Well this sucks, having a 980ti. if you have an gpu affected from this, is there anything that can prevent the memory vrm from blowing up? Does a bios flash for gpu core power affect this?
How do you know if your card is of reference design ? I have a Evga 780 6gb Super clocked ACX cooler card that I have been hold onto hopefully until the end of the mining crazy. Is this card affected ?
This happens a lot, I've been fixing GPUs for years now, and this is a typical problem with memory on reference boards, especially on this GK110 and GM200 reference boards like EVGA, or reference design based boards, like some models from ZOTAC, MSI, etc.. The problem happens when typically a high side MOSFET from memory VRM goes short. The lack of a good fuse (10A white fuses) instead a inductor on the 12V line that feeds the m-VRM just make things worse. The memories goes short, almost instantly, there is no repair for this. I already tried (many times) changing ALL memory modules, fix the VRM by replacing the bad MOSFET, the PWM (NCP81172) and the burned inductor, sometimes the burned shunt resistor. In some cases after all this, the board returns to output video, but with artifacts, in other cases without artifacts, but when driver installs, artifacts and crash instantly. I've discovered that in this cases, that dont give artifacts without video driver, if the board dont give erros on Hirens Video Memory Stress Test, its possible to edit BIOS to runs memory at freqs between 810-1100MHz, instead the normal 3500/3600MHz. In this cases, the board works normally, but with half or less of the power (40/50%), that is a side effect of running memory with slower speeds. The problem is in the GPU chip it self, that is permanently damaged by the short that happened on the memories. I already changed the GPU chip and this solved the problem, but by this point, its too much work and GPU chips are no cheap, a GM200-310 (980TI) can cost almost 250$... So that it, 12V going in to memory, do not only completely damage the memory modules, but partially the GPU itself. Maybe good fuses could save the GPU and maybe the memories on some cases, but I've never seen reference Nvidia board with good fuses. Never.
Felipe Muniz - Nice explanation. Thank you for your time. Would it be possible to improve a working card by replacing the weak components with better specced ones? I've got a Zotac 980ti reference.
Ah you sound like you have experience. Message me. I bought a “like new” gtx 980 ti off eBay. Bastard repaired it and resold it. Crashed constantly Took it apart. And this r33 was hanging off. Soldered it back down myself. It ran passmark perfectly. Then failed the 2nd time. And 3,4,5th time. Took it back apart my repair had popped off “cold solder joint?” Did reflow this time. Took a lot to get it to flow. Worried it was too much. Now get no image.
I was about to replace all the inductors. But I am not sure about downstream stuff. Is it just toast. I’m chasing the dragon of that time it worked perfectly for about 20 min when i soldered the bastard back down. I’m pretty sure it was repaired when I got it. I’m about to send it back. It’s mostly curiosity at this point though. If I can fix it. Or is the vram/ gpu permanently damaged? After this goes. Resistance reads 0 across all the inductors now. Is this normal. I don’t think they did befor I cooked them. But then spec sheet lists max resistance at like 120 mOhm. So. 0 sounds normal. How else to test? No I don’t have signal generator.
I have now this problem replaced the inductor and the power Mosfet 4901NF every time I start the PC the Mosfet got burned, what should i replace more???? Can you help me Felipe??
agreed, the power draw new with new caps may have been fine, tho after a few years of age on the caps, the leakage current may have pushed the amps over the limit for the inductors. Combined with downsized components for the minimum BOM costs, yeah, sounds about right. lol.
These GPU's are the most expensive ones you can buy when they are released. There is no excuse for cheap parts, not at the prices charged for flagship models.
facecrusher agreed, if not for one thing, Profits, lol. As a consumer, I agree that at some level, we all expect the stuff we spend money for to be good and work without bad things happening. Sadly tho from a manufacturers perspective, it's about trimming products down for the highest profits possible, and balancing minimum quality with quantity of sales. It's a fine line, and many things are in the end, built to within an inch of there lives, constantly on the edge of failing. If the stuff was selling at MSRP, I would understand the profits thing to a point. at going prices tho, I would expect the cards to be built to Spaceflight grade redundancy, lol.
My EVGA 980 ti SC+ 6GB popped just the other day.... Im gonna open it up and see if it puked all over the card... Hopefully I can bake it but I doubt it
I had a reference GTX 1070 that failed because one of those inductors blew. I know because I did tear it down before RMA'ing it. That in of itself wouldn’t be that significant... however the replacement card failed the exact same way after 3 months of normal use. When I returned it to Zotac, they would not replace the card with another reference design. It took 4 months of back and forth to get the issue resolved. I got frustrated after 2 months and purchased a GTX 1080. In the end they sent me a Zotac 1070 mini that has been ran for one hour to test that it works, and now it resides on a shelf collecting dust. To date the GTX 1080 has run flawlessly, and well outlasted the 1070's they replaced.
Damn, that's unfortunate. Bought a 980ti recently. Zotac reference. Well, Luckily it has space for a 10mm inductor. Would you make a video about soldering techniques next? lol Thanks for the video
I actually had this exact same issue. I was using a EVGA Gefore GTX 780 SC with the ACX cooler, I was playing some EVE Online and then all of the sudden I heard a loud bang and then smelled the magic smoke. I took my card apart and found exactly what you showed here.
980 Ti here, started as the strangest artifacting as memory corrupted due to this issue. Was a black mass that started in one area of the screen then quickly spread over the whole thing then the PC would lock for a second and reboot. This was 1 month from going out of warranty. Gigabyte was fast in fixing it and returning, but my fear is that the replacement part is from the same bin. What burns me is this replaced a 680GTX that failed in warranty and EVGA wouldn't cover it. Will never do business with them again.
