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

[Valantar]

I wonder how much thermal pads damp coil whine.

Essentially not at all. I guess it would be worse if the coil was vibrating against a bare aluminium heatsink, but the coil manages to produce plenty of noise on its own.

 

[lexluthermiester]

Will leave this to dry overnight and report back.

If you used super glue, it should be solid by now. 24hours curing time is for full solidification. Once it hardens it's structurally set.

Essentially not at all. I guess it would be worse if the coil was vibrating against a bare aluminium heatsink, but the coil manages to produce plenty of noise on its own.

This.

 

[Mussels]

I wonder how much thermal pads damp coil whine.

The noise is a high pitch echoing inside a sort of container over the coil

So this is the noisy bit

and they wear fancy hats

So a thermal pad only covers noise that goes "up", while the glue seeps in and seals the bottom, stopping the noise leaving out any gaps (and ideally, soaks up onto the coil a little reducing the vibration at the same time it traps it inside the fancy hat)

 

[Valantar]

The operation seems to have been a success. While there's still a slight audible buzz, I now have to bring my ear pretty close to the system (running open-bench right now) to properly hear it as it's essentially drowned out by the NH-A9i's fan. So it's gone from "clearly audible at all times at unacceptable noise levels" to "it's there, but you kind of have to listen for it". Definitely happy with that!

On a side note, the delid seems to have been a success as well. Haven't run much of a torture test, but the A9i heat soaks quickly, and after 10 minutes of OCCT (+Heaven to get some load on the GPU as well), my CPU core temps are sitting between 62-66°C, with the package temperature at 65-66°. Previously this test brought everything to 79-80°C quite quickly. Oh, and did I mention this is with a ~300rpm reduction in fan speed? Yep, that TIM was clearly pretty crap, and conductonaut is clearly quite nice (even if applying it is anything but fun).

 

[Bill_Bright]

Has anyone tried clear nail polish? I heard that works pretty good.

It has to be hot pink, not clear!

Seriously, I cannot speak for this coil application, but we have tried nail polish on a noisy transformer before. I do not recommend it. It did work, but the noise returned after about a week. Upon close inspection, we could see the polish became brittle and started to flake off.

To make matters worse, in order to implement a permanent solution (with epoxy putty), we had to use acetone (essentially nail polish remover without all the extra additives/contaminants). The problem there was the acetone started to breakdown the resins applied to the transformer plates at the factory, creating a gooey mess that was even harder to clean thoroughly.

That said, the epoxy putty worked great. We just squished it in between the plates like you would spackling a crack, let it cure over the weekend and we were good to go.

 

[Shrek]

The problem there was the acetone started to breakdown the resins applied to the transformer plates at the factory, creating a gooey mess that was even harder to clean thoroughly.

I'd just fork out for the real deal, if the plates touch eddy currents can build up.

 

[Valantar]

the noise returned after about a week. Upon close inspection, we could see the polish became brittle and started to flake off.

Sounds reasonable - it's being hammered by high frequency vibrations after all. And nail polish is notorious for flaking too. For anything to withstand those vibrations it would either need to be slightly flexible or very hard - like superglue. I would guess that insulating varnish sets similarly hard?

 

[Bill_Bright]

I'd just fork out for the real deal, if the plates touch eddy currents can build up.

Not likely. Remember, typically transformer plates are stacked on top (or side by side) of each other at the factory. This is different from a coil where the windings are spaced some distance apart.

But regardless, as I noted above, squishing epoxy putty between the plate would prevent any unwanted contact.

or very hard - like superglue.

Except superglue has low ductility - becomes brittle too.

If you want to use superglue, I recommend one of the newer, 2-part hybrid epoxy superglues that provide the best of both worlds. See this and scroll down to "Are there two-part cyanoacrylates?"

 

[Valantar]

Except superglue has low ductility - becomes brittle too.

If you want to use superglue, I recommend one of the newer, 2-part hybrid epoxy superglues that provide the best of both worlds. See this and scroll down to "Are there two-part cyanoacrylates?"

True, but it cures much harder than nail polish, and at least from what I've seen has much less of a tendency for cracks to spread, meaning it chips rather than flaking, and damage is less likely to spread catastrophically, at least over relatively short time spans.

 

[lexluthermiester]

But regardless, as I noted above, squishing epoxy putty between the plate would prevent any unwanted contact.

Perhaps, but it's also VERY messy and cumbersome to apply. Super Glue is not. It is easily applied and mostly mess free when using a nozzle applicator. Furthermore, epoxy does not wick into small spaces and crevices, thus failing to fill spaces that Super Glue can do natuarlly.

Except superglue has low ductility - becomes brittle too.

