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

[Count von Schwalbe]

Ferrite chokes can not do harm

I meant if your incoming voltage was low already.

 

[lexluthermiester]

I meant if your incoming voltage was low already.

Still doesn't work that way. Ferrite chokes will do nothing to a brown-out situation, good or bad.

 

[Count von Schwalbe]

Still doesn't work that way. Ferrite chokes will do nothing to a brown-out situation, good or bad.

Well, mebbe the ferrite ones work differently.

I can assure you that the chokes I am more familiar with would cause voltage drop if applied to an AC wire.

Which could lead to issues if your voltage is at the low end of what is OK.

 

[lexluthermiester]

Well, mebbe the ferrite ones work differently.

I can assure you that the chokes I am more familiar with would cause voltage drop if applied to an AC wire.

Then you're correct in that you're not talking about ferrite chokes.

 

[Mussels]

I can't say really. Can't replicate the same environment, as I changed all the fans, and did other stuff to the system, and I'm lazy to do the paper cone thing. But I'll update when I do.

I meant to put them on a psu-gpu cable, as Mussels mentioned a while back, but a few on the socket-psu part wouldn't do any harm neither.

I meant the power cable coming into the PSU.

Both!
On the PSU's power cable can remove electrical noise sources before the PSU turns it into DC power, the one before the GPU removes radio interfere/EMI before the GPU turns it into lower voltage DC power
It's simply that straight wiring works as an antenna, and can pick up or transmit noise.

Spoiler: DSL stuff, how i learned all this

The original installer had put the cables in un-evenly. Really bad job terminating by hand with a screwdriver because he didnt bring tools, and he'd dropped the backplate inside the wall cavity.
Because the cable length wasn't identical the two directions had uneven electrical resistances that had my speeds down to 55Mb.


One ferrite choke on my DSL cable at the modem gained me 3Mb sync boosted that to a total 5Mb gain. Both helped alone, but removing that source of uneven resistance and reflected noise gained me 25Mb, getting me from the original 50Mb to 82Mb in the end. Practical experience that put electrical noise into a real-world perspective.

One of my friends houses had serious issues with their VDSL - they were running homeplug wifi extenders/repeaters through their home, and they were jamming the hell out of the VDSL - you could start a file transfer and watch the DSL drop seconds later. Ferrite chokes stopped the dropouts, but still had speed loss issues - his solution was that we got new Cat6 cabling for the DSL line and mesh wifi removing both the noise source, and how it was getting 'absorbed'

Electrical signals can both be transmitted and received by the same wiring, and they can also reflect internally with the skin effect - that surface level skin energy can bounce back out of sync to the main signal and cause a little hiccup, an echo of noise. If you're unlucky that returning echo can also transmit in the radio sense too, making one sound echo echo echo echo....

Imagine you're in a room with a microphone and a speaker getting mic feedback, but it's also a tiled bathroom with an echo - under a certain noise level you're all good, but pass a certain threshold and you get noise that starts to snowball and escalate itself, instead of fading away

I tested my previously whiny GTX 1080 and the new RTX 3070 in my ITX system, and saw over 2000 FPS in one loading screen - and had zero whine anywhere in the PC. I wanted to test out these problems and damn well can't reproduce them now.

Those GPU's need a 6+8 and i did the "bad" thing and used both ends of one pigtail cable instead of two separate ones, where previously I'd used the first connector to the GPU, and tied the far end one up behind the first one, to hide it.

I'll take better images soon, hopefully these make sense in the meantime

Old image of my main PC - i used the 'close' connector and then hid the 'long' one behind the RGB cable, cable tied back onto itself. That would literally send its own noise into itself twice - once by reflecting off the unused connector at the end, and then as radio interference back into its earlier half, tied beside it.

If you're only using a single PCI-E connector on a cable, use the far end one - not the closer one. See if it helps.
avoid sharp bends to the cable - cable management has us fold them over flat, or pulled tight around parts of the case.
Simply try stretching it out with a big gentle curve and see if anything changes for a test run.



