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

[lexluthermiester]

Just re-registering to show my results with ASUS TUF OC 7900XT.

This glue is not perfectly fluid it is slightly thick and stringy. But I applied a heck of a lot.

View attachment 297286View attachment 297287

Reassembled and up and running, glue is not fully dried but I gave it 30 mins, put Borderlands 3 no Vsync on 370 FPS on the menu, the coil whine still exists but massively reduced, once it dries it will likely be almost non existent.

So a positive change.

Sweet! Out of curiosity, was the high pitched sound reduced, the low pitched or was it a general reduction in volume?

I ask because some cards have been showing the high pitched fall off but low pitched remain more, other cards vice-verse. Most see a general reduction in volume.

 

[RDNAuser]

Sweet! Out of curiosity, was the high pitched sound reduced, the low pitched or was it a general reduction in volume?

I ask because some cards have been showing the high pitched fall off but low pitched remain more, other cards vice-verse. Most see a general reduction in volume.

Just quieter overall.

 

[Mussels]

I hope that's thermal paste residue and not glue around the socket, as those should not be making any noise at all (and can get really hot)


Next time you feel like ripping it apart, try copper shims instead of thermal pads? Just get the right thickness and use thermal paste with them
I use them instead of the thermal pads on my asus motherboards chipsets because asus didn't bother with heatsinks the correct size and re-used intel ones with super thick pads, and it took a good 15c off.
This is a photo from someone using them on VRAM, but they can also be used on VRM's as well - they absorb heat, and they're solid so they can absorb vibration and noise too

I've been pondering carefully cut and trimmed copper sheets, combined with thermal pads/paste

They'd absorb heat, vibration and noise, and EMI/RFI would become 'shared' between all the components under it, preventing it spreading to other nearby components at the same time it controls those under it
(See the previous page with the toroid vs the cubes magnetic fields, even something as simple as rounding the corners changes it massively)

I've used adhesive copper tape on a ton of devices around here, but on lower wattage devices only - the adhesive they use needs to be able to handle the heat, or you're better off with the plain copper sheet and a slightly thinner than stock thermal pad on top to hold it in place

Eco-Fused Adhesive Copper Foil Tape - Double-Sided Conductive - EMI, Rf Shielding, Paper Circuits, Electrical Repairs, Grounding, 1 Roll - Copper Adhesive 0.5 inch / 12.7 mm
The danger ofc is conductivity - but we have many methods to keep it safe regardless.

I love that the secondary use is as slug repellant? Imagine fixing your GPU coil whine with "slug repellant"




Edit: Oh on my previous musings about grounding out ferrite beads, the key is to make sure any shielding you use is only grounded at one end. You could for example use a simple ass loop of plain copper wire, wrap it around your cable a bunch of times and have the far end connected to the PC case - but only one part should be connected, not both. That earths out the EMI into the case itself, but without an open circuit by being connected at both ends it can't mess with anything other than EMI/RFI.
I'm definitely using the wrong search terms to get the answers i'm looking for on this one


This is just the materials i had on hand at the time, use the power of imagination.
It's from a solar panel and a pipe cleaner, seemed close enough to get the idea through.

PCI-E cables with a copper wire (stolen from a cat6 cable, is my idea) wrapped around as many times as possible before one end goes off to be earthed somewhere on the case.
This evens out that EMI "AC" voltage differential between the power cables, but it's not an open circuit so it doesn't allow power to flow through it for any reason

I'm unsure if this could potentially build up any heat - from what i've read with only one end connected it should not. Anyone who tests this, should be careful that the end around the cables is covered completely, and check for heat buildup if it has a lot more EMI to deal with than we expect.
(The less time's it's wrapped around, the less it absorbs - start small and work up with more turns?)

 

[Fishtank23]

Greetings. I have a problem with crackling (very similar to the sound of a hard drive) and squeaking in the socket area. When buying a motherboard (used), there were no sounds.

Motherboard Gigabyte B550 Aorus PRO V2
Processor AMD Ryzen 7 5800X
DDR4-3733 memory with forced installation of Infinity fabric at 1866
The sound does not depend on the power supply, I tried three different ones. Now Seasonic pime titanium 750
I tried different video cards, depending on this, the squeak in the socket area either appears or disappears in games. Now Gigabyte RTX 3070 Ti Gaming OC with it, the squeak is the strongest in 3d.

I decided to superglue my motherboard inductors as suggested here. Please give advice on how best to do this for the motherboard? Is it more profitable to remove the heatsinks from the VRM or the cooler (BeQuiet Dark Rock Pro 4) from the processor? Do I need to somehow protect the VRM transistors from getting superglue on them, because if you do not remove the VRM heatsinks, they can stick, and if you remove them, the glue can prevent you from putting them back. In any case, when the VRM is heated, the adhesive may stink. Although my motherboard model is cold, only the processor is hot, its temperature in AIDA64 is up to 89 degrees.

