people thought, nor the GPUs and other boards from back then. For it to die outright is surprising though..could be a shitty PCB!
@st3althyone
Жыл бұрын
@@ChaosHusky Considering he was missing half the PCI lanes, a re-ball was the right thing to do. Unfortunately, whatever he tried seems to have fried some of the memory modules or even the controller and maybe even the vBIOS. His mistake was putting the card directly on the heat plate. He should've used a silicone pad or put it in an oven to heat it uniformly before using the heat gun to attempt the re-ball method he used. Also, if you notice at the end of the video, when he shows GPU-Z, it shows that the bus interface is showing PCIe X16 4.0. Unfortunately, his memory information is missing, suggesting the memory controller/module issue. So what he did worked, but he now has other problems. Either way, my suggestion stands if he can’t find a fix after all our suggestions. Edit: clarity and spelling.
@JMUDoc
Жыл бұрын
10:48 To be fair, Roman, you _did_ fix the x16 issue🤣
@bencesas3482
Жыл бұрын
Lead free solder's melting point is between 220C-230C, depending on what alloy they use, but the temperature has to be even and the heat up and cooling down has a trapeze-like temp profile and I think u need 220-230C for maybe 10 minutes I'm not sure.
@benjaminchung991
Жыл бұрын
Peak reflow temperatures are usually only held briefly; most datasheets call for about 10-30 seconds depending on the part. The rampup/rampdown that you describe takes most of the time.
@robertmarsh1338
Жыл бұрын
@@benjaminchung991 I've fixed GPUs multiple times with the oven method.10 minutes at 200 degrees C.
@benjaminchung991
Жыл бұрын
@@robertmarsh1338 Lead-free solder won't have reached its liquidius point at 200C; as an example, Chip-Quik SMD4300SNL10's liquidius point is 217C with a 210 second ramp followed by a 30 second ramp to 249C sustained for 10 seconds then cooled back to 217C over another 30 seconds. An oxide layer as exists in completed solder joints would further inhibit good melting performance. I suspect that the successes that you've had are due to the differential thermal expansion of the joints closing a gap rather than remelting of the solder. 200C just isn't hot enough for lead-free solder, as is used in essentially all GPUs these days, to melt.
@JJFX-
Жыл бұрын
@@benjaminchung991 Not my area but I would assume the oven also causes all of the BGA underfill to hit well above its glass transition temp. This combined with the thermal expansion may allow poor contact areas to become 'locked' in placed once everything cools and constant pressure is applied. Seems like the original issue would eventually undo all this but if you're really lucky and keep the GPU cooler going forward maybe it could keep it going longer than you'd expect?
@imangreen9043
Жыл бұрын
When you won the battle but lost the war. Sorry to hear that Roman.
@Moon___man
Жыл бұрын
send it to northwestrepair, he seems to know a thing or two about bringing cards back to life
@MatthewLenz
Жыл бұрын
100% that dude can fix anything.
@arch1107
Жыл бұрын
iirc he revivied gpus by reballing ram and other simlar fixes, sounds like a great idea, if he doesn't contact asus
@primus711
Жыл бұрын
He doesn't have the skills to reball a gpu
@Moon___man
Жыл бұрын
@@primus711 watch for yourself hes definitely done it. one time unsuccessfully that ive seen but ive also seem him succeed
@primus711
Жыл бұрын
@@Moon___man he doesn't have the skills nor know what he is doing this video plus what he said himself is the proof
@Betaobonzao
Жыл бұрын
I just received the Asus Z790 Extreme (11/28/22) with a NEW MANUAL that clearly shows that Video Card on PCIEX16(G5)_1 connected together with M.2_1(CPU) makes the video card work at 8x
@hakylymypa
Жыл бұрын
did you tried to put it in an x8 slot? maybe the last x8 lanes prevent booting now
@primus711
Жыл бұрын
Doesn't matter if its 1x or 16x its all the same
@johnk5148
Жыл бұрын
Try moving you’re m.2 drive one slot down, if you have it plugged into the top slot will force you’re video card in 8x mode.