I'd love for someone who's got a problematic, whiny system to try the following:

1. Ferrite choke on the AC cable
2. Ferrite choke(s) on the GPU power cable(s)
3. Try the various combinations of power cable(s) - use the far end connectors, avoid sharp bends. Try using both connectors on one cable instead of two seperate ones, short term
(If it's a modular PSU, disconnect the unused one at the PSU!)

Well, mebbe the ferrite ones work differently.

Oh yeah, entirely different then. they arent in-line resistors or capacitors or anything

They used to be *EVERYWHERE*, even built into USB cables and such

Another VDSL related comment but also explains how big scale this can get: One persons old television sent out EMI surges taking out an entire english villages internet every time it powered on or off

It turned out that at 7am every morning, the occupant would switch on their old TV which would in turn knock out broadband for the entire village.

You could have a source of interference a centimeter away from tucking cables against each other, from a noisy device in your home, or be blasted by interference well outside your control.


Example photos of the GTX 1080 and corsair HX750i? that had really bad whine for me in the past.

1. The setup i had with the worst whine ever - and it fit my conclusions above perfectly.
I used the close end of the cable and tied them back onto themselves.
When i moved ITX it only had one cable to use, so i used that - and had no noise. I don't think it was the PSU change as much as the cabling change that helped.

2. These are messy as heck because they were tied together and left in the PSU box for a year or so, but this is how it should be done for a high wattage GPU.
Use both ends of the same cable for something under 250W, or use the far ends of both cables. Avoid folding the cable back over itself!

Heck, technically even using the farthest SATA and molex connectors could help reduce whine that comes from the PSU itself.
A single noisy device like an RGB controller or motor from a pump or fan could 'echo' and make one noisy signal repeat like many smaller noises and be greatly reduced just by making sure it's at the furthest plug, so theres no echo and the noise suppresion only deals with it once.

This is something i bet a lot of users never care about at all and could explain why people have such wildly different experiences with EMI and coil whine - the larger their case, the more likely they are to use the far ends of a cable.

 

[Count von Schwalbe]

Oh yeah, entirely different then. they arent in-line resistors or capacitors or anything

They used to be *EVERYWHERE*, even built into USB cables and such

USB is DC as well.

The only use I have seen for a choke on AC before this was as an easy way to limit voltage.

I guess the ferrite ones are insensitive to 60Hz fluctuation?

 

[Mussels]

USB is DC as well.

The only use I have seen for a choke on AC before this was as an easy way to limit voltage.

I guess the ferrite ones are insensitive to 60Hz fluctuation?

USB is DC, but it could pick up interference easily because it's long straight wires - think a USB powered speaker, they were super prone to electrical noise with pops and crackles when a phone recieved an SMS nearby

As a really obvious example, USB wall chargers are AC-DC and the inline filter prevents noise from the charger reaching the device, unless you're older than me you should be able to stick your ear next to the various chargers in your house and hear the coil whine from them, up close

 

[Nonam3]

Got myself MSI Gaming X Trio 4090 and this became very quickly nightmare and worst purchase in 20 years that I build custom PCs (for myself).
Purchased used one on warranty, didn't saved much, just so I could buy, using that price difference, water block. GPU by seller was purchased in end of 2022 just after release of 4000 series, first batch.

Stupidly, and this is on me, straight away stripped GPU and put water block. Reason - didn't had quick disconnect at the time and was eager to see 2x timer performance increase in compute tasks (Blender, Ai) and indeed performance is there... but at what cost.

Coil whine was problem from the get go. Compared to mine 3090 in FurMark is lauder like 10 times (can't hear 3090, easily 4090) using OpenPose 3090 is audible but 4090 is screaming !

Spec:
CPU: Threadripper 3960X
GPU: MSI Gaming X Trio 3090 and now MSI Gaming X Trio 4090
PSU: be quiet! Dark Power Pro 12 1500W + Seasonic PRIME TX-1000 80Plus Titanium 1000W
Case: be quiet! Dark Base Pro 900 Rev.1 (silent focused case)
Cooling: custom water cooling (CPU+GPU) with MO-RA3 420

What I tried:

1. Attempting to dampen case using glass
Got 2 glass sheets 4mm 60x50 cm + 16x30cm smaller sheets (covers for kitchen ovens) and put on sides and top - 0 difference, tried different arrangements with styrofoam in between etc.