Should the adhesive be poured from the socket side or from the VRM side or between the coils? Also does anyone know how thick the thermal pad of this motherboard is so I can buy it in advance?

 

[Yougotonejob]

thermal pad can be reuse if not too much damage.
I would suggest identify which part is making sound. Then use a toilet paper roll to check which coil making sound.
btw watch this might help you.

Edit : a few month ago I accidentally take off one of the coil they are all sealed on my MB I assume yours is not much different

 

[Fishtank23]

Replacing the coil is a very time-consuming operation that requires tools. Fixing it with superglue is much easier. I plan to apply superglue today, it should dry by tomorrow. Perhaps I'll try to protect the VRM with cling film.

 

[Shrek]

If manufacturers were to take a similar approach, what would they use given that superglue might not be safe in the case of a fire due to the fumes?

 

[P4-630]

How to fix coilwhine

 

[Shrek]

I do belive I can hear the owner whining...

 

[lexluthermiester]

If manufacturers were to take a similar approach, what would they use given that superglue might not be safe in the case of a fire due to the fumes?

That's a good question! It's greatly easier to apply a sealant during manufacturing than after, but it's also an expensive process which is why they don't do it.

 

[Mussels]

Greetings. I have a problem with crackling (very similar to the sound of a hard drive) and squeaking in the socket area. When buying a motherboard (used), there were no sounds.

Motherboard Gigabyte B550 Aorus PRO V2
Processor AMD Ryzen 7 5800X
DDR4-3733 memory with forced installation of Infinity fabric at 1866
The sound does not depend on the power supply, I tried three different ones. Now Seasonic pime titanium 750
I tried different video cards, depending on this, the squeak in the socket area either appears or disappears in games. Now Gigabyte RTX 3070 Ti Gaming OC with it, the squeak is the strongest in 3d.

I decided to superglue my motherboard inductors as suggested here. Please give advice on how best to do this for the motherboard? Is it more profitable to remove the heatsinks from the VRM or the cooler (BeQuiet Dark Rock Pro 4) from the processor? Do I need to somehow protect the VRM transistors from getting superglue on them, because if you do not remove the VRM heatsinks, they can stick, and if you remove them, the glue can prevent you from putting them back. In any case, when the VRM is heated, the adhesive may stink. Although my motherboard model is cold, only the processor is hot, its temperature in AIDA64 is up to 89 degrees.

Should the adhesive be poured from the socket side or from the VRM side or between the coils? Also does anyone know how thick the thermal pad of this motherboard is so I can buy it in advance?

You do not glue the top of these devices - you're glueing the gap between them and the motherboard so it soaks inside them.
I don't think you should be gluing that board, see the PSU Comments below.

An important question here is, what PSU's did you actually use? You partially named one - and that's an old PSU from 2016 - no wonder it has issues with a modern setup, it's going to have aged and it's likely not designed for modern hardware with such low idle wattages and such fast transitions between load and idle.

You can possibly fix your low idle wattage squealing by disabling C-states in the BIOS and adjusting the PSU idle control setting

 

[Fishtank23]

Thanks, but I don't think the problem is with the power supply. The power supply is ATX 12v2.4, and the motherboard was created when ATX 12v3.0 was not yet created. On the Fractal design ion + 2 power supply with ATX 12v2.52, the same problem.

I glued the induction coils and I didn't use cling film because I figured out in time that it would cause the superglue to spread all over the VRM. However, due to a moment of inexperience, some superglue did get on the VRM chips. What can be done to clean them now?

 

[Yougotonejob]

i guess do nothing? VRM MOSFET are about 26 W power loss at most.
26 W heat is nothing to worry about won't make your glue to go smoke
----------------------------------------------
Edit here 5800x is about 100W + but sure less then 200W.
and CPU voltage is about 1.3 so your Amp is about 200A at max and worst case.
these Mos are usually 10% power loss(you need to know datasheet of the mos to know exact number but 10% is average for all these kind of Mos).
So you can think about 20W heat is generated on worst case. Your mos are not empty and naked they got heat sink too.
So temp wise nothing to worry about no fire hazard here. If the worry is from other part, i have no comment.

 

[aanti]

Hi TPU

My MSI 4090 Gaming Trio is a great card. Have it in a custom loop, but the coil whine is quite annoying.
Just upgraded motherboard, cpu and ram, am super happy with the performace of the rig, but the CW kind of ruins the experience.
I was thinking about this mod for quite some time and decided to give it a try.