@q1337
Жыл бұрын
for me and my 1060 3gb , it was a single strand of hair that made it run at 8x , it's funny but i spent so much time trying to figure it out.
@speederooa3846
Жыл бұрын
sad to see this new "mini-supercomputer" broken, hope you can fix it, good luck!
@SimonZerafa
Жыл бұрын
Hi Roman, I would recommend sending the card back to the vendor and asking for a post mortem. They may well find the initial PCI lanes disappearing symptom interesting enough to investigate as it might also happen to someone else . Clearly the various environments it's experienced may not be within the warranty but if you smile sweetly and say please they might send you another one? 😉
@tazogochitashvili6514
Жыл бұрын
He said 3.0x16 I think? So lanes didn't disappear, at least not between the card and the mobo. Still you're right, could be useful info on the card.
@dainluke
Жыл бұрын
KrisFix is your man. Dude is a magician.
@zMeul
Жыл бұрын
the use of kapton tape was a horrible idea as the temperature rises the tape shift and it will pull components with it when the solder reaches the melting point
@EvL--_--
Жыл бұрын
Whelp, I just hope these videos get more than 2k in revenue so at least you have a free paperweight. Thanks for the videos Roman! Someone has to do these experiments :D
@Biker_Gremling
Жыл бұрын
Roman: "I don't know why it no longer works" *Me thinking after all those "I think", "this should be fine" and "I'm not an expert at this": WOW! The RGB somehow still works!
@infango
Жыл бұрын
maybe ask buildzoid ? he is good in GPU necromancy ..
@w04h
Жыл бұрын
No offense, that guy maybe knows a thing or two about power stages but he is totally clueless about anything else gpu or electronics related. 99% of his viewerbase don't know what he is talking about.
@AI_MicroSolder
Жыл бұрын
Your next step is to have the cpu on the gpu taken off and reballed being you put solder in the cpu sides and tried to reflow it. That for sure linked some of the traces. Talk to Alex at NRidge Fix and he will h3lp you get it fixed. He eoesnt do the reballing of the cpu but knows people to trust to do the same professional work. By the 2a6. When 8 removed the m.2 out of lane 1. I installed 8t into lane2 for the second pcie x16 lanes. Thats where my windows os system lives and everything works perfect now. I hope you good luck and love your show and products. I try to buy most of my componets from you.
@PolskiJaszczomb
Жыл бұрын
I kinda wonder why didn't you RMA it.
@der8auer-en
Жыл бұрын
It's a sample. I did not buy it
@LolJolk
Жыл бұрын
I wonder why... "Mod-BIOS allows 1000W Power Target - RTX 4090 Strix with Water Cooling"
@PolskiJaszczomb
Жыл бұрын
@@der8auer-en Of course, but why not ask Asus to replace it or at least check? I've smashed MSI monitor sample once (long live the absolute garbage stand they provide) and they agreed to replace it.
@PolskiJaszczomb
Жыл бұрын
@@LolJolk iirc PCIE lanes stopped working even before E-bios mod.
@LolJolk
Жыл бұрын
@@PolskiJaszczomb custom bios voids the warranty regardless of how the pcie lanes broke. especially the extreme overclocking and putting the 1000w limit come on thats just abusing the card
@v3rnus733
Жыл бұрын
I did put my gtx760 into the oven at 130C for like 20min when it started artefacting, then i left it in the oven to cool down and tested it in PC like 3hours after baking it and it worked for another 3 years :). I think u might kill it by moving the card when it was still not cooled down enough and some parts could move or de-solder. Try to put it to oven again on since u have nothing to lose at this point :)
@meeponinthbit3466
Жыл бұрын
Can also confirm ovens can fix bad BGAs. No it won't fully reflow, but it may at least sinter the bad solder together and make it work again. I fixed three free TVs this way. But yeah thermal cycles are huge issue since it's so fragile, but it's better than nothing you can always rebake it.