2. Using different thermal pads on water block
Did not make a difference, used gelid. Coil whine got noticeable even in idle when used pads of improper thickness (gpu pcb bent I suspect). Got different thicknesses and tried different arrangements. At the end not difference.

3. Checking GPU water block
Assembled back GPU, installed quick disconnects for water coling, tested using stock cooler - exactly same coil whine

4. Checking house power
Unplugged most of appliance etc. no difference (didn't seen any interferences in operation so I doubt that this could cause issue)

5. Get additional GPUs to test
Purchase 2 gpus:
shop 1 - MSI Gaming X Trio 4090 (sample2) - exactly same as original one
shop 2 - MSI Suprim X 4090 - exactly same, to be honest even worse

6. Checking PSU
Hooked up second PSU (Seasonic 1000) only to GPU, rest of PC was running out of be Quiet 1500 - exactly same coil whine

7. Glued original GPU
Waiter 24h for cure, assembled water block - exactly same coil whine
Done some other testing, point 4, so waited yet another 24h - as above

Some pix from the work, crappy quality of pix (and work) using old iPad Air 2:




All above attempts resulted in no difference in perceivable acoustic operation of GPU aka. coild whine didn't changed.
When power limiting 4090 to 40% I can get to 3090 acoustic levels, but doing so does performance so...

Contacted MSI regarding warranty sticker and that it was torn - warranty is voided.
Did noticed that I still have 3090 sticker (dunno why it was intact) so most likely will either a) try to RMA it, with glue , b) sell (dont want to do that as I don't want to sell malfunctioning gpu - to me coil whine is malfunction)

If anyone from NVidia and MSi read this message I have a message from the bottom of my heart: https://tenor.com/pl/view/linus-torvalds-linus-nvidia-fuck-you-gif-19475186

If anyone have any suggestions I'm all ears.

 

[Mussels]

Damn man, that sucks - have you tried any of the other tricks i suggested in recent posts, with the ferrite beads and checking how you're powering the device with the cables?

 

[Veseleil]

If anyone have any suggestions I'm all ears.

What @Mussels said.

 

[Nonam3]

I have power strip with EMI/RFI protection, this one:

Premium-Protect-Line 60.000A extension lead with surge protection 8-way 3m H05VV-F 3G1,5

Premium-Protect-Line Power Strip 8-way with surge protection (3m cable, with switch, Made in Germany)

www.brennenstuhl.com

When tried using GPU, alone, on Seasonic PSU, there were no additional cables only 3 directly connected to the GPU and those are PCIe (8/6 pins) straight connector (just 8pin on PSU end and 6+2 on other - GPU, nothing, as in no connector, in between).

beQuiet PSU: checked A and B connectors on PSU side (that's how they marked them) and only GPU hooked up on those cables, end of those cables is 2 x 6+2PINs and I used only single 1 x 6+2PIN for any given connector (so I ended up with 3 cables from the PSU -> GPU for each individual line and 6 endings).

On that wall power outlet I had:
MSI GTX Gaming 970 + i7 4770 - no coil whine (at least I couldn't hear it, was using fans at the time)
MSI GTX Gaming X Trio 1080 Ti + i9 1090X - no coil whine (at least I couldn't hear it, was using fans at the time)
MSI RTX Gaming X Trio 3090 + TR3960x - moved to the water cooling, could hear coil whine, for me acceptable
MSI RTX Gaming X Trio 4090 + TR3960x - water, screams like world ends
MSI RTX Suprim X 4090 and sample 2 of MSI RTX Gaming X Trio 4090 as above.

Because I was, until now, happy with MSi products, did loosened my guard and carried out (among other reasons) with water block install.

Ferrite beads - personally I'm sceptic about that approach. I mean power strip that I use already has EMI/RFI and I don't hear any distortion etc. from mic / speakers / headphones so I doubt that these could cause issue. Also I read about them and in many cases, that people reported that they worked for them, they did said that i.e. my TV went berserk when microwave was turned on or noise was comming from the speakers etc. I didn't noticed such issues in my case.

