How to quickly & easily fix coil-whine(coil choke noise)

[Wolverine2349]

There is no reason to wait that long. The application is usable within 30 minutes to an hour, tops.


True, this does happen.


That's just weird. The backplate shouldn't have anything to do with it...

Can it hurt waiting that long? Like putting it back together sooner could the almost dried super glue get pressure and mush better to rid the whine. Or does it make 0 difference?

 

[Shrek]

Or use an activator to accelerate setting

 

[Wolverine2349]

What about replacing the loud inductors with quiet ones? Is that possible to do?

I could test the card for loud inductors while taking the risk of course of frying the card. And maybe replace the loud inductors?

I am so sick of coil whine and so badly want a quiet 4090 ASAP and call it a system for a few years at least.

What are the dimensions of the AD102 die for a temp heatsink to stick on it. so I can find the loud inductors. Obviously something almost as big but not bigger or I know it could short it as it is direct die. How the heck else am I going to find and deal with noisy inductors if I cannot find where they are.

 

[Calenhad]

So i dont think the glue had anything to do with this entire situation. I did add more glue after my test this morning where the card was silent without the backplate. After putting more glue on, waiting for it to dry, then putting everything back together including the backplate, the whine was back same as ever. Im almost certain its got to do with the slight pressure adjustment from having a backplate, to just pcb and waterblock. The chokes have a little more pressure on them now and thats stopping some vibration. Im no engineer so i cant prove my theory but process of elimination tells me the glue didnt do much.

Edit: So i noticed there are rubber gromets below the standoffs of the backplate. My guess is with the backplate, the screws stop at the top of the standoffs. Without it, im able to apply just slightly more pressure, pushing down on those gromets. Again, just a theory

Resonance can be fun. Sounds like your chokes and backplate like to sing in harmony. If you have a support (or vertical mount) for your GPU and don't need the backplate for looks, you should be able to run it just fine without. Since it sounds like it is possible to properly mount the cooler without the backplate. If you really want to keep experimenting, it is theoretically possible to dampen the backplate with a couple thermal pads, but where you have to place them is a very good question. And make sure they are non-conductive if you try.



Why alot of people in this thread is willing to both void the warranty and possibly ruin a very expensive GPU at the same time, instead of having it RMA'd is beyond me. But it is not my job to stop you guys from experimenting.

 

[lexluthermiester]

Can it hurt waiting that long?

No, it doesn't hurt anything, but is just a waste of time.

Or does it make 0 difference?

It make zero difference. Chemically, superglue is set within 5 minutes, but with the thermal differentials that exist on a video card, the final stage of the curing process needs an hour or 2 to finish.

Why alot of people in this thread is willing to both void the warranty and possibly ruin a very expensive GPU at the same time, instead of having it RMA'd is beyond me.

Because no one will accept an RMA for choke coil noise. EVGA once did, but sadly..

 

[Brothanumsie]

That's just weird. The backplate shouldn't have anything to do with it...

I agree. Its not so much the backplate thats "fixing" or even causing it. Im fairly certain its just mounting pressing. And without the backplate, im able to apply just a little more and thus pushing those chokes a little tighter against the waterblock. So maybe the vibration isnt from the bottom of the chokes on the pcb, but instead maybe the coils are vibrating the ever love crap out of the top of the housings? Just throwing something out there since gluing the bottom seemed to have no effect in my situation. And since the whine seemed to be more in the lower frequencies than normal, this might just be an outlier.

 

[athenaesword]

This was the case with 3090 last gen too. Ekwb backplate needed longer screws to keep it looser against the card to reduce whine. I got the longer screws and it was quieter, tho not silent. There’s a Reddit thread on this.

 

[Brothanumsie]

This was the case with 3090 last gen too. Ekwb backplate needed longer screws to keep it looser against the card to reduce whine. I got the longer screws and it was quieter, tho not silent. There’s a Reddit thread on this.

Hmmm, so youre saying its actually the opposite of what im suggesting. The whine im hearing is actually the chokes vibrating slightly against the waterblock and that without the backplate, the fit is looser and not touching?

 

[ad1]

I have come up a plan for a less intrusive, but more expensive fix for my screaming cat 4090 Spurim X.