my gtx 780 is still kicking, no problems. I use msi afterburner for fan curve and set pretty high lol. The first couple months owning it back in 2013 i felt like my video card was defective when i finally found out after replacing my hdmi to dvi cable for a hdmi cable alone and i still got this weird static green outline in moving objects when i outline my wallpaper background on desktop or watching videos or playing games. I found out my asus mx239H had a problem where the monitor i had trace free and sharpness enabled to 100%, this is the only case scenario where i use my monitor settings at just for games and i leave it at one setting all the time. Well i would get weird problems, like artifacting and goofing around i reset my settings in my monitor and change it back to the way i like it and it fixes it for couple days. Its very weird. Sometimes my monitor would go black for 2 seconds Been like this for 5 years
This might be one of those times were the engineers spec the correct inductor but them the accontant/contablist saw it costed too much and "suggested": "that 330nH inductor is a bit too expensive for production, could you use this one instead?"(of course the "could" is a "CHANGE THEM IMMEDIATELY")
I know this is several years late, but my 780 died this week due to an inductor failure. I am just happy it kept running as long as it did (almost 6 years)
This was kind of alarming for me at first. My 1080ti has what is probably the worst custom pcb available that isn't a downgrade from reference, the gigabyte gaming oc version, and I run it with a custom cooler. I knew nothing about the pcb before I bought it because no one else who cared had it, so I only have my own pictures to work with. Basically, the inductors in its 8 phase vrm are also these small ones (assuming the powerstages it uses are the standard size, which they should be), similar to the aorus pcb except there's 8 instead of 12. These are R22, so I put that into the same website you used in your video, and the highest current rating for that size is 16.8A. So according to that information, there's no way this thing would survive at full load since that would require around 32A per phase, but it clearly does without problems, so instead of getting worried I'm going to call your method into question. Your reasoning for this particular issue is sound, but either there's way better inductors somewhere else on the market, or something else is wrong with the numbers here. Or I made a mistake because its 7 in the morning and I haven't slept, wouldn't surprise me .
Just bought one sold as "used, working" that has this problem. The inductor furthest right is cracked and the shunt next to it is burnt out. A Vcore fet also has those overheat puked out solder balls. Don't buy from epc-texas on eBay.
the replaced one is actually a EVGA card and didnt work either after "replacing" i researched that one tryed to figure out what happend. turns out the weird replacement resistor or w/e it is, fried the owners card more AND his PSU. so.. PEOPLE, IF you REALLY want to replace an inductor, buy the RIGHT part with the same specs as the original inductor, or else everything goes to trash.
What is funny is I bought a 980ti with that choke missing. It worked, but with problems. Once I replaced it, everything worked fine. like you said it will probably blow again. sold it due to mining crazed prices
Huh, thanks for the video - this is interesting, I will take a look under the heatsink of my 780 GTX that blew up just a week ago. In my case the issue were a couple of SMD caps (c523, c525) blew up (violently). Apparently that's a common problem. I removed the c524 and c526 - from my measurements they are both 10 uF 0805 X7R's, so I am going to replace them and see what happens. However, now that you pointed out the inductor problem, I am going to check them to see if they look suspicious as well. If they look iffy, I may as well replace them too.
I think they inductors might have cracked due to thermal cycling or some other poor QC, when the inductor has air gaps it makes the inductor drop in inductance to the point where they cannot carry the magnetic field anymore and they start letting DC through. Though I'd expect that would make the high side fail as well.
TheLawnWanderer has it right. While it's true that inductors are not DC blocking, they are not just filters - they're not filtering the output of SMPS's, they *are* the SMPS. That's why they let 12 volts straight through when they fail.
inductors alone are NOT switch mode power supplies. I have designed quite a few buck mode switching power supplies and i can guarantee you that that is not the case. The passives in a switching power source are there to supplement the the driver. The inductor here is there to prevent sudden changes in current, and when paired with a capacitor, can give a stable DC output voltage and current source from what is basically a pwm voltage source from the driver ic.
I was making a point that the inductors are integral parts of the power supplies. You can't have a classical SMPS without an inductor. Your statement was this: "they are made to filter out high frequency noise which in this case is generated from a switched mode power source" - which is still false. They are not filtering noise, they are themselves integral to generating the supply in the first place. Inductors for ripple filtering would have different values and would not be placed close to the source. Nobody said that indoctors are DC blocking, so I'm not even sure why you wrote that in the first place.
Huh, interesting. I used a 780 reference model up until September 2017 (got a Vega 56) and it still worked just fine after being in two separate gaming rigs with long gaming sessions put on the card each day. I believe it was one of the first cards made since my cousin got it in May 2013 with some PC he built and ordered off this site.
Honestly, I'd say it's possibly a poorly assembled batch. Otherwise, we'd see a lot more by now. Even though it's the memory inductor, I'd be interested to see how many people with failures were using the voltage mod for circa 1.3v-1.35v. There's not a lot of tolerance on the Flextronics cards AFAIK
Two critical things: 1.) what is the manufacturer (e.g. Asus, Gigabyte, MSI, Zotac etc) as they receive the chips from Nvidia (not defending Nvidia in the least especially with their GPU monopoly incentive program). 2.) For those with dead cards taking off heatsinks just requires a screw-driver. I'm constantly fixing issues with computers on the side and I'm always thankful (for myself and my clients) when someone goes out of their way to properly post the necessary information the first time around.
I have a 980Ti refrence Model wich, causes the psu to shutdown via shortcurcitprotection. The funthing about it, only with a 3D load for a few minutes. On 2D evryting sems to be fine.
Happened the same to me three weeks ago. With my 980ti. PSU wouldn't send any current, then installed my old 780 (which I once changed for failing as well) and the pc runs well only in 2D app with the 780 (the 980ti is burnt, not booting and preventing my psu to turn on as I said before) but the funny thing is that now the 780 is showing some little burn marks behing the PCB just where the inductors are! Actually I'm waiting for a 2080 super to arrive home but I'm a little worried if my pci-e slot is damaged tho. Don't know if those shorts may caused problems to to pci-e slot
I wish I'd never heard of this video but one of my EVGA GTX 980 TI just burned. The location seems different though, the burnt chips say 024 and U9, closer to the edge, on the left of the power plug. The annoying part is that for some reason the other card doesn't get detected by windows (I was running them in SLI) and I'm stuck in 1024x768 right now.
I'm sure you know this, but the inductor has to have its coils insulated from each other, otherwise it just acts as a wire and will not oppose changes in current as will an inductor. Since it is in a sintered metal box, the coil also has to be insulated from the box. This is acomplished by a coating on the wire, which if it melted off would produce the liquid like residue outside inductor. I would blame either the coating material, its thickness, or the methode of attaching it to the wire as a first step.
Could it be the coating turning into snot from the card running too hot? I guess inductors generally don't reach high enough temps to require heatsink cooling?