That argument is flawed in both it's presumption and it's presentation. Videocards do NOT undergo flexing frequently enough nor to a degree that ductility would be any great factor with superglue that would not also be experienced by epoxy. And while it is known that superglue does become a little bit more brittle after curing, it's not enough to be considered fragile nor hinder it's application in any manner in the context of this use-case-scenario. Also...

True, but it cures much harder than nail polish

...this

 

[Bill_Bright]

True, but it cures much harder than nail polish,

This is true. No argument there. Frankly, I am not a fan of using nail polish. But I note so does epoxy cure harder than nail polish. If I had nothing else, I would use nail polish. If super glue was handy, then for sure that. But for a truly permanent fix, epoxy - or the hybrid superglue epoxy mentioned earlier.

Perhaps, but it's also VERY messy and cumbersome to apply.

Not really. You only need a pea size blob of each part, a wooden Popsicle/glue stick, and a 3x5 index card to mix the two on. Hardly messy or cumbersome - with no danger of supergluing your fingers together, or your fingers to your eyelid.

But I will admit that superglue, with its handy little dispenser bottle, is less messy and easier to apply - if it has not dried up already in the previously opened tube - a MAJOR complaint of mine and others.

Furthermore, epoxy does not wick into small spaces and crevices

Which is exactly why I said you need to squish it in between the plates. So in that respect, it is messy for your thumb - but if you have a box of surgical gloves handy, or a small piece of plastic wrap - that problem is easily solved too.

I will give you that superglue looks nicer - if that is important to you.

That argument is flawed in both it's presumption and it's presentation.

There is no presumption. Look it up. Superglue becomes brittle and can easily break free when stressed by "shearing" and "twisting" forces - that is, any force that is not 90° (perpendicular) from the bond. Epoxy does not. Some may remember the old Crazy Glue commercials where a construction worker hung from an I-beam girder by his hardhat that was superglued to the girder. That demonstrates superglue's superior bonding characteristic as the force of gravity is straight down - 90° - perpendicular to the bonded surfaces.

And while it is known that superglue does become more brittle after curing, it's not enough to be considered fragile nor hinder it's application

First, thanks for admitting - after first stating my argument was flawed! - that it does indeed becomes brittle. But note I NEVER said or implied it was either fragile or a hinderance. What I am saying, and still contend is if you want greater assurance of a more permanent bond, epoxy putty is the better, solution.

 

[Mussels]

You can argue for alternatives - but for those of us that have TRIED most of them, superglue genuinely seems to be the best real-world choice

Hot glue didnt get far enough inside to help
Nail polish flakes off too fast

Superglue sucks itself into where its wanted and dries fast, and i've seen no flaking or cracking in my little SFF server after a few months of abuse

 

[Valantar]

Which is exactly why I said you need to squish it in between the plates. So in that respect, it is messy for your thumb - but if you have a box of surgical gloves handy, or a small piece of plastic wrap - that problem is easily solved too.

I don't know what you mean by 'plates' here - the rest of us are talking about inductors on motherboards and GPUs, typically of the encased kind. The noise mostly comes from the coil vibrating inside of the casing. There is no physical way of "squishing" anything in between there unless you desolder the inductor - and typically not even then. That's why you need a low viscosity fluid that can move up inside of the inductor through capillary action. Epoxy putty is the exact opposite of what is needed here. It's a decent insulator, and cures very strong, but at best you'll dampen external vibrations (between the casing and PCB) and possibly add some noise isolation for the internal ones, but you won't actually fix the problem that way.

I will give you that superglue looks nicer - if that is important to you.
There is no presumption. Look it up. Superglue becomes brittle and can easily break free when stressed by "shearing" and "twisting" forces - that is, any force that is not 90° (perpendicular) from the bond. Epoxy does not. Some may remember the old Crazy Glue commercials where a construction worker hung from an I-beam girder by his hardhat that was superglued to the girder. That demonstrates superglue's superior bonding characteristic as the force of gravity is straight down - 90° - perpendicular to the bonded surfaces.

While that is true, I don't see the relevance here? There will be no major shearing or twisting forces applied to these inductors. Sure, the vibrations will cause movement in all kinds of directions, but the forces involved are tiny and chaotic. Can this cause cracking over time? Possibly, sure. I'd be more than willing to take my chances. As discussed above, cured superglue generally doesn't see cracks propagate and start flaking (unless the forces applied are quite extreme). So it's possible the vibrations will cause some cracking over time, but it will be gradual and very small.

 

[izy]

There should be some downside to this as im thinking that the OEMs would have fixed this by now. I think that superglue may damage the PCB and i am not sure if that coil gets very hot but that can be a problem too if its melting the glue.

 

[Valantar]

There should be some downside to this as im thinking that the OEMs would have fixed this by now. I think that superglue may damage the PCB and i am not sure if that coil gets very hot but that can be a problem too if its melting the glue.