Personally I think that these GPUs whole batches are defective, most likely were produced during pandemic (rushing them out due to crypto, low QA to accommodate lower labour forces etc.) and now companies pushes narratives that this is "normal" operation. I was researching coil whine when purchasing 1080Ti / 3090 and kinda skipped when purchased 4090. Now when I face same issue I see that with 4090 there is much higher amount of people reporting that those cards whine. But maybe it was like that always? Dunno, i.e. der8auer in his Asus 4080 review mentioned that it had terrible coil whine, but his Asus 4090 was OK, but some people in the comments stated that they had coil whined Asus 4090. And yes I know that from EE POV this is "normal" but many would disagree with this and call it sloppy design.

Don't want to be thin foiled guy, but to Me it seams like this coil whine will become next business opportunity, pay +30% for "silent version".

Anyway I will try ferrite beads, not so expensive, but I have low hopes for it to do anything.

Alternative, that I have higher hopes for, is building black box around case, literally I seen black wood that could be used to build cabinet and will re-use that purchased glass and buy some foam etc. and maybe / hopefully it will kill that noise.

Thanks everyone for writing theirs finding as this thread is great source of knowledge and help!

PS. something that I may try is to use UPS and, after charging it 100%, unplug and run tests. It should hold 400W load for 5-9mins just to perform those test. But then what would that prove? If it is silent that I'm dead in the water as my power line is basically useless and if this is still whining GPU then I'm scr3w3d as I did voided warranty ;/. No matter outcome I'm scr3w3d ;/

PS2. I was also researching PUS and, please do correct me If I'm wrong, but technically speaking we don't need 2-3 cables to deliver load > 200W. Usage of 2-3 cables is for safety reasons i.e. melting point of wire at diameter that PSU manufacturers use. If they used i.e. >0.5/1cm thick cables it should be fine to use single cable to deliver i.e. 600W. But again I may be wrong on this one.

PS3. Regarding EKWB waterblock for 4090 and coil whine going away after removal of backplate - EKWB used incorrectly thick thermo pads and IIRC 6 or 7 mm screws added to much pressure and solution is to either a) loosen screws b) remove back plate c) use 7-8mm screws. This info is from EKWB (run into it researching my problem and some youtube video, in comment section, had that info) and I did seen, in this thread, that someone else reported on this issue, so that's the reason why removal of back plate helped, coil whine on gpu itself will not change.

PS4. When applied glue I did done so only on inductors (highlighted in red) and that done nothing. Did seen video from diyPerks and on his card mosfets caused noise, not inductors (marked in picture below in blue). NOTE: please to correct me if my terminology is incorrect.



The question is did anyone tried to glue those? It appears that those are NCP303151A and can handle up to 140*C (well they will throttle at that temp) but the problem would be kind of the glue I guess as those will get toasty and probably not every super glue can handle such temps?

 

[Mussels]

A good summary I read on the EMI issue and ferrite beads is really simple
You can use it on the AC input to prevent external interference coming in, or you can use it on the internal components to prevent it going out.
The problem with a PC is that you've got dozens of components raising and lowering voltages and transmitting heaps of data constantly - and modern cases arent all 1.0mm thick steel to keep it contained any more.

It's extremely common for front panel audio in cases to have EMI noise issues for example and adding a ground wire or using ferrite beads can fix it. That sort of noise could absolutely be involved in coil whine, no one but you can have any idea how your PC is assembled - we don't know how well it's all connected, how your case and hardware connect for earthing, and so on... no one really does, until they spend a lot of time investigating a problem like the audio crackling.

Example 1, Example 2, Example3

This is someone advertising a product so their claims are "worst case scenario" but the testing is valid.
At the 25 second mark you can hear the EMI noise from a cordless phone, that when amplified up sounds very much like the coil whine we hear - it's a lower pitch because it's a lower frequency.
You can see him moving the USB cable toward and away from the phone and hearing the pitch change (and see it on the graph) and see just how easily a straight wire can pick up interference.