12V turn into 1.6V? I was under the impression that a "transformer" does that but not on DC. I always thought that inductor "stores" current like capacitor "stores" voltage and both of them are use as filters. Could you explain pls?
Hi. Thanks for the link. In the linked video He himself calls the inductor a filter. I also think his understanding on how inductors work is incorrect. en.wikipedia.org/wiki/Inductor Inductors change current, not voltage.
Got a couple of MSI N780 3GB cheap for my latest build. V_V Really hope the waterblocks help keep this in check. Could it have been a thermal runaway issue, or a BIOS flash and badly monitored overclock......maybe?
does anyone know if PALIT also uses this kind of PCB design on their 780s and 780tis? Ive seen way too many of them on ebay listed as broken, like about 40% of all broken cards from this generation are from PALIT and PALIT isnt even that popular in germany.
Armin Berkemeier ich sag ja nicht, dass die Karten schlecht sind, hab selber ne Gainward Phantom GTX 780, aber es fällt eben auf dass so viele als kaputt verkauft werden
On the worst one of the bunch, a Titan Black I have blew on the inductor and shunt Resistor you pointed out but not the other two inductors (not probed it to confirm but yeah) so a design flaw in long term use? Not sure as it’s the PCIE side that’s blown going by the breakdown, having seen this and listened thoroughly I’m going to take another look as I dint remember inductors memory side being blown (maybe swollen) I have another Titan Black I’ve yet to test (missing a bit of it’s cooler, gotta take bits off of my dead one). I’m hoping it doesn’t happen again but there’s not too many working ones still around to have much hope, I didn’t see bulging at least not to the same degree on the dead one but they had it in a custom loop so I’m to assume they oc’d it Though saying that, I’m super concerned about the 2nd card simply blowing cause of previous usage /wear or age
Please help, I have a dead r33 on my 980ti, it looks to be the only thing damaged and want to try and replace the inductor I have pads the same size as the R22 on the right picture @ 23:20. Can you please give me a link to a good replacement. I know how to solder but nothing about inductors. Thanks in advance!
Buildzoid, you probably have already satisfied your curiosity, but I just ran across a blown inductor on a 780 card. Was wondering if you wanted it before I tossed it.
Paddydapro If they're under water, the temperatures will be significantly lower, and will help with the underpowered memory vrm. I wouldn't worry if I were you.
My 980Ti died 4 days ago and I took it apart found just the R33 component fried and everything else is fine. Where do I get that component? I would like to try to re-solder it and maybe fix it.
Is the design similar on the 10xx series of cards? If so, do we see a potential risk in memory overclocking (or buying used cards whose RAM was overclocked continuously), a thing that has been considered harmless until now, on NV cards? Possibly affects AMD designs too?
I have a 980ti with no signs of damage at all im overthinking this board for sure and causing myself a handicap problem with diagnostic I think the 3v pcie is grounded but get lost after that
My GTX 780 AMP had a short in the PCB underneath a GPU phase... I got rid of the short by drilling a 10mm hole in the PCB, i just have to epower it and hope that it didn't send 12V to the GPU...
my geforce gtx 780 from msi got burned out yesterday. I swear this is the saddest day of my life?? what is the solutions now? buy a new 1200$ card? what the fuck is wrong with companies where days. I can even smell the burnt part if the gpu... hughhhh I guess this is just life. the card is still wicked clean from the outside tho...
Hi do know where is located circuit of fan controller/voltage ,on me titan x fan do not work I replaced it to new one but still fan do not work the, rpm is working an I have 12v on pin 1 and 2 .
My 980ti blew up, something around the middle, maybe closer to the side with the power plugs died. I saw smoke coming from it. How do I send it to you? I have the 980ti Hybrid from EVGA with a waterblock on it. I will soon be disassembling my entire loop just to see what it looks like under the system
How do I check if mosfet is intact? How do I check if the rest is intact. I used to get video, it would just crash. After trying to reflow it, I get nothing. I’d be happy to push 10 A through a new one of these. I think my PSU only goes up to 10A
I got this thing on a Tesla m40, is very similar pcb to Titan X. Shunt bulged and inductor cracked (L14 on the Titan X and closest shunt). My motherboard VRM got toasted with magic smoke. Memory VRM doesn't looks damaged (mosfets look ok and inductors as well from the outside). Would have to test it electrically. Is this repairable? I hope 12v didn't come into my VRAM. (no fuse no shutdown if something goes out of specs.) Pff awful nVidia design.
I found my vrms seems to be fine however my 12v pciex lines look shorted giving 3.9 ohms with ground. Any Idea on what feeds that L14 inductor? I am trying to figure out what is shorted because that is likely what caused cracked inductor and bulged shunt.
I have a gainward phoenix gtx 980 ti reference pcb ( reference clocks ) and under load pc will reboot . The fix was to mod bios and decrease core clock and voltage (-200mhz , -100mv) . PSU is Seasonic 620w. Are inductors failing on this card ?
Actually Hardcore Overclocking msi is so strange.... They made mobos they blew up (think it was mostly people OCing on low cost boards).....yet gpus that are solid as hell My msi 990fxa dg80 v1 also lived from phenom2 6 core launch day till a month after ryzen dropped 1.55 volts.....that vrm was a time bomb Got quite warm
My friend's 6970 sapphire reference died in 2 months and had lots of driver issue. He is not a power user though. So he got annoyed with issues and got a 760. According to him, he had zero issues for 3 + years before he sold the 760 msi. What's your opinion on that?
True AF. my gtx 660Ti PE was super solid whereas msi z68 gd65 died. 3 of my friends replaced and went through like 5 p67/z68 gd65s. All died within 1-2 years. msi the weirdos
Is it possible to damage motherboard components with this problem? My PC tried auto-rebooting after titanXM started to fry and now I can't even post to bios with integrated graphics...
I had a evga 780ti reference that released smoke about 6 months after i purchased it, i never looked where the smoke came from but I wonder if it could have been this :thinking:
I have seen three 970s with burnt out inductors already, though not entirely sure it was mem ones. Could be a coincidence, but that's a pretty bad track record for 3 yo cards =/
Is there a way to modify your card to help avoid this? I'm no computer wiz, so it would hopefully have to be simple(ish). I have a gtx 980ti hybrid card.