Just would like to have a confirmation if this kind of superglue should be appropriate:

super glue 351

www.danalim.com

Also, could be any other parts than the chokes be relevant for superglue application?
Here are links to the front and back of the PCB
front
back

Would appreciate if someone could quickly mark the relevant parts

 

[Yougotonejob]

The green box with LR22 is a choke that matters too.

 

[Mussels]

Thanks, but I don't think the problem is with the power supply. The power supply is ATX 12v2.4, and the motherboard was created when ATX 12v3.0 was not yet created. On the Fractal design ion + 2 power supply with ATX 12v2.52, the same problem.

I glued the induction coils and I didn't use cling film because I figured out in time that it would cause the superglue to spread all over the VRM. However, due to a moment of inexperience, some superglue did get on the VRM chips. What can be done to clean them now?

Assumptions is what will trip you up and stop you finding a fix.
You can think it unlikely, but it's a combination of the PSU's and hardware - they don't exist in a vacuum.
Definitely do your testing based on likelihood of success mixed with how easy it is to test, but never rule something out entirely based on a feeling or assumption.

I run a Fractal Ion2+ myself, they're amazing PSUs - but they cant fix a problem where your motherboard/CPU/GPU are causing the noise - the BIOS settings can possibly help, otherwise it's physical modding time with superglue, ferrite beads, or shielding.

JonnyGuru made that quote about how they changed their PSU's to suit Nvidia 20 series at the cost of coil whine on AMD vega and Nvidia 10 series, and that nothing would solve it other than a different PSU or a different GPU (with the ferrite beads being one exception he suggested)

You can use acetone on cotton balls to wipe superglue off the external surfaces but go easy on the stuff and dont leave any behind.

 

[Simone]

Steeve the glyphid

.. I've been dying to know what that was, but just didn't have it in me to ask.. >xp

 

[Mussels]

.. I've been dying to know what that was, but just didn't have it in me to ask.. >xp

Beast Master - Deep Rock Galactic Wiki (fandom.com)
I play with the googly eyes mod, and he lives on my mic boom.

A good simplified way to explain the noise issues is how PWM fans work, or a modified sine wave vs simulated
The power is pulsed, and those pulses are timed. Having too much amps arrive early or late results in the voltages rippling up or down, as the phases turn on/off to adjust for the changes in power demand.
remember that they're pulsing it to match changes in demand.

Depending on where in the setup you're at, power could be coming in or out as either side of this image - being converted from one the other.
multiple phases are used so the square setups compensate for each other, so the steps are smaller - but they can go out of phase (hah) with each other too, if the incoming voltages are different between them.
You'll notice it's the current that varies here, not the voltage.

They use multiple phases and try to time them to provide as smooth a curve as possible, but when power consumption (current/amps) goes above or below the happy range for the components or voltages are higher or lower than expected - you get noise like a squeaking door as they're forcing themselves to shut against a strong flowing current

Actually, closing a door and having air whistling past is also a great example of how to explain this.
Low voltage with a higher current, the door/gates have trouble opening, while high voltage and low current you get sqeal as some of the current sneaks past after its meant to be closed.

The timings being off results in the square looking output, and voltage ripples. Capacitors can smooth it out again, but only so much.

On the CPU and mobo side, definitely look into altering how the system idles - even something as simple as the high performance power plan in windows could have an effect there, as it definitely did back on the x58-UD3 gigabyte boards back in the day where low idle wattages resulted in single core CPU usage being audible over the onboard audio as you moved a USB mouse around.

For GPU's we're a bit more limited, as we can only try changing cabling to reduce resistance (smoother/higher voltage inputs), or altering clock speeds and voltages up or down to get them out of whatever unhappy state they're in, without physical modifications

 

[Yougotonejob]

Cable mod is done, bought the cable mentioned at #655 and #662.
Bought 5 of them. plan to use 2 on my rx580 and 1 on mother board 8 pin and 2 for addition mod.
And after that I replace the default capacitor with RR71C331MDN1
Did some test, the sound is lower but not the level like night and day. I couldn't feel a difference the sound still there. But I record the sound, and watch the graph is lower for sure. I guess it's milage will vary depends on your system.
Not the perfect way to deal and not cost effective, but this works in some way.
Hope this helps, I will keep figure more ways.

Before

https://imgur.com/oPTJ2HN

-----------------------------------------
After

https://imgur.com/a/U5T3rST

The p to p value is lower and average is also lower.

 

[Mussels]

So while it's small, you definitely noticed a change from the capacitor cables?

 

[Yougotonejob]

So while it's small, you definitely noticed a change from the capacitor cables?

Hard to say, my ear is not "calibrated" I can't confirm the change, I can feel it's there or not but how may DB I can't tell.
BTW small change forget to mention is when loading CS GO the logo period, before the change have noise already, after change logo period not audible at all. That's a huge improvement.