@akumarus
Жыл бұрын
Love the conclusion @9:00 "seems like this method is not the correct one.." :)
@WouterVerbruggen
Жыл бұрын
Wouldn't the flux make things way worse here? Flux reduces surface tension of the liquid solder, this might make the solder bumps to flow, merge and create shorts everywhere underneath the chip...
@big0bad0brad
Жыл бұрын
Flux decreases the surface tension next to metal by removing the oxides on the surfaces. It doesn't reduce the surface tension against everything else because there's no fresh metal being exposed there to stick to.
@kakkoishonen
Жыл бұрын
Why did you decide to do a reflow in first place and why did you do it this way? Doing a reflow process requires important tools like a decent preheater (a Puhui T8280 at least) and a real BGA rework station (a 850D heatgun at least). Those are basic tools but not the best if you really want to do a reflow (at least for a 2000 dollars GPU you should've used those). But the real question is what did cause the problem? For a PCX limited to 8x instead of 16x the problem is around the capacitors linking the lines to the GPU. Now what you had done with the heatgun is you destroyed the solder or even the silicon inside the GPU chip. What you should do now is to try to reball the GPU (of course using the right tools) or send it to Krisfix Germany, he will fix this GPU perfectly for you.
@earthtaurus5515
Жыл бұрын
Others have already posted thing's I'd have suggested so all I can see good luck! Hope you're able to revive the card somehow. Thank you for sharing your experiences, it provides a invaluable resource in what to do and especially what not do 😁. The latter is very important as at least others can avoid making the same mistakes lol.
@AgentHeX_0007
Жыл бұрын
My 3080Ti has been in my pci-e riser from the day I bought it, plugged in only once when new. Last week it started only running in x8 and the PC was crashing with artefacts only when running games. I removed the card cleaned the pci-e contacts, x8. Cleaned the riser contacts, x8. Plugged the card directly into the motherboard, x8. Tried it directly in different pci-e x16 slots, x8. I then gave it a few hard taps onto the cooler and plugged it back in to the riser, x16!!! Working fine since then. I don't know what got it working, the hard tap to the cooler or the inserting and removing of the card repeatedly, who knows.
@Fryk1990
Жыл бұрын
I remember baking old GPUs before 10x0 series :D
@davidsimpson2967
Жыл бұрын
Should add that whilst I repair GPUs myself, there are guys I can put you in touch with around the world, including Europe (I’m in the U.K. myself) who would be far better placed than I am at the moment to help diagnose and fix this and maybe do a vid collaboration.
@sspacegghost
Жыл бұрын
i feel your pain - at 2am last night my robo vac ate the usb cables on the floor pulling my backup TB hard drives off the table smashing them. hundreds of dollars...
@michaelthompson9798
Жыл бұрын
Sorry to hear about this 😢. As an Asus TUF RTX4090 owner, I’m now goin* to check to see if I have the PCI-e 4.0 x 8 issue too 😇🫡
@calop002
Жыл бұрын
How did it go? I have the TUF 4090 non-oc, and i have this x8 issue. The problem is that 1 out of 10 boots it manages to go to x16.
@dougler500
Жыл бұрын
Was waiting for this video! Sad it didn't make it.
@roboman2444
Жыл бұрын
Heatgun isn't precise enough. You need a good re flow oven (aka a toaster oven with some temp probes in it), and you need to generally heat and cool the card slowly.
@thomasdb.eu.
Жыл бұрын
i put a little lead solder wire on top of the die and then just heatgunned it , with a macbook once. the solder melted then i stopped adding heat. it can also help making a tinfoil square cup around the gpu. the repair was sucessful and nothing fell of the back of the board
@rossk7927
Жыл бұрын
if nothing else works, reflow it again at a higher temperature. Maybe you cold fractured the joints by getting hot enough to cause maximum thermal expansion, and greatly soften the solder, without reaching hot enough to cause uniform reflow. I've never worked with BGA parts, but I've done a fair bit of designing and hand assembling PCBs with small pitch no lead packages, so this is just a hunch.