Do you think this is just limited to higher end cards, or are entire series of cards have underspec components? What about mobile chips? I assume whoever designs laptop motherboards uses whatever reference design they're given.
Now are these actually cards used as intended, or are people abusing the shit out of them, frying them, then complaining? Is there any way to verify it if they did abuse it?
I have a dead 980 ti Inno3D Hybrid S one here and wanna fix it. Same R33 error and wanna replace the inductor. Any suggestions? Maybe a better inductor to not let this happen again. Thanks in advance folks
Ive experienced and Issue with R9 Fury cards (AMD pcb) After mining (undervolted, high fan speed and lots of cool air), the card crashes in games, and under mining load i have observed that the card is unable to hold stable voltage/clock rates above 940mhz. If the clock speed is left at default and a load is put on the card the voltage and core clock fluctuate rapidly. 1.05-1.2v and 900-980Mhz. This issue is experienced on 2 separate cards (first one was RMA'd) in 2 different systems. I am wondering if there is a Capacitor problem with the card somewhere. the card I currently have is out of warranty so i was unable to rma it. i took the heatsink off of it and could not see any apparent physical evidence of a problem. Ive looked online and have been unable to find anyone with a similar issue (which may be due to cards being used to mine rather than game). any idea what i should look for?
my 980ti kept crashing my pc the last few weeks when playing games, but last week it crasht hard and my pc would not boot ad all. After i put my old card back in it was able to boot again. but with the 980ti i could use it for websites, but every thing that whas asking more from the card it would crash i dont know what to look for and dont know how to fix this if it's even able to be fixt
Hello man, i have a 980ti with r33 fried and 1r0 that shaterred. i've purchased them to replace them, once i solder them, will it work in your opinion? thank you
Yep. My Gtx Geforce 780 Physx Cuda OC died about 6 months ago. Now i looked it up on youtube and someone recomended the "oven solution" by putting the GPU inside the oven for 8 minutes on 200 degrees celsius. I thaugt why not? So I unscrewed the 780 and saw this burned up little square thing that said "R33" on it and it looks just the same like GPU's in your video. I tried it out anyways. when i installed it in the pc and started it up, the computer actually booted up, but there was no picture on the monitor. As i was going to shut the computer down i heard thi wierd noice and a wierd smell came out of the GPU. When i uscrewed it again i saw another "memory inductor" was blown up... Is there any way to replace theese memory inductors? Btw, sorry for my English writing. I really do suck in English. :) Hope for reply from you @Actually Hardcore Overclocking.
I'm guessing this is probably a heat issue of some sort. High heat breaking down the internal wire insulation of the inductors. I took a photo of one of my old 780Ti's, they ran highly overclocked for like 4 years, but on a fullcover water block, so they look pristine. i.imgur.com/N5ghZA0.jpg
I burnt out my graphic card gtx 760 blower type in April 2018, its vram is faulty and i concerned the inductors were blowning out as you mentioned. Do you have any model references of the replacement of these memory inductors because i gonna replace them out to revive my gpu back to the life.
Not trying to stir up controversy here. I've just seen way too many of these in way too short a time period to ignore it.
clearly an amd fanboy because a nvidia fanboy would've kept quiet about this /s
Well this sucks, having a 980ti.
if you have an gpu affected from this, is there anything that can prevent the memory vrm from blowing up? Does a bios flash for gpu core power affect this?
How do you know if your card is of reference design ? I have a Evga 780 6gb Super clocked ACX cooler card that I have been hold onto hopefully until the end of the mining crazy. Is this card affected ?
Superclocked cards are reference PCBs
either undervolt the memory or replace the chokes before they go nuclear
This happens a lot, I've been fixing GPUs for years now, and this is a typical problem with memory on reference boards, especially on this GK110 and GM200 reference boards like EVGA, or reference design based boards, like some models from ZOTAC, MSI, etc.. The problem happens when typically a high side MOSFET from memory VRM goes short. The lack of a good fuse (10A white fuses) instead a inductor on the 12V line that feeds the m-VRM just make things worse. The memories goes short, almost instantly, there is no repair for this. I already tried (many times) changing ALL memory modules, fix the VRM by replacing the bad MOSFET, the PWM (NCP81172) and the burned inductor, sometimes the burned shunt resistor. In some cases after all this, the board returns to output video, but with artifacts, in other cases without artifacts, but when driver installs, artifacts and crash instantly. I've discovered that in this cases, that dont give artifacts without video driver, if the board dont give erros on Hirens Video Memory Stress Test, its possible to edit BIOS to runs memory at freqs between 810-1100MHz, instead the normal 3500/3600MHz. In this cases, the board works normally, but with half or less of the power (40/50%), that is a side effect of running memory with slower speeds. The problem is in the GPU chip it self, that is permanently damaged by the short that happened on the memories. I already changed the GPU chip and this solved the problem, but by this point, its too much work and GPU chips are no cheap, a GM200-310 (980TI) can cost almost 250$...
So that it, 12V going in to memory, do not only completely damage the memory modules, but partially the GPU itself. Maybe good fuses could save the GPU and maybe the memories on some cases, but I've never seen reference Nvidia board with good fuses. Never.
Felipe Muniz - Nice explanation. Thank you for your time.
Would it be possible to improve a working card by replacing the weak components with better specced ones?
I've got a Zotac 980ti reference.
Ah you sound like you have experience. Message me.
I bought a “like new” gtx 980 ti off eBay. Bastard repaired it and resold it. Crashed constantly
Took it apart. And this r33 was hanging off. Soldered it back down myself. It ran passmark perfectly. Then failed the 2nd time. And 3,4,5th time. Took it back apart my repair had popped off “cold solder joint?” Did reflow this time. Took a lot to get it to flow. Worried it was too much. Now get no image.
I was about to replace all the inductors. But I am not sure about downstream stuff. Is it just toast.
I’m chasing the dragon of that time it worked perfectly for about 20 min when i soldered the bastard back down.
I’m pretty sure it was repaired when I got it. I’m about to send it back. It’s mostly curiosity at this point though. If I can fix it. Or is the vram/ gpu permanently damaged? After this goes.