Having a power strip with EMI filtering is going to help from sources from the AC line, but it wont do anything to EMI interference picked up or created inside the PC itself.
personally, removing the noise is going to be a ton more effective than trying to dampen it - anything you do to absorb the noise is also going to obstruct airflow.

#1. Running a UPS battery can only prevent noise from the AC power input - it can verify noise is there (like a bad ground) but it wont do anything for EMI/radio interference, nor interference created within the PC itself.

#2. You are meant to spread the load out over multiple cables on high wattage GPUs, but when those cables have more than one connector, which one you use matters.
For short term testing, it doesnt matter at all.

#3. Adding or removing cooling won't ever remove coil whine, when people get lucky with coil whine after changing things like that, they're also reinstalling the entire card, wiring it differently and changing how it's fitting in the PC (usually a lot straighter, with less weight flexing the PCB) so it's impossible to say how they reduced whine, often it doesnt help at all.

4. You dont wanna glue your VRMs, they need thermal pads and cooling.



Someone beat me to it back in 2020
Edit: It was JonnyGuru, the quotes stated his name. He's well accepted as an expert on all things power supply.
If jonnyguru says its worth doing, it's worth doing.
He's basically the brains behind corsairs current power supplies.

Look at that advice above - unfortunately his view on this is a bit frustrating, because he assumes you buy a PSU *After* you have the GPU not the other way around.

They changed a PSU so it had no noise issues with a 2070 super, but it still had noise with 10 series GPU's and AMD vegas because of it's design - and the answer there was "this PSU and GPU basically arent compatible"

 

[Yougotonejob]

Register to reply this. (Sorry about broken English)
I am having the same issue too.
Coil Whine is coil get vibrate under some load. At some work load human can hear it. Simple is that.

The complex part of this is the power goes in have 2 source.
1 PSU -> GPU
2 PSU -> Mother board -> PCIe Lanes -> GPU
If PSU have low Ripple and remain DCish input all the time then you can get rid of some noise in the power input.
we know only choke make noise, there is Core VRM and Memory VRM. Memory frequency and power draw is somewhat stable. we can ignore them now.

Example AMD 580 will make coil vibrate at 1.1V * 100A = 110W and we lock the clock to 1366 Mhz to make it constant 110W here.
We know even if it draw a huge load but in office work and casual game it does not make sound.
that's where the refresh rate is at 60 FPS then it's about every 16.667 milliseconds to make it vibrate once, the rest of the time GPU idle.
Next Test is we cap 120 FPS is when you play some game FPS game like CS GO, we draw same power 110W the only difference is FPS is 120 right now.
120 FPS make each frame generate with a smaller gap, make it 8.33 ms per frame.
GPU get work done per frame is well under 1 ms and rest of 7 ms is wait for CPU to tell what next frame looks like.
Next test is the same just change to cap 240 FPS, then the gpu need about 3 ms in each frame. GPU will make a sound every 3 ms.
So, we got a different frequency here 60 Hz is just a bit above human hearing range (20Hz to 20,000Hz). human are not that sensitive to pick up.
Human voice are 85-225 Hz for male and 165-255 Hz for female.
And if you covert 16.67ms 8.33ms and 3ms back to Hz you get
Frequency (Hz) = 1 / Time period (seconds)
60 Hz 120Hz and 240Hz
Sorry at earth this point we dont have display over 20,000Hz so cap your FPS 60 is make coil whine less "pick up" by your ears.

Talk about ripples
check the middle section of this page https://www.tomshardware.com/reviews/corsair-rm750x-2021-power-supply-review/4
This is what I mean, Junny Guru work at Corsair now, and RM750x and other RMx are the best ripple control PSUs we got. When you talk about good power supply you can't just look at 80 gold vs 80 platinum.
That is one dimension of the many spec. Low ripple need filter caps and they will lost efficiency in 80 plus rating.
In GPU VRM you got mos + chock + Caps. Mos can have noise too. don't believe me? check here

Clean 12V will make Mos have less noise. Does it work for coil ? out side of my knowledge range. But as a guess I think very small effect at most. Coil is acting like a wind mill. when wind push the propeller it has inertia to slowly ramp up and when wind stop it has inertia to slowly stop. Coil make Amp goes in and out smooth.