I will transplant it into a "be quiet!" Silent Base 601 solid padded case, with only 2 fans in for SSD and motherboard VRMs, everything else closed off. Cooling will be provided by a Mo-Ra3 360. Without RGB and fancy glass case to look at, I can move the PC under my desk or perhaps even further away. If this doesn't solve the problem, I will try to add more foam around the card, and if that doesn't work, I will throw the whole thing off a bridge.

I chose this particular case because it should let me mount my gpu waterblock horizontally, since it has a max CPU cooler height of 185mm (other cases are usually 155-165). Kind of a shame to hide the custom loop, but at this point, I don't really care how it looks anymore, I just want a card that will let me focus more on a game than the changing pitch of the buzzing as I move around.

 

[Wolverine2349]

It seems like the Super glue method does not work thus far to get rid of annoying whine/buzz from the inductors/coils?

Or is it only the cards that scream so badly that it helps.

Lex stated a Zotac and Palit still had some but were quieter and blended in well with case fans after super glue method. Those cards themselves especially Zotac have really loud fans. So I wonder if both the Palit and Zotac were Duds that had far worse screaming and whine than most others of same brands.

IU wonder if the reality is what someone considers bad whine is different to others.

I wonder if so called good 4090s that are deemed quiet really are just far less bad than others.

After try8ing so many Gigabyte and PNY cards, I can say there is no such thing as a good 4090 in terms of coil whine. They are just good compared to Asus and MSI cars, especially the bad samples of them and even good compared to the less bad samples.

Though there has to be some unicorns out there as some say they do build very silent PCs and their card is practically silent.

Such a shame as I replaced 4 140mm Be Quiet Silent Wings 3 1000RPM fans (they were pretty quiet though had a slight audible harmonic noise, but not at all bad). Though I researched fans and discovered Shadow Wings 2 were even quieter and the name was misleading. They were also newer released in 2019 as a opposed to Silent Wings 3 in 2016.

So I gave them a try and I am so impressed and they were so worth it. They are slightly quieter and more importantly they sound smoother and have modestly or maybe even significantly less harmonic noise.

So my PC while no whisper quiet, it is pleasant sounding and smooth and pretty quiet.

And these 4090s now have such big coolers with such large dies and so many have good fans the dream silent smooth sounding build that would be nearly inaudible across the room is ruined by a noise no one would have ever thought many years back. Electronic resonance/coil whine not even supposedly from physically moving parts.

I want that unicorn 4090 and my build is a dream quiet smooth sounding even though not completely silent build even when under intense load. Such a shame its ruined by noise that is not even what is associated with noise like fans or spindle HDDs. Even high TDP from VRM on motherboard barely causes any if at all coil whine when torturing CPUs with 250 watts to 300 watts of power.

 

[athenaesword]

Hmmm, so youre saying its actually the opposite of what im suggesting. The whine im hearing is actually the chokes vibrating slightly against the waterblock and that without the backplate, the fit is looser and not touching?

I thjnk we’re talking about the same thing. The backplate needs to be looser/off. That’s what happened with the 3090. When the screws are too short it somehow warped or applied pressure in the wrong way which resulted in noise. I’ve sold mine when I upgraded to the 4090 so I can’t test now. And this thread makes me abit sad that most people haven’t been able to make much out of this fix with the 4090. In fact there hasn’t been any success in weeks it seems like.

 

[mouacyk]

No success with my Suprim X Liquid either, used copius amount of glue and cured for 12hours+, with hair-dryer treatment at the end. Wicked all the inductors/chokes in yellow/red on front and the one on back. Starting to think it's a back-plate issue also. May have to try loosening it some time.

 

[lexluthermiester]

No success with my Suprim X Liquid either, used copius amount of glue and cured for 12hours+, with hair-dryer treatment at the end. Wicked all the inductors/chokes in yellow/red on front and the one on back. Starting to think it's a back-plate issue also. May have to try loosening it some time.

Or just remove it. There is also thin thermal pads you could place inbetween all contact points to act as a cushion..

Lex stated a Zotac and Palit still had some but were quieter and blended in well with case fans after super glue method. Those cards themselves especially Zotac have really loud fans. So I wonder if both the Palit and Zotac were Duds that had far worse screaming and whine than most others of same brands.