The clips are there feel free to check with headsets, I wish you can find difference in audible feelings. (Sorry to my ear I can't)
Or you download the clips and analysis in software like I did.
Both clip are in CS go 180 - 200 FPS.
I am quite sensitive about the noise, any level is bad to me, I want it gone 100%.

 

[Shrek]

Beast Master - Deep Rock Galactic Wiki (fandom.com)
I play with the googly eyes mod, and he lives on my mic boom.
View attachment 298963


A good simplified way to explain the noise issues is how PWM fans work, or a modified sine wave vs simulated
The power is pulsed, and those pulses are timed. Having too much amps arrive early or late results in the voltages rippling up or down, as the phases turn on/off to adjust for the changes in power demand.
remember that they're pulsing it to match changes in demand.

View attachment 298964

Depending on where in the setup you're at, power could be coming in or out as either side of this image - being converted from one the other.
multiple phases are used so the square setups compensate for each other, so the steps are smaller - but they can go out of phase (hah) with each other too, if the incoming voltages are different between them.
You'll notice it's the current that varies here, not the voltage.

They use multiple phases and try to time them to provide as smooth a curve as possible, but when power consumption (current/amps) goes above or below the happy range for the components or voltages are higher or lower than expected - you get noise like a squeaking door as they're forcing themselves to shut against a strong flowing current

Actually, closing a door and having air whistling past is also a great example of how to explain this.
Low voltage with a higher current, the door/gates have trouble opening, while high voltage and low current you get sqeal as some of the current sneaks past after its meant to be closed.

The timings being off results in the square looking output, and voltage ripples. Capacitors can smooth it out again, but only so much.

On the CPU and mobo side, definitely look into altering how the system idles - even something as simple as the high performance power plan in windows could have an effect there, as it definitely did back on the x58-UD3 gigabyte boards back in the day where low idle wattages resulted in single core CPU usage being audible over the onboard audio as you moved a USB mouse around.

For GPU's we're a bit more limited, as we can only try changing cabling to reduce resistance (smoother/higher voltage inputs), or altering clock speeds and voltages up or down to get them out of whatever unhappy state they're in, without physical modifications

What frequencies are we talking about here? as those outside the audible may be of less issue; modulation from the video card may be the problem.

So while it's small, you definitely noticed a change from the capacitor cables?

Electrolytic capacitors are not very good at high frequencies; that is why they are often accompanied by smaller value by-pass capacitors.

 

[Mussels]

What frequencies are we talking about here?

That's not something i can answer - I've been learning as we go in this thread, and much of the information comes as second-hand snippets from more experienced people who don't give full explanations.
(Jonny gurus self-deleted posts, YT videos of people replacing capacitors on GPU's without explaining why they used the ones they did and accidentally fixing their whine, etc)
I aint graduated from homeschool e-lec-trici-tah 101 yet, a lot of what i've put here has been guesswork that i've googled to find out how accurate the guesses were and slowly we got a few hits.

Some of what i wrote recently has been after using a modified sine wave inverter and hearing the fun noises from devices that didn't like the way they produce the power - they really had a lot in common with the coil whine we get, at a lower speed.
Same with over-using a 24v to 12v step down converter, once it reaches its amperage limit it whines, and my solar MPPT controller develops a high speed clicking when it rapidly goes on/off as the sun sets, where it's voltage and the battery voltage meet. JG mentioned that as a problem in his quotes as something people mistake for whine because it's rapid, but the tone is totally different.
(The clicking is from when voltages drop too low, specifically)

I'm pretty sure a big part of the noise is the common, the ground. Anything that keeps the voltages even on all the rails and the grounds even, prevents a lot of noise.
When a design has paired up components, they're not expecting any variances in power - but it can happen when they've got two sets of power cables with 3-4 pairs of power+grounds, with potentially different resistance due to cable damage, bends, etc.
I assume those capacitors help that in some way, by smoothing out some spikes in the power, and keeping things even between whatever rails they're bridging.

 

[Shrek]

Me too, the learning part.

 

[Yougotonejob]

Steve Burke aka Gamer's Nexus Steve.

PowerColor Factory Tour

We visit the brand-new factory of PowerColor in Taiwan. Here the latest and greatest AMD Radeon graphics cards are built, like the Radeon RX 7900 XTX and RX 7600. We'll take you through every production step and also show you how these cards are tested before they are shipped off. Show full review

www.techpowerup.com

der8auer visited power color factory too.
we need Steve complain about coil whine. If you can email him this thread that would be great.
also confirmed from both TPU and der8auer's visit, there is 0 care about NVH in testing.
and someone mention about the coil whine in first link too, with a factory setup they can't hear it I guess.