Resistance reads 0 across all the inductors now. Is this normal. I don’t think they did befor I cooked them. But then spec sheet lists max resistance at like 120 mOhm. So. 0 sounds normal. How else to test? No I don’t have signal generator.
So I shouldnt try to replace the inductor? It wont work after the replacement?
I have now this problem replaced the inductor and the power Mosfet 4901NF every time I start the PC the Mosfet got burned, what should i replace more???? Can you help me Felipe??
Maybe its planned obsolescence.
Not maybe, its pretty much guaranteed that if you oc your card it will fail, given enough time the un-oc cards will fail too.
cheaping out on inductors -> artificially increasing GDDR5/GPU demand --> profit
Thats why they are making records in profit
Ngreedia cought again...
_Planned obsolescence_ seems what Nvidia is expert with. :D
agreed, the power draw new with new caps may have been fine, tho after a few years of age on the caps, the leakage current may have pushed the amps over the limit for the inductors. Combined with downsized components for the minimum BOM costs, yeah, sounds about right. lol.
Zarcondeegrissom talk out ur ass much
These GPU's are the most expensive ones you can buy when they are released. There is no excuse for cheap parts, not at the prices charged for flagship models.
facecrusher agreed, if not for one thing, Profits, lol. As a consumer, I agree that at some level, we all expect the stuff we spend money for to be good and work without bad things happening.
Sadly tho from a manufacturers perspective, it's about trimming products down for the highest profits possible, and balancing minimum quality with quantity of sales. It's a fine line, and many things are in the end, built to within an inch of there lives, constantly on the edge of failing.
If the stuff was selling at MSRP, I would understand the profits thing to a point. at going prices tho, I would expect the cards to be built to Spaceflight grade redundancy, lol.
Anyone still remembers the famous 'nvidia bug' (Nvidia bump issue) in laptops from 2007-2010?
I lost my thinkpad T61p due to it :(
The magic smoke has escaped!
RELEASE THE SHMOO
if you want a dead 780 that died in a similar way to this , i have one here , you can have it for free/ maybe shipping costs
Justagermannerd I'm in
Justagermannerd ye I will pay you for shipping if you will send to my of course if you still have that card
it was more directed towards Buildzoid , since hes in the UK and i can ship it there easely
Justagermannerd he Iam from UK as well my friend and I focusing on repair gpu as well
This happened to me and yes i have pics. MY Evga 980ti from September 2015 caught fire in exactly this way in January 2018.
I have a Zotac 980ti. Now I feel frightened.
Zotac non ref 980Ti was v good.
My EVGA 980 ti SC+ 6GB popped just the other day.... Im gonna open it up and see if it puked all over the card... Hopefully I can bake it but I doubt it
I had a reference GTX 1070 that failed because one of those inductors blew. I know because I did tear it down before RMA'ing it. That in of itself wouldn’t be that significant... however the replacement card failed the exact same way after 3 months of normal use. When I returned it to Zotac, they would not replace the card with another reference design. It took 4 months of back and forth to get the issue resolved. I got frustrated after 2 months and purchased a GTX 1080. In the end they sent me a Zotac 1070 mini that has been ran for one hour to test that it works, and now it resides on a shelf collecting dust. To date the GTX 1080 has run flawlessly, and well outlasted the 1070's they replaced.
Seems Nvidia is taking notes of Apple's business model
Top end, most expensive cards => cheaping out on the inductors.
smh
This was helpful in confirming my K4200 is screwed. Once I got the heat sink off, I could see the inductor was burnt, and the MOSFET beside it.
Damn, that's unfortunate. Bought a 980ti recently. Zotac reference.
Well, Luckily it has space for a 10mm inductor. Would you make a video about soldering techniques next? lol
Thanks for the video
I actually had this exact same issue. I was using a EVGA Gefore GTX 780 SC with the ACX cooler, I was playing some EVE Online and then all of the sudden I heard a loud bang and then smelled the magic smoke. I took my card apart and found exactly what you showed here.
980 Ti here, started as the strangest artifacting as memory corrupted due to this issue. Was a black mass that started in one area of the screen then quickly spread over the whole thing then the PC would lock for a second and reboot. This was 1 month from going out of warranty. Gigabyte was fast in fixing it and returning, but my fear is that the replacement part is from the same bin. What burns me is this replaced a 680GTX that failed in warranty and EVGA wouldn't cover it. Will never do business with them again.
my gtx 780 is still kicking, no problems. I use msi afterburner for fan curve and set pretty high lol. The first couple months owning it back in 2013 i felt like my video card was defective when i finally found out after replacing my hdmi to dvi cable for a hdmi cable alone and i still got this weird static green outline in moving objects when i outline my wallpaper background on desktop or watching videos or playing games. I found out my asus mx239H had a problem where the monitor i had trace free and sharpness enabled to 100%, this is the only case scenario where i use my monitor settings at just for games and i leave it at one setting all the time. Well i would get weird problems, like artifacting and goofing around i reset my settings in my monitor and change it back to the way i like it and it fixes it for couple days. Its very weird. Sometimes my monitor would go black for 2 seconds Been like this for 5 years
Oh, so that's why my friends 780 died...
You should still check it before assuming it.
Yeah checking seems wise
under spec pcb in 999$ card (titan XM)...
well, component, doesnt count for the rest of the pcb
Aw heck, just when i bought a used 980ti G1 GAMING and got it under a H55 with a kraken g12 - gotta hope I don't run into these issues.
This might be one of those times were the engineers spec the correct inductor but them the accontant/contablist saw it costed too much and "suggested": "that 330nH inductor is a bit too expensive for production, could you use this one instead?"(of course the "could" is a "CHANGE THEM IMMEDIATELY")
I know this is several years late, but my 780 died this week due to an inductor failure. I am just happy it kept running as long as it did (almost 6 years)
This was kind of alarming for me at first. My 1080ti has what is probably the worst custom pcb available that isn't a downgrade from reference, the gigabyte gaming oc version, and I run it with a custom cooler. I knew nothing about the pcb before I bought it because no one else who cared had it, so I only have my own pictures to work with. Basically, the inductors in its 8 phase vrm are also these small ones (assuming the powerstages it uses are the standard size, which they should be), similar to the aorus pcb except there's 8 instead of 12. These are R22, so I put that into the same website you used in your video, and the highest current rating for that size is 16.8A. So according to that information, there's no way this thing would survive at full load since that would require around 32A per phase, but it clearly does without problems, so instead of getting worried I'm going to call your method into question. Your reasoning for this particular issue is sound, but either there's way better inductors somewhere else on the market, or something else is wrong with the numbers here. Or I made a mistake because its 7 in the morning and I haven't slept, wouldn't surprise me .