It's likely more a problem of the levels of QC/research/manufacturing complexity needed to eliminate whine. It could probably be solved if inductor manufacturers did extensive research on a potting material that eliminated vibrations between the coil and casing, if they did extensive QC testing for whine across a wide range of switching frequencies and noise/ripple levels, or if they added an additional manufacturing step of adding a low viscosity potting material (which is what the superglue is acting as) after the one currently used in order to entirely eliminate air gaps and room for movement. The problem is that either of these approaches (or any combination of them) is going to be expensive, and inductors are a mass produced low cost, low complexity component. Most likely this would drive prices up far above competitors. And, let's face it, most users in most scenarios neither notice nor care about coil whine, and it is only really noticeable and annoying in rather extreme cases (like cpu and gpu VRMs).

Superglue will not damage the PCB unless it contains acetone (none that I know of do) or other very powerful solvents not typically found in cyanoacrylate glue, and won't melt within the operating temperature range of most VRMs.

 

[izy]

It's likely more a problem of the levels of QC/research/manufacturing complexity needed to eliminate whine. It could probably be solved if inductor manufacturers did extensive research on a potting material that eliminated vibrations between the coil and casing, if they did extensive QC testing for whine across a wide range of switching frequencies and noise/ripple levels, or if they added an additional manufacturing step of adding a low viscosity potting material (which is what the superglue is acting as) after the one currently used in order to entirely eliminate air gaps and room for movement. The problem is that either of these approaches (or any combination of them) is going to be expensive, and inductors are a mass produced low cost, low complexity component. Most likely this would drive prices up far above competitors. And, let's face it, most users in most scenarios neither notice nor care about coil whine, and it is only really noticeable and annoying in rather extreme cases (like cpu and gpu VRMs).

Superglue will not damage the PCB unless it contains acetone (none that I know of do) or other very powerful solvents not typically found in cyanoacrylate glue, and won't melt within the operating temperature range of most VRMs.

If i remember correctly some years ago i tried to fix a broken amplifier and the superglue i used affected the PCB ( i didnt had a solder gun at the place i was) but i dont remember if i cleaned it with something that was not ok before gluing or not.

There should be like cheap and quality ones and in a premium GPU , lets say RTX 3090, they can afford using the more expensive one. You can find other components with different levels of quality why not this too.

 

[Valantar]

If i remember correctly some years ago i tried to fix a broken amplifier and the superglue i used affected the PCB ( i didnt had a solder gun at the place i was) but i dont remember if i cleaned it with something that was not ok before gluing or not.

Sounds weird. I guess it's possible that it used some sort of low-grade PCB that was sensitive to weaker solvents, but even that sounds unlikely IMO. PC components definitely don't. Or, of course, it's possible that wasn't CA glue but something else.

There should be like cheap and quality ones and in a premium GPU , lets say RTX 3090, they can afford using the more expensive one. You can find other components with different levels of quality why not this too.

There obviously are, but I think you're underestimating the cost difference of additional R&D or manufacturing steps like this. One thing is designing to a higher spec, another is producing something with sufficient precision to eliminate an elusive and hard-to-remove issue - and the latter will be more expensive in the vast majority of cases.

And this is made all the more complicated by coil whine being dependent on factors outside of the manufacturer's control - coil whine is typically dependent on (among other things) the specific PSU used, the GPU or CPU used and their clock speed (and its stability/variability), the current draw (and thus the width of the VRM), the amount of filtering capacitors both on the 12V input and between the inductor and the CPU/GPU, and even soldering quality - and a lot more. The exact same system otherwise might see no coil whine with one PSU, and unbearable whine with another. The exact same inductors might whine like crazy installed on one GPU, while they're dead silent on another with similar on-paper specs (same die, power draw, clock speed) but different board layouts, filtering, and BIOS tuning. And so on, and so on.

 

[izy]

Sounds weird. I guess it's possible that it used some sort of low-grade PCB that was sensitive to weaker solvents, but even that sounds unlikely IMO. PC components definitely don't. Or, of course, it's possible that wasn't CA glue but something else.

There obviously are, but I think you're underestimating the cost difference of additional R&D or manufacturing steps like this. One thing is designing to a higher spec, another is producing something with sufficient precision to eliminate an elusive and hard-to-remove issue - and the latter will be more expensive in the vast majority of cases.

And this is made all the more complicated by coil whine being dependent on factors outside of the manufacturer's control - coil whine is typically dependent on (among other things) the specific PSU used, the GPU or CPU used and their clock speed (and its stability/variability), the current draw (and thus the width of the VRM), the amount of filtering capacitors both on the 12V input and between the inductor and the CPU/GPU, and even soldering quality - and a lot more. The exact same system otherwise might see no coil whine with one PSU, and unbearable whine with another. The exact same inductors might whine like crazy installed on one GPU, while they're dead silent on another with similar on-paper specs (same die, power draw, clock speed) but different board layouts, filtering, and BIOS tuning. And so on, and so on.

