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

[ad1]

And how would we apply any one them? Sure, put the stuff on in the factory before the parts are soldered in place and we can call it a great solution. Not so much post-consumer delivery. None of those compounds have a wicking action while liquid that allows them to seep into the spaces inside the coil and into the spaces where the coil housing meets the PCB.

Any adhesive that has wicking action before curing, that is PH neutral(or close to it) and chemically inert once cured would work as a great solution for this problem. No wicking qualities, it's going to be difficult(cumbersome) to work with and much less effective.

Out of curiosity, have you ever tried to remove a choke or cut it open after applying the glue, to see how much of the glue actually got wicked up into the coil itself? Just wondering where the primary noise reduction comes from... The coil vibrating less? The sound not escaping through air gaps? The coil casing vibrating less since it's "one with the pcb" once it's glued down? Obviously I wouldn't try this on a working part, since it would be destructive, but if someone has broken stuff lying around that has these types of chokes, it could be interesting to check.

 

[lexluthermiester]

Out of curiosity, have you ever tried to remove a choke or cut it open after applying the glue, to see how much of the glue actually got wicked up into the coil itself?

I have. A fair amount of it.

Just wondering where the primary noise reduction comes from... The coil vibrating less? The sound not escaping through air gaps? The coil casing vibrating less since it's "one with the pcb" once it's glued down?

Yes, yes and yes.

Obviously I wouldn't try this on a working part, since it would be destructive, but if someone has broken stuff lying around that has these types of chokes, it could be interesting to check.

I did. It was a 9800GTX that was squealing like a banshee. The coils were desoldered, then glued on the inside, then glued back onto the PCB and resoldered. As the card was also artifacting my crazy, but still worked otherwise I thought it would be a good test case. After the treatment it was totally silent. It was otherwise still defective but was silent. It was recycled later.

 

[Bill_Bright]

I meant no offense.

None taken. Just confused, hence the " ".

Out of curiosity, have you ever tried to remove a choke or cut it open after applying the glue, to see how much of the glue actually got wicked up into the coil itself?

Cut it "open"? I have to assume you are not familiar with what the vast majority of inductor coils look like.

I might have got myself into trouble by assuming again - that is, by assuming everyone knew what the typical inductor coil looked like.

Most look like this - a ferrite rod or loop with a conductor, typically copper wire, wound around it. There's really nothing to cut "open" - though certainly, they can be cut "through" for inspection or curiosity's sake - but that would permanently destroy the device. If the sound is originating from the coil vibrating against the PCB where it is mounted to the board, again, there's nothing to cut "open". But you sure can unmount the device to see if the sealant/adhesive got underneath.

There are some small inductor coils that are enclosed/sealed in cases, as seen in this illustration. But they are "sealed" so not sure how you would inject any sort of sealant/adhesive - thus not sure cutting them open would reveal anything useful.

Remember as a kid turning a nail into a magnet by wrapping a wire around the nail and attaching a battery to the wire's ends? Same principle.

In case there's a misunderstanding ( another bad assumption on my part) about transformer "plates", they are essentially sheets of ferrite materials that make up the "core" as seen in this illustration. They typically are bonded together and occasionally, the bond breaks and the plates (sheets) vibrate against each other and "chatter", commonly at 50 - 60Hz, thus "hum". Note the term "core" might be misleading. In this case, it does NOT mean center, like the core of the earth. That said, the center "limb" of the core does indeed go through the center in this example.

 

[Shrek]

Low frequency transformers use steel plates

High frequency transformers use ferrite cores

 

[Bill_Bright]

When I said ferrite, I meant it in general term for "iron" based - not a specific compound of elements.

Ferrite | Metallurgy for Dummies

 

[ad1]

None taken. Just confused, hence the " ".


Cut it "open"? I have to assume you are not familiar with what the vast majority of inductor coils look like.

I might have got myself into trouble by assuming again - that is, by assuming everyone knew what the typical inductor coil looked like.

Most look like this - a ferrite rod or loop with a conductor, typically copper wire, wound around it. There's really nothing to cut "open" - though certainly, they can be cut "through" for inspection or curiosity's sake - but that would permanently destroy the device. If the sound is originating from the coil vibrating against the PCB where it is mounted to the board, again, there's nothing to cut "open". But you sure can unmount the device to see if the sealant/adhesive got underneath.

There are some small inductor coils that are enclosed/sealed in cases, as seen in this illustration. But they are "sealed" so not sure how you would inject any sort of sealant/adhesive - thus not sure cutting them open would reveal anything useful.

Remember as a kid turning a nail into a magnet by wrapping a wire around the nail and attaching a battery to the wire's ends? Same principle.

I was assuming they looked something like this...

and I simply meant to try and break the casing off, in case it'd be hard to remove it from the board, but I suppose you could just rip it off and look.

And as I wrote, it would be out of curiosity, a destructive procedure, on an already broken component, just to see if any glue got wicked onto the coil itself (assuming the bottom of it is exposed to the PCB side).

I don't want to get into a technicality argument. I know, I clicked on your signature, you are very smart because you've worked for 60 years (as most old people also have).

 

[Bill_Bright]

and I simply meant to try and break the casing off, in case it'd be hard to remove it from the board, but I suppose you could just rip it off and look.