Just bought one sold as "used, working" that has this problem. The inductor furthest right is cracked and the shunt next to it is burnt out. A Vcore fet also has those overheat puked out solder balls. Don't buy from epc-texas on eBay.
the replaced one is actually a EVGA card and didnt work either after "replacing" i researched that one tryed to figure out what happend. turns out the weird replacement resistor or w/e it is, fried the owners card more AND his PSU. so.. PEOPLE, IF you REALLY want to replace an inductor, buy the RIGHT part with the same specs as the original inductor, or else everything goes to trash.
What is funny is I bought a 980ti with that choke missing. It worked, but with problems. Once I replaced it, everything worked fine. like you said it will probably blow again. sold it due to mining crazed prices
Huh, thanks for the video - this is interesting, I will take a look under the heatsink of my 780 GTX that blew up just a week ago. In my case the issue were a couple of SMD caps (c523, c525) blew up (violently). Apparently that's a common problem. I removed the c524 and c526 - from my measurements they are both 10 uF 0805 X7R's, so I am going to replace them and see what happens.
However, now that you pointed out the inductor problem, I am going to check them to see if they look suspicious as well. If they look iffy, I may as well replace them too.
I think they inductors might have cracked due to thermal cycling or some other poor QC, when the inductor has air gaps it makes the inductor drop in inductance to the point where they cannot carry the magnetic field anymore and they start letting DC through.
Though I'd expect that would make the high side fail as well.
Inductors are not dc blocking, they are made to filter out high frequency noise which in this case is generated from a switched mode power source.
TheLawnWanderer has it right. While it's true that inductors are not DC blocking, they are not just filters - they're not filtering the output of SMPS's, they *are* the SMPS. That's why they let 12 volts straight through when they fail.
inductors alone are NOT switch mode power supplies. I have designed quite a few buck mode switching power supplies and i can guarantee you that that is not the case. The passives in a switching power source are there to supplement the the driver. The inductor here is there to prevent sudden changes in current, and when paired with a capacitor, can give a stable DC output voltage and current source from what is basically a pwm voltage source from the driver ic.
I was making a point that the inductors are integral parts of the power supplies. You can't have a classical SMPS without an inductor. Your statement was this: "they are made to filter out high frequency noise which in this case is generated from a switched mode power source" - which is still false. They are not filtering noise, they are themselves integral to generating the supply in the first place. Inductors for ripple filtering would have different values and would not be placed close to the source.
Nobody said that indoctors are DC blocking, so I'm not even sure why you wrote that in the first place.
Yep. Now my next card will be an RX Vega.
I have over 2000 700 and 900 series running on my farms and they are lasting just like normal. As they die you can have them all.
Huh, interesting. I used a 780 reference model up until September 2017 (got a Vega 56) and it still worked just fine after being in two separate gaming rigs with long gaming sessions put on the card each day. I believe it was one of the first cards made since my cousin got it in May 2013 with some PC he built and ordered off this site.
Honestly, I'd say it's possibly a poorly assembled batch. Otherwise, we'd see a lot more by now. Even though it's the memory inductor, I'd be interested to see how many people with failures were using the voltage mod for circa 1.3v-1.35v. There's not a lot of tolerance on the Flextronics cards AFAIK
Two critical things: 1.) what is the manufacturer (e.g. Asus, Gigabyte, MSI, Zotac etc) as they receive the chips from Nvidia (not defending Nvidia in the least especially with their GPU monopoly incentive program). 2.) For those with dead cards taking off heatsinks just requires a screw-driver. I'm constantly fixing issues with computers on the side and I'm always thankful (for myself and my clients) when someone goes out of their way to properly post the necessary information the first time around.
You can find broken reference PCB 780's instantly on ebay, normally there are not that many broken cards with the same GPU on there.
My Gigabyte Windforce 780 Ti is still running strong :D Been used a hell of a lot. Changed the paste out recently and it runs cooler than new lol
Unhappy 'lil inductor :(
good to know about the 1.4v on vcore being a problem, i run my 780 at 1.36 24/7 and was debating going farther....
This is interesting and as I focus o card repair as well This will definitely help me thanks for sharing
Just brought a used 780 and bingo, same thing in the same place.
I have a 980Ti refrence Model wich, causes the psu to shutdown via shortcurcitprotection.
The funthing about it, only with a 3D load for a few minutes. On 2D evryting sems to be fine.
Happened the same to me three weeks ago. With my 980ti. PSU wouldn't send any current, then installed my old 780 (which I once changed for failing as well) and the pc runs well only in 2D app with the 780 (the 980ti is burnt, not booting and preventing my psu to turn on as I said before) but the funny thing is that now the 780 is showing some little burn marks behing the PCB just where the inductors are! Actually I'm waiting for a 2080 super to arrive home but I'm a little worried if my pci-e slot is damaged tho. Don't know if those shorts may caused problems to to pci-e slot
I found on my 980Ti PCB one of the inductors for the memory is cracked (the botom one).
I wish I'd never heard of this video but one of my EVGA GTX 980 TI just burned. The location seems different though, the burnt chips say 024 and U9, closer to the edge, on the left of the power plug.
The annoying part is that for some reason the other card doesn't get detected by windows (I was running them in SLI) and I'm stuck in 1024x768 right now.
I'm sure you know this, but the inductor has to have its coils insulated from each other, otherwise it just acts as a wire and will not oppose changes in current as will an inductor. Since it is in a sintered metal box, the coil also has to be insulated from the box. This is acomplished by a coating on the wire, which if it melted off would produce the liquid like residue outside inductor. I would blame either the coating material, its thickness, or the methode of attaching it to the wire as a first step.
Could it be the coating turning into snot from the card running too hot? I guess inductors generally don't reach high enough temps to require heatsink cooling?
There is an optimal temp for ferrite inductors of 90c, it falls off below 80c and above 100c.