@RATTU5_
Жыл бұрын
Northwest Repair is good and is even risky enough to bore out a section of boards just to get them working
@crf637
Жыл бұрын
"x8 to x0" Ouch!
@Gamingtechgg
Жыл бұрын
You can revive that card! I believe in you
@pow9606
Жыл бұрын
In my not very experienced opinion: - 1. A GPU reflow should be the last thing a person should do. 2. The hot plate is good for single sided PCB's. Dual sided PCB's when manufacturing the initial side. 3. The heat gun can reach temperatures of about 550c which would be too much heat too quickly. 4. Kapton tape can cause issues when components under it go in to a reflow state and surface tension can't do it's job due to the tape. I would inspect the PCB using a loupe or scope to make sure those tiny caps and resistors are all in places where they should be.
@Wimpzilla
Жыл бұрын
My first reaction was in line with what others said, why would de memory subsystem or the GPU memory initialization fail. It seems that the GPU can't initialise properly either because: -something related to the memory or memory controller, check mat/mods -the bios was/is corrupted, re-flash a fresh one. -i noticed that the GPU die sit on some kind of pads on its corners, reflowing may have lifted the die, check the balls using some light. -a component moved during the reflow process, check thoroughly the card.
@mibfox
Жыл бұрын
I baked once hd7970 Put it in the oven, 180C for 10 minutes. Card was not working at all. After baking, card was working without errors for another 9 months :)
@orated762
Жыл бұрын
"AGP PCI-E X16 -Purpose Display Video Card Checker Tester Card Diagnostic" will tell you if you have a short on your PCI lanes. Other than that, you need to test for shorts on the card, apply voltage to the short and monitor with a thermal camera to find the problem. The GPU is not dead, just has a short somewhere. This doesn't seem like a memory failure, look at the mosfets really good and see if there are any bulges cracks. Seems like a power problem. Just just be a surface mount fuse or could be a mosfet.
@danytoob
Жыл бұрын
Roman - you might try reaching out to Louis Rossmann (here on KZitem). He specializes (or used to) in component level repairs - primarily Mac Books - . While he is moving on with other things in the last year he truly is quite knowledgeable and capable.
@Gazereths1234
Жыл бұрын
Has a £200,000 CNC machine, doesn't have a £50 temp controlled Chinese heatgun/soldering station.
@arch1107
Жыл бұрын
well, to be fair, he doesn't repair gpus, so why have that?
@Gazereths1234
Жыл бұрын
@@arch1107 They are handy for loads of different things, from desoldering and replacing smd components to just removing stickers and shrinking heatshrink tubing. Most can go from 100c to 400c plus. Guy seems to spend a fortune on other tools, just thought he would have at least a cheap Chinese version. I mean jeez, he has a PCB heater plate which is designed specifically for surface mount/reflow work on PCB's.
@MrGeneralScar
Жыл бұрын
I once fixed one using the same method as you just did that "killed" the card, but with a difference. I didnt have the heater plate, and I left the cooler attached (just heatsink only) with stock thermal paste and pads as I didnt want the chips to move. I then basically cooked the top of the heatsink only with a heat gun and then left it to cool down to room temp by itself. I then removed the heatsink and changed the paste and pads for new stuff reassembled it and perfect, working and still working to this day. Granted it was a 1080Ti, so not exactly the same beast as a 4090. With that siad, I am also no where near as experienced as you at this kind of repair and the 1080Ti was completely dead before I started so at least I couldnt make it worse. Perhaps what I had was a fluke.
@DkGrRo
Жыл бұрын
Thanks God that KrisFix knows his job :))
@UnexplainedHoveringSpheroid
Жыл бұрын
I would think that either the temp probe was reading too low because of placement / mounting, or the tape had a gap / insufficient coverage to protect components elsewhere on the PCB. Not sure what you could do, short of getting a schematic for the strix card and providing it along with the card to an experienced microsolderer / board repair specialist to fault find for you.