Fan noise was acceptable and not really louder than any other brands except Galax(which have crazy quiet fans). I was referring only to the audible high-pitched coil whine.

 

[ad1]

No success with my Suprim X Liquid either, used copius amount of glue and cured for 12hours+, with hair-dryer treatment at the end. Wicked all the inductors/chokes in yellow/red on front and the one on back. Starting to think it's a back-plate issue also. May have to try loosening it some time.

My Suprim X was identical with and without backplate, with original and waterblock backplate.

 

[Wolverine2349]

Or just remove it. There is also thin thermal pads you could place inbetween all contact points to act as a cushion..


Fan noise was acceptable and not really louder than any other brands except Galax(which have crazy quiet fans). I was referring only to the audible high-pitched coil whine.

Interesting. I had 2 Zotac AMP Airo 4090s and fan motors were horribly loud even at only 30% which it forces the card to use at minimum if you try and set a custom fan curve and cannot manually set fan any lower than 30% unless on auto and at idle. Any intense load after like 30 seconds to 1 minute or less fan spins up and it was horribly loud on both

Gigabyte, Asus, MSI, and PNY and even Founders fans at only 30 to 50% and sometimes even 60% were so much smoother and better sounding and usually very quiet. Well Gigabyte default fan BIOS is too aggressive and ramps fan speed up too fast and on high it gets loud, but manual fan curve sets it and the fan is fine on Gigabyte Gaming OC at manual fan curve and fan not running too fast and even ok at 50 to 60% though a little louder than MSI, Asus and PNY. The Windforce Gigabyte 4090 fan was kind of bad, though not like Zotac. It did have a slight scratching, but far more bearable. Though it did make a click noise every like 5 to 10 seconds.

I have never tried Palit or Galax, or Gainward or any of the brands sold more outside the United States.

I wonder is there really a fan lottery too on these GPUs?

 

[Brothanumsie]

Pretty ridiculous to be having these kinds of issues on cards so damn expensive

 

[Wolverine2349]

Pretty ridiculous to be having these kinds of issues on cards so damn expensive

Exactly.

The manufacturers should be embarrassed and ashamed that they cannot resolve coil whine and make a card as inaudible regarding coil whine/electrical buzz as motherboards are with high end CPUs overclocked under intense load which produce no audible motherboard coil whine in most well designed mobos.

The fact they get away with selling this crap to us is a shame especially given they give no alternative for so many of us that care to pay a premium for a binned card that is guaranteed to be whine/buzz free using a high quality power supply and components and mobo. Instead we are left to nothing but a stupid lottery which means buying and having to return cards to the point we could get flagged and banned form returns at a store. Or having to sell cards for a loss after they were opened to finally find a card. Lots of people built quiet and even silent systems these days and have for 10 years or more now. And you can build a very high end system with high end power hungry CPU with the water cooling solution these days especially if you do it right and still have it silent under full load or at least very quiet.

I am so ready to hear but we are enthusiasts and a small sliver of AIB and NVIDIA and AMD and other manufacturers customers. Well guess what. We enthusiasts are one of the only few who actually buy 4090 cards. So there is certainly a market for premium 4090 cards marketed and tested with rigorous QA process to have almost no to 0 coil whine/buzz under intense loads. A $500 or even $1000 premium would sell as I would have paid for such a card over 3 months ago and been done with this nightmare and had a quiet 4090. If they put effort into such large coolers and fans that are quiet to cool these power hungry cards with little noise there why the heck would they invest in that without testing for coil whine/electrical buzz???

Could using super glue actually damage something??

Accoridng to Igor's lab it can.

Power Supply in Graphics Cards explained - Switching Voltage Regulators, PWM Controler, true Phases and the annoying “Coil Whining” | Reminder | Page 4 | igor´sLAB

I wrote today’s article less than a year ago, so it’s not really “retro” yet, but somehow it’s time to bring it back up in this section on Saturday. I receive so many letters and enquiries on the…

www.igorslab.de

Though it mentions plastic plates between coils, and I do not think your method Lex has that?

Though is it actually bonding coils together or just sealing the inductor to PCB so no noise can leak out? And thus is what you describe on how to super glue different than what Igor is describing about bonding of coils with super glue?

 

[Shrek]

The manufacturers should be embarrassed and ashamed that they cannot resolve coil whine

I agree and can't see why it is so hard to fix; but what would I know?