12V turn into 1.6V? I was under the impression that a "transformer" does that but not on DC. I always thought that inductor "stores" current like capacitor "stores" voltage and both of them are use as filters. Could you explain pls?
he has a video or two on how vrms work. heres one ruclips.net/video/oDRHV3qtSWc/видео.html
Hi. Thanks for the link. In the linked video He himself calls the inductor a filter. I also think his understanding on how inductors work is incorrect.
en.wikipedia.org/wiki/Inductor
Inductors change current, not voltage.
Got a couple of MSI N780 3GB cheap for my latest build. V_V Really hope the waterblocks help keep this in check. Could it have been a thermal runaway issue, or a BIOS flash and badly monitored overclock......maybe?
does anyone know if PALIT also uses this kind of PCB design on their 780s and 780tis? Ive seen way too many of them on ebay listed as broken, like about 40% of all broken cards from this generation are from PALIT and PALIT isnt even that popular in germany.
possibly. This is the Nvidia reference design so it should be very common.
Luis Hagner Palit ist doch eigentlich echt super. Die 10er Reihe von Palit / Gainward war eine der wenigen guten
Armin Berkemeier ich sag ja nicht, dass die Karten schlecht sind, hab selber ne Gainward Phantom GTX 780, aber es fällt eben auf dass so viele als kaputt verkauft werden
Luis Hagner I think páliť is good brand
Frog - Tech - Tips Gigabyte and MSI arent "bad brands" either, yet they still have those weak/bad parts on their cards
On the worst one of the bunch, a Titan Black I have blew on the inductor and shunt Resistor you pointed out but not the other two inductors (not probed it to confirm but yeah) so a design flaw in long term use? Not sure as it’s the PCIE side that’s blown going by the breakdown, having seen this and listened thoroughly I’m going to take another look as I dint remember inductors memory side being blown (maybe swollen)
I have another Titan Black I’ve yet to test (missing a bit of it’s cooler, gotta take bits off of my dead one). I’m hoping it doesn’t happen again but there’s not too many working ones still around to have much hope, I didn’t see bulging at least not to the same degree on the dead one but they had it in a custom loop so I’m to assume they oc’d it
Though saying that, I’m super concerned about the 2nd card simply blowing cause of previous usage /wear or age
Please help, I have a dead r33 on my 980ti, it looks to be the only thing damaged and want to try and replace the inductor I have pads the same size as the R22 on the right picture @ 23:20. Can you please give me a link to a good replacement. I know how to solder but nothing about inductors. Thanks in advance!
Buildzoid, you probably have already satisfied your curiosity, but I just ran across a blown inductor on a 780 card. Was wondering if you wanted it before I tossed it.
tfw have 2 980ti reference in my rig but under water.. they oc pretty nicely but hope they wont die on me lol
Paddydapro
If they're under water, the temperatures will be significantly lower, and will help with the underpowered memory vrm. I wouldn't worry if I were you.
This happened to me on my EVGA SC GTX 780 back in 2015
Glad I went with a classy
My 980Ti died 4 days ago and I took it apart found just the R33 component fried and everything else is fine. Where do I get that component? I would like to try to re-solder it and maybe fix it.
Hey, can you recommend one R33 and 1R0 inductor?
The same ones as everybody elses' blew up
I have a 780ti that recently stoped working and i could send it to you for you to take a look at and see wants wrong
I just bought a 980ti Evga from offer up and I got screwed with the same problem . Where can I get the R33 inductor ? Help please
Is the design similar on the 10xx series of cards?
If so, do we see a potential risk in memory overclocking (or buying used cards whose RAM was overclocked continuously), a thing that has been considered harmless until now, on NV cards? Possibly affects AMD designs too?
I have a 980ti with no signs of damage at all im overthinking this board for sure and causing myself a handicap problem with diagnostic I think the 3v pcie is grounded but get lost after that
My GTX 780 AMP had a short in the PCB underneath a GPU phase...
I got rid of the short by drilling a 10mm hole in the PCB, i just have to epower it and hope that it didn't send 12V to the GPU...
how i can remove the things near the R22 stuffs there i guess they are inductors, well the chips near them are in short in my only GTX780 Ti from EVGA
my geforce gtx 780 from msi got burned out yesterday. I swear this is the saddest day of my life?? what is the solutions now? buy a new 1200$ card? what the fuck is wrong with companies where days. I can even smell the burnt part if the gpu... hughhhh I guess this is just life. the card is still wicked clean from the outside tho...
Why can't you bios mod an nVidia GPU? Because the ticking time bomb is counting down to cause an overvoltage situation, so you buy a new GPU.
Hi do know where is located circuit of fan controller/voltage ,on me titan x fan do not work I replaced it to new one but still fan do not work the, rpm is working an I have 12v on pin 1 and 2 .
just had my gigabyte g1 980 ti burn up similar to this i have pics if you want to analyze it
My 980ti blew up, something around the middle, maybe closer to the side with the power plugs died. I saw smoke coming from it. How do I send it to you?
I have the 980ti Hybrid from EVGA with a waterblock on it. I will soon be disassembling my entire loop just to see what it looks like under the system
So would undervolting your memory 15% (and downclocking to stable mem speeds) solve this problem if you are willing to take a performance hit?
How do I check if mosfet is intact? How do I check if the rest is intact. I used to get video, it would just crash.
After trying to reflow it, I get nothing.
I’d be happy to push 10 A through a new one of these. I think my PSU only goes up to 10A
I got this thing on a Tesla m40, is very similar pcb to Titan X. Shunt bulged and inductor cracked (L14 on the Titan X and closest shunt). My motherboard VRM got toasted with magic smoke. Memory VRM doesn't looks damaged (mosfets look ok and inductors as well from the outside). Would have to test it electrically. Is this repairable? I hope 12v didn't come into my VRAM. (no fuse no shutdown if something goes out of specs.) Pff awful nVidia design.
I found my vrms seems to be fine however my 12v pciex lines look shorted giving 3.9 ohms with ground. Any Idea on what feeds that L14 inductor? I am trying to figure out what is shorted because that is likely what caused cracked inductor and bulged shunt.