Coil can vibrate in out side of 12 V (like 12.2V ) so some effort make input as clean as possible is working for sure. Just need make sure the sound is coming from what work load.
I guess we can pick up a R15 coil and run it by different Hz see what range it get noise. then replace it with better coil is the solution??

 

[Mussels]

The advice from JG above makes sense too, as it can be the primary *capacitor* in a PSU as the cause of the whine and not the coils - the input or the output, depending on the load and hardware used.
I find it fascinating that he states some GPU's are fine with certain PSU's even at the same wattages, then a newer series suddenly isn't.

They chang the PSU design to support the new GPUs, and then people buying a PSU get whine from the older card. What a nightmare.

 

[Shrek]

Talk about ripples
check the middle section of this page https://www.tomshardware.com/reviews/corsair-rm750x-2021-power-supply-review/4

Amazing that the video card noise comes from the MOSFETS (toward the end of the video)

 

[Mussels]

As to that wooden PC, that goes into a bit of what I mentioned earlier with crackling audio and front USB issues - the board there did have a metal backplate to screw onto, but the PSU and GPU were seperated and not in direct metal to metal contact with each other.

Some of these electronics parts designs rely on and assume there will be metal-to-metal contact between every component and even out their ground that way, including the PCI brackets on graphics cards, and every screw hole on the motherboard.

This is one *potential* cause of EMI and noise, because without a common ground, small differences make the parts work harder to even those differences out, or focus the noise in one place, instead of spread out between all the FETs chokes and capacitors.


he picked up the noise like this - note the GPU isn't wired in at the time for whatever reason, and he also bravely desoldered one of the mosfets

In the blue area he had no noise, but lower down he did - the problem is that he didnt specifically test which part made the noise in the closeups, it could be from the capacitors, the low or high mosfets - he only ruled out the chokes at the end of the setup - which of course don't do anything to prevent noise from the devices UPstream of them

JonnyGuru mentioned The PSU's meant to control the noise here at the primary output capacitor before sending to the motherboard and GPU and it's seemingly failing miserably - so a ferrite bead on the AC input could help if its noise from external sources making the PSU struggle, or over the appropriate power cable to the motherboard (like the 12V EPS cable) if they're picking up local EMI from another source. Wood isn't going to filter out any sort of EMI, period.

EK have a great example of this on their website to explain VRM cooling - note how the PSU power comes in two stages, the PWM controller is powered first, then the power is merged from the PSU and the drivers independently. Any difference in resistance, voltages, heat, or grounding between those may cause one part to work harder than the rest - whereas having them all receive equal power and equal grounding means the noise is spread out between them all, rather than focused in one spot.

The low and high FET's are what came up in buildzoids VRM breakdowns of motherboards in a lot of his reviews as if those parts aren't matched up properly, certain combinations of power are just very inefficient if the loads aren't balanced- like onboard audio whining at extremely low load as you move a mouse around, or a GPU that whines with a very easy GPU load that uses lots of GPU power, but no VRAM power (loading screens)

FET's are a high speed electrical switch like how PWM fans work - they get 12V in and pulse it to give out the desired voltage, then chokes and capacitors give the final smoothed out product to the CPU/GPU. If they're forced to work too slow they click, and too fast we hear them whine. They also have higher resistance and work harder when hot, so they can get noisier if they arent cooled - which I feel that youtuber had issues with because he had zero airflow, a single heatpipe pre-heated by the GPU won't be enough there.

 

[Yougotonejob]

https://thepcenthusiast.com/wp-content/uploads/2017/03/Zotac-GTX-1080-Mini-Review-16.jpg

not all has high and low mos it can be DrMOS too.
picture above is Zotac 1080 mini it has empty spot on one of the mos location. I guess this AIC could use this PCB for a different card like 1070 or 1060 mini.
I am not saying ferrite is not working, it just need confirm what is the root cause for each situation. if it come from the feed back/bounce back of internal pc or external device and it happen to be high frequency I think ferrite ring can be useful here.
Not only coil can make noise.

 

[lexluthermiester]

Amazing that the video card noise comes from the MOSFETS (toward the end of the video)

This makes me wonder..