Well.. i am a bit ironic here but why dont they just factory glue them , just in case of coil wine happens, as its cheap and as you guys are saying, effective

 

[Valantar]

Well.. i am a bit ironic here but why dont they just factory glue them , just in case of coil wine happens, as its cheap and as you guys are saying, effective

One possible reason: soldering temperature tolerance: adding CA glue after a component is mounted is no problem, but adding it beforehand is likely not possible due to the heat of SMD component mounting temperatures with unleaded solder (which is required for ROHS compliance). At that point that glue would likely see temperatures way higher than it could tolerate, causing it to melt, deform, or even catch fire. And there would of course be some unpleasant fumes involved either way.

Another reason might be production cost as I mentioned above: if this adds another step to production, and particularly if that step requires waiting for something to set or cure beforehand, then that's a very significant additional cost to production, requiring production line expansions, new tools, and possibly temporary storage space for curing components.

 

[Shrek]

One possible reason: soldering temperature tolerance: adding CA glue after a component is mounted is no problem, but adding it beforehand is likely not possible due to the heat of SMD component mounting temperatures with unleaded solder (which is required for ROHS compliance). At that point that glue would likely see temperatures way higher than it could tolerate, causing it to melt, deform, or even catch fire. And there would of course be some unpleasant fumes involved either way.

They would not use super-glue

 

[Valantar]

They would not use super-glue

True, but that varnish is still just rated up to 180°C, nowhere close to enough. And it dries in ~48h at room temperature (for a thin surface coating?) or 1h at 80°C, which is rather impractical. The datasheet doesn't say anything about curing for using it in voids/in between parts, which might be even longer. And no word on shrinkage. That's the struggle here: finding a potting compound for the inductors that's non-toxic (or close to it), doesn't shrink noticeably when cured, is sufficiently hard and strong, can take ~300°C, and sets quickly, ideally at room temperature. And of course has good insulating properties. That sounds pretty much like a miracle material - while adding superglue after the fact counteracts a lot of the problems with finding such a material. Or, of course, varnish like that, which would probably work just as well - I can't say as I have zero experience with it.

 

[Shrek]

Electrolytic capacitors are good to 105°C but are fine for soldering; the 180°C is service temperature.

One could use epoxy; the boards are made of epoxy so there should be no temperature issue.

 

[Valantar]

Electrolytic capacitors are good to 105°C but are fine for soldering; the 180°C is service temperature.

True, but service temperature for a capacitor is the temperature it can operate under for X thousand hours before failing, while the "operation" of a varnish is... well, covering something while not conducting electricity. I highly doubt there's any major difference between operating and installation temperature for that varnish, unlike most electronic components. That isn't to say it'll immediately break down or catch fire at 300°C obviously, but if it could handle that, it seems reasonable to expect that it would be rated for it IMO.

One could use epoxy; the boards are made of epoxy so there should be no temperature issue.

Definitely a good candidate, and frankly I expect the potting compound they already use to be some form of epoxy. The problem with epoxy is typically shrinkage though, which is difficult to avoid, and inevitably leaves gaps in a use case like this.

 

[Shrek]

Bob Smith Industries BSI-205 Clear Slow-Cure Epoxy (4.5 oz. Combined): Amazon.com: Industrial & Scientific
This has an upper temperature rating of 180°C which happens to be the same as for the insulating varnish.

Or maybe a low viscosity epoxy
FDA-Bond 2 Food & Drug Administration Medical Grade Epoxy Adhesive, Low Viscosity RT Cure: Amazon.com: Industrial & Scientific

I'd just go with the varnish as no mixing is needed (I just made a repair with epoxy today, and, in this heat, things cured way too fast).

It also reminds me of a tour I did some time back of a hydroelectric power station where I was told they used to use epoxy for the windings but now use tar; not sure they believed me when I told them they may have got things the wrong way round.

 

[Mussels]

There should be some downside to this as im thinking that the OEMs would have fixed this by now. I think that superglue may damage the PCB and i am not sure if that coil gets very hot but that can be a problem too if its melting the glue.

superglue can trap heat, and long-term i dont expect it to last 5-10 years without flaking or cracking, so it's no good as factory option

Thats why we arent dumping it over heatsinks or high-heat components, only on specifically noisy coils
The ceramic covers *is* their attempted fix, that mostly works

Its a nightmare to test at the factory, my 750W corsair had no coil whine, but it caused coil whine on an aorus 3080
Then a reference 3090 had none but the PSU did

Now with a new PSU and no coil whine, I cant trigger the whine with that older corsair PSU with any other hardware in the house



Oh and i did try epoxy, but its so thick it's difficult to get it where it matters - superglue has a wicking effect and gets sucked in, which is why it's so simple and easy to use on the covered coils