I have a gainward phoenix gtx 980 ti reference pcb ( reference clocks ) and under load pc will reboot . The fix was to mod bios and decrease core clock and voltage (-200mhz , -100mv) . PSU is Seasonic 620w. Are inductors failing on this card ?
and my launch month msi lightning 6970 is still chugging along.....
but.... i am going to replace it as soon as the prices even out
well it's a lightning. Those are basically the definition of overkill
Actually Hardcore Overclocking msi is so strange....
They made mobos they blew up (think it was mostly people OCing on low cost boards).....yet gpus that are solid as hell
My msi 990fxa dg80 v1 also lived from phenom2 6 core launch day till a month after ryzen dropped
1.55 volts.....that vrm was a time bomb
Got quite warm
My friend's 6970 sapphire reference died in 2 months and had lots of driver issue. He is not a power user though. So he got annoyed with issues and got a 760. According to him, he had zero issues for 3 + years before he sold the 760 msi. What's your opinion on that?
True AF. my gtx 660Ti PE was super solid whereas msi z68 gd65 died. 3 of my friends replaced and went through like 5 p67/z68 gd65s. All died within 1-2 years. msi the weirdos
Is it possible to damage motherboard components with this problem? My PC tried auto-rebooting after titanXM started to fry and now I can't even post to bios with integrated graphics...
Nvidia's superior products in action. Maybe another "bumpgate"?
I had a evga 780ti reference that released smoke about 6 months after i purchased it, i never looked where the smoke came from but I wonder if it could have been this :thinking:
I have an gtx 780 has similar board but there is tantalum capacitor in it and and one is shorted I can't find similar help me
I have seen three 970s with burnt out inductors already, though not entirely sure it was mem ones. Could be a coincidence, but that's a pretty bad track record for 3 yo cards =/
My VEGA 64 scoffs at the thought of this!
My Vega 64 popped an inductor too, that's why I'm watching this video...
@@A86140 mine did exactly the same not long after I posted this comment lmfao.
Attack on Titan's inductors
Is there a way to modify your card to help avoid this? I'm no computer wiz, so it would hopefully have to be simple(ish). I have a gtx 980ti hybrid card.
What about ASUS Strix GTX 980s? (Not the ti model) Are these affected too?
7:33 that fine dust on capacitors, etc, is that ok? can it ever short or cause overheating or even micro fires?
got a gtx titan x and my voltage controller is getting realy hot and got no display what can it be?
i have 2 reference 780tis laying around, neither of them have their mem vrm blown up though
Do you think this is just limited to higher end cards, or are entire series of cards have underspec components? What about mobile chips? I assume whoever designs laptop motherboards uses whatever reference design they're given.
Now are these actually cards used as intended, or are people abusing the shit out of them, frying them, then complaining? Is there any way to verify it if they did abuse it?
9:43-10:20
If the high side failed closed and the inductor blew and 12v went through, wouldn't the memory caps also blow aswell?
Depends on its max voltage rating
I have a dead 980 ti Inno3D Hybrid S one here and wanna fix it. Same R33 error and wanna replace the inductor. Any suggestions? Maybe a better inductor to not let this happen again. Thanks in advance folks
Ive experienced and Issue with R9 Fury cards (AMD pcb) After mining (undervolted, high fan speed and lots of cool air), the card crashes in games, and under mining load i have observed that the card is unable to hold stable voltage/clock rates above 940mhz. If the clock speed is left at default and a load is put on the card the voltage and core clock fluctuate rapidly. 1.05-1.2v and 900-980Mhz. This issue is experienced on 2 separate cards (first one was RMA'd) in 2 different systems. I am wondering if there is a Capacitor problem with the card somewhere. the card I currently have is out of warranty so i was unable to rma it. i took the heatsink off of it and could not see any apparent physical evidence of a problem. Ive looked online and have been unable to find anyone with a similar issue (which may be due to cards being used to mine rather than game). any idea what i should look for?
What about the newer 1080ti reference etc?
my 980ti kept crashing my pc the last few weeks when playing games, but last week it crasht hard and my pc would not boot ad all.
After i put my old card back in it was able to boot again.
but with the 980ti i could use it for websites, but every thing that whas asking more from the card it would crash
i dont know what to look for and dont know how to fix this if it's even able to be fixt
Hello man, i have a 980ti with r33 fried and 1r0 that shaterred.
i've purchased them to replace them, once i solder them, will it work in your opinion?
thank you
hi mate i have a 980ti that has blown a mosfet just wondered if you could help me out on fixing it if it can be fixed i can send photos if my card :)
Yep. My Gtx Geforce 780 Physx Cuda OC died about 6 months ago. Now i looked it up on youtube and someone recomended the "oven solution" by putting the GPU inside the oven for 8 minutes on 200 degrees celsius. I thaugt why not? So I unscrewed the 780 and saw this burned up little square thing that said "R33" on it and it looks just the same like GPU's in your video. I tried it out anyways. when i installed it in the pc and started it up, the computer actually booted up, but there was no picture on the monitor. As i was going to shut the computer down i heard thi wierd noice and a wierd smell came out of the GPU. When i uscrewed it again i saw another "memory inductor" was blown up... Is there any way to replace theese memory inductors?
Btw, sorry for my English writing. I really do suck in English. :) Hope for reply from you @Actually Hardcore Overclocking.
I'm guessing this is probably a heat issue of some sort. High heat breaking down the internal wire insulation of the inductors. I took a photo of one of my old 780Ti's, they ran highly overclocked for like 4 years, but on a fullcover water block, so they look pristine. i.imgur.com/N5ghZA0.jpg
I have this issue with a PNY XLR (lifetime warranty that they never responded) gtx 700 series.
I burnt out my graphic card gtx 760 blower type in April 2018, its vram is faulty and i concerned the inductors were blowning out as you mentioned. Do you have any model references of the replacement of these memory inductors because i gonna replace them out to revive my gpu back to the life.
What about the reference 980 non ti? I have 2 of them in sli.
How tho.
I have never seen a inductor on a switch mode power supply fail.
Join the club. Until I saw several of these I thought it was a freak one of accident.
Is the MSI gtx-1080 okay or not okay,.. that's the question.
would it be possible to buy that memory inductor and replace it myself? my 980ti went toast tonight
how i can find the sam inductors and how i can now the low mos from the high
Can burned out or not-fully dead memory inductor cause an artifacts in the game? Or the card will be completely dead at this moment?