 

[Wolverine2349]

I agree and can't see why it is so hard to fix; but what would I know?

Exactly and with all the effort and research and engineering that goes into making these video cards, coil whine should be a minor thing in QA testing. And it should be something they can fix easily compared to the other things. I do not care if it increases the costs. They already cost enough to begin with and its just lazy engineering the fact that GPUs have any audible coil whine under any condition without and I mean literally 1 inch your ear being next to the card.

 

[Frag_Maniac]

The manufacturers should be embarrassed and ashamed that they cannot resolve coil whine and make a card as inaudible regarding coil whine

I'm not convinced it's so much can't as won't do what it takes to, or at least do what it takes to minimize it considerably. Apparently they think we're going to accept high power graphics cards as if they were high powered sports cars, kicking out lots of volume when the revs crank up.

 

[Wolverine2349]

I'm not convinced it's so much can't as won't do what it takes to, or at least do what it takes to minimize it considerably. Apparently they think we're going to accept high power graphics cards as if they were high powered sports cars, kicking out lots of volume when the revs crank up.

Exactly and they should be ashamed that they won't do it., Charging us these insanely high prices for lazy engineering in what is a defective product with unwanted noise for silent and quiet PC builders (who by the way often only think about fans and HDDs and think electronics with non moving physical parts are silent until reality strikes hard with coil whine) and ripping us off to no end thinking it will go unnoticed and trying to rationalize it as normal when it is just lazy crap engineering!! Its quite frankly embarrassing and shameful and a class action lawsuit against them should come against them marketing their cards as quiet.

Just like how Intel had a one for the Pentium 4 Williamatte and AMD for the Bulldozer CPUs with false marketing claims that was settled over a decade later.

Intel Pentium 4 Class Action Settlement Pays Consumers $15 Each

How to claim $15 from the Intel Pentium 4 class action settlement.

topclassactions.com

AMD Bulldozer 'Core' Lawsuit: AMD Settles for $12.1m, Payouts for Some

www.anandtech.com

 

[lexluthermiester]

Well Gigabyte default fan BIOS is too aggressive and ramps fan speed up too fast and on high it gets loud

That isn't a bad thing. Keeping things cool is FAR more important than keeping things quiet.

I wonder is there really a fan lottery too on these GPUs?

This is true with any one can buy.

The manufacturers should be embarrassed and ashamed that they cannot resolve coil whine

I agree and can't see why it is so hard to fix; but what would I know?

Also agree. They really should be embarrassed. It's the 2020's for fraks sake. Coil whine should have been a thing of the past 25 years ago when the process for making chokes was greatly improved.

 

[Wolverine2349]

That isn't a bad thing. Keeping things cool is FAR more important than keeping things quiet.


This is true with any one can buy.

Yes but it does not need to do that to keep things cool enough. Other cards fans are much quieter and spin much less fast and stay within reasonable temps.

 

[Wolverine2349]

View attachment 255768

*Sploosh* on all sides and hope for the best

So basically on that type of inductor do not put it where the metal thing is between the red lines?

Also, sticking one's ear up against it can be reasonably definitive.

Today's Low Tech Solution(tm)

Yes in cases the noise coming clearly from the card. And I have tried different PSUs on past 4090 cards and it made no difference. I currently have an eVGA SuperNOVA 1600 T2 which is top tier for specs and ripple suppression. And I put my ear right next to card under load and noise clearly coming from the 4090 cards I have tried not the PSU when the cards were under intense load.

As before, the ones highlighted in red are the most likely culprits but the ones in yellow are possible too.
View attachment 285831
I would do them all, just to be safe.

For gluing the chokes to get the glue to wick in, do you want it applied around or on top of the metal lever looking things sticking out of them particularly visible on the R15 ones.

I am considering to try this on a Gigabyte Gaming OC 4090.

Like in my attachment modified what you made I circled top left 3 metal things sticking out of chokes in dark red. They are on each side. How do I do the super glue with those. Around them or just on top of them into the choke.

 

[lexluthermiester]

do you want it applied around or on top of the metal lever looking things sticking out of them particularly visible on the R15 ones.

No, just on the side of the housing. You won't hurt anything gluing on top of those metal contacts, but it isn't needed.