@yatsukiko
Жыл бұрын
with that heat, not much should change, the only thing that changed is flux under BGA, from my experience isopropyl alcohol under chips can get them "shorted" or at least not working properly i can image same goes with flux to some extend, or the specific flux you used. I would try to remove that flux under chip with something like iso or my fav pcc-15 then as stupid as it sounds, blow it out with air compressor.
@dreamcat4
Жыл бұрын
yes makes sense to try. but maybe roman didnt saw such comment(s) after reading about some other type of things to do. this is a good idea however too
@SquintyGears
Жыл бұрын
we don't know if the solder did flow and then some of the contacts shorted. I'm pretty sure people who actually do board repairs put loads more flux , apart from that who knows.
@spinkey4842
Жыл бұрын
it's all good my dude, you got money unlike most of us
@farbrorfjellstrom
Жыл бұрын
I baked my old Radeon 9800 XT back to life back in the days. I was very careful to wrap the underside of the card in aluminium foil and use a toothpick to get it as snug around all components of the card to make sure the foil would hold them in place in case any solder would dislodge. I just put it in the oven and let it run for 200°C for an hour.
@AdamShumpis
Жыл бұрын
Should have probed the PCI pins first back to source to see if any were shorted or disconnected. Could have done that with a cheap multimeter. Call Rossman or NorthridgeFix.
@MeitsMSX
Жыл бұрын
I wonder where the myth of reflowing a flip chip on a substrait comes from.
@glenharland5287
Жыл бұрын
You should have used at minimum a rework station. That thing is not for soldering. More flux, if it dries up add more and get it up to 220c steady then move in closer to 260c for around 10 seconds. Use a swirling motion to evenly heat. Did you check the pcie pins in diode mode first?
@michaelklit7799
Жыл бұрын
Doing the bake more than once with iMAC cards, and it was 230 celsius for 8 minutes, then let it cool down in the oven slowly. And those card worked just fine for a few more years until the iMACS went on pension
@robertmarsh1338
Жыл бұрын
I've fixed GPUs multiple times with the oven method.10 minutes at 200 degrees C. And turning the oven off and opening the door before taking it out. Different problem though. Heat gun doesn't work I tried that. You need all of it hot for an extended period and heat gun is not uniform.
@McTroyd
Жыл бұрын
I'm fairly certain you would have checked this, but in the interest of being thorough: when you did the hot air reflow, did any of the passive components (resistors, capacitors, etc.) move out of place? Were there any solder bridges created? It could be the GPU itself is fine, but we've lost some signaling lines, or they're being pulled to the high/low rails when they shouldn't be.
@mikmoody3907
Жыл бұрын
At least you fixed the problem :)
@Marc_Wolfe
Жыл бұрын
Flash it again, what if the heat corruption vBIOS. I'd also go for a hell of a long alcohol bath to make sure that flux is out; like days in your alcohol filled cleaner.
@Morpheus-pt3wq
Жыл бұрын
IIRC Northridge Fix mentioned something about kapton tape not being foolproof - while it can deflect heat, it can also shrink and cause your heated components to move, once the solder weakens enough. He often uses metallic shielding, when using heat gun near heat sensitive components like capacitors and low melt solder, when working near connectors.
@Morpheus-pt3wq
Жыл бұрын
@Michaels Carport thank you for clarification 👍 edited my top comment to avoid any confusion.
@MeguRED
Жыл бұрын
Maybe BGA reball the thing? You could have made some kinda minor short between em...
@reed-young
Жыл бұрын
I am not at all surprised that heat more than 2x the rated operating temp did more harm than good. It's unfortunate. A really careful examination of the components that control how many PCIe lanes are used, could have been interesting.
@MJohonez299
Жыл бұрын
Good way to go from x8 to x0. Nice 👍 I enjoy modding and messing with my cards, but I do understand why my friends think I'm insane for shunt modding a 3070ti.. expensive paperweight is definitely a possibility
@greenbow7888
Жыл бұрын
If all else fails try sending it to one of the KZitem repair channels. Wish I could remember which channel, but have never seen soldering skills like it.
Пікірлер: 1 М.