 

[Yougotonejob]

This makes me wonder..

yes I am wonder too, can some one bereave enough to glue them?
I checked with some data sheet MOS are working under sub 100 C.
in theory it won't have any problem.

@Mussels
I checked AX1600i tear down and found there is individual magnetic ring in each out put. I am more lean toward it's going to be help suppress some ripple.

 

[lexluthermiester]

yes I am wonder too, can some one bereave enough to glue them?

That's what I was wondering. The GTX560 shown in the OP still makes a bit of noise. I'm wondering if it's the MOSFETs AND if gluing them would have an effect. I already have some ferrite chokes coming to experiment with. I think Mussels is on to something that could be one variable in the total equation.
This is the set I've ordered.

https://www.amazon.com/gp/product/B07CWCSNW9

EDIT:
Here's the progress:

I've glued the MOSFETs to seal them as well. Also glued the three open choke coils up near the display connectors.
They are curing now. Will update after a test run.

EDIT2:
This rendered a result. This GTX560 was much more quiet after the chokes were glued, but still made some noise. After now gluing the MOSFETs, there is even less sound. With the side of the case on, it is only just audible. So I'm left wondering how common MOSFET noise really is.

Bump

 

[Chocostick]

This rendered a result. This GTX560 was much more quiet after the chokes were glued, but still made some noise. After now gluing the MOSFETs, there is even less sound. With the side of the case on, it is only just audible. So I'm left wondering how common MOSFET noise really is.

Hello,

I am planning to seal the chokes of my RTX 2080 ti with liquid silicon glue following your tutorial to reduce coil whine.

Now you are mentioning that MOSFETs should be sealed as well.
Are there any known risks for doing that?

As far as I know silicone is non conductive and can handle temperatures much higher than the MOSFETs can reach so in theory I should be fine I guess.

Looking at your picture it seems you applied the glue around the MOSFETs just like you did around the chokes right?

BTW I have heatsinks glued on top of the MOSFETs for cooling and I wonder is this could worsen an eventual coil whine coming from there.

Thanks.

 

[Yougotonejob]

As far as I know silicone is non conductive and can handle temperatures much higher than the MOSFETs can reach so in theory I should be fine I guess.

I can help you out here.
Silicone acting like a rubber it damping the vibration, the sad part is you almost can't get it under the chip. the bound is only around chip. it will be a poor bound.
But great thing about damping is you will only hear the noise once, it obsorb the echo and diminishing vibration. when you using any rubbery type of thing as damping, less is more. You want to the vibration force stretchy the rubbery thing to the maximum. If you apply too much it will have less ideal effect. Watch this

CA glue is not acting like rubber it try to make chip and PCB as one it will transfer the small vibration to PCB and since the mass of the PCB is greater then small chip, it will still vibration but vibration together with PCB at a different frequency.

In theory they both work, just depends on how you do it.

BTW I have heatsinks glued on top of the MOSFETs for cooling and I wonder is this could worsen an eventual coil whine coming from there.

If you already have this and still hearing noise, i am guessing you can leave MOSFET for now??
Glue on top and with a heatsinks will damping some sound for sure.
If I were you I would try glue MOS last.

 

[lexluthermiester]

with liquid silicon glue

NO!! Don't use this kind of glue. Silicone glues contain acidic solvents which WILL DESTROY your card. Acids will severely corrode the metal pathways and components.

Do not use silicone glues with electronics, EVER!!!

 

[Yougotonejob]

not as a ADs but this work.

Chip Quik - Electronics Grade Silicone Adhesive Sealant (Clear) 20g (0.7oz) Squeeze Tube

there is electronic grade stuff that release no acid in cure time.
I think the glue stuff in power supply is the same type.

 

[lexluthermiester]

not as a ADs but this work.

Chip Quik - Electronics Grade Silicone Adhesive Sealant (Clear) 20g (0.7oz) Squeeze Tube

there is electronic grade stuff that release no acid in cure time.
I think the glue stuff in power supply is the same type.

That's some good glue, but I still can not recommend it for the application subject of this thread. The wicking action of super glue is an important factor. Silicone glue has no wicking action.