I was just tying to understand where you were coming from - and I do and don't disagree.

I am just saying for those type inductors that are enclosed in casings, I don't believe you can inject a sealant/adhesive inside to stop any whining - at least not without destroying the device. HOWEVER, I've never tried it so I could be wrong.

For our systems, if the whine became objectionable, we just ended up replacing the device - unlike inductors with exposed windings. With those, we typically would first use epoxy or hot glue to silence them.

So as you noted, it would be out of curiosity - for me, mostly to see how they are constructed. Nothing wrong with that! I've torn apart many things to see what's inside and how they worked. In fact, I seem to remember getting my a$$ whooped a time or two for that when I was a kid!

It should be noted that a whining coil is simply an annoyance for dogs and others sensitive to the noise. The whine does not change the electrical properties of the device or circuit it is in.

 

[KLiKzg]

Nonsense.

I hate to burst your bubble KLIKzg, but contrary to your belief, there are many high tech fields (in and out of IT) with much tighter and stricter tolerances and compliance levels than automotive certs.

The main reason some compliance requirements in the automotive industry are so strict is NOT due to the electronics. Rather it is the destruction potentials the speeds combined with the sheer mass of the vehicles can cause when they collide with other objects.

While your reaction can rip your rotator cuff apart if you grab on to a spark plug wire, it won't kill you. Stick your hand inside a live PSU, and it might be Tango Uniform for you.

Ever tried to grab a 800V cord on a car? Check your facts & see where do I come from...maybe you will add 2 & 2.

 

[Bill_Bright]

Check your facts

I do, all the time. I might suggest you do the same. Note doing so is so important, I even put it in my signature block as a reminder to us all, including myself.

Ever tried to grab a 800V cord on a car?

Gee whiz, dude. Don't be silly. Note I specifically said "spark plug wire". How many EV (not hybrid) cars do you know have spark plug wires?

That said, assuming the cable is not damaged, you can grab the 800V cable from an EV battery (if you can access it - they, by design, are tucked away) and not feel a thing! Why? Because it is only (by comparison) 800VDC.

And of course, if you had checked your facts you would have seen that Level 3 EV Charging, also known as "DC Fast Charging" supplies DC voltage directly from the charging stations at voltage levels up to 1000VDC. And guess what? Those EV car owners handle those cables all the time! And don't feel the slightest tingle.

A spark plug cable, on the other hand, will carry 15,000 to 20,000 volts and in some cases on modern cars, 45,000 all the way up to 60,000 volts!! And it is not traditional DC from the battery, but pulsed DC. And unlike with EVs, spark plug cables are exposed to road grime, salt water and weathering (sans direct sunlight) which breaks down the insulation.

see where do I come from

No clue what universe you come from but I don't think it is the same as the rest of us.

That said, this topic is about coil whine in computer electronics. So please - stop trying to make moot the whole point with your OT obfuscations. Let's get back to the OPs topic, or move on.

 

[lexluthermiester]

Guys, we've strayed from the thread topic. Let's rope ourselves in. Thanks.

 

[lexluthermiester]

Bump for Edit #2 to add a video from Adam Savage into the thread header post.

I'll post it here for those who have been here before.

 

[Zangief]

Hey guys.

My "GeForce RTX 4070 SUPER 12G VENTUS 2X OC" was always making a high-pitched noise at idle, but I have greatly improved it by installing the ferrite beads mentioned in this thread.
The original volume was unbearable, but now it's drowned out by the quiet fan noise at idle.
Thanks for the great information you provided. I created an account to say a word of thanks.

 

[Shrek]

beats voiding the warrantee

 

[lexluthermiester]

Hey guys.

My "GeForce RTX 4070 SUPER 12G VENTUS 2X OC" was always making a high-pitched noise at idle, but I have greatly improved it by installing the ferrite beads mentioned in this thread.
The original volume was unbearable, but now it's drowned out by the quiet fan noise at idle.
Thanks for the great information you provided. I created an account to say a word of thanks.

View attachment 361788

Glad that worked for you! Simple & easy is always the better way to do things when possible. Nice "Y" power cable as well!

Also, Welcome to TPU!

 

[Bill_Bright]

The thing is, ferrite beads are used to suppress EMI, not sound created by things vibrating and physically chattering. So for sure, I am glad they worked for you, but your problem most likely was not coil whine.

 

[Count von Schwalbe]

The thing is, ferrite beads are used to suppress EMI, not sound created by things vibrating and physically chattering. So for sure, I am glad they worked for you, but your problem most likely was not coil whine.

TBF, the noise in the power from the PSU can cause the main VRM chokes to vibrate. Coil whine can have many sources.

 

[Shrek]

Hey guys.

My "GeForce RTX 4070 SUPER 12G VENTUS 2X OC" was always making a high-pitched noise at idle, but I have greatly improved it by installing the ferrite beads mentioned in this thread.
The original volume was unbearable, but now it's drowned out by the quiet fan noise at idle.
Thanks for the great information you provided. I created an account to say a word of thanks.

View attachment 361788

@Zangief
I think this is a very important development; could we ask you to record the sound with and without ferrite cores?

 

[Bill_Bright]

TBF, the noise in the power from the PSU can cause the main VRM chokes to vibrate. Coil whine can have many sources

True, but if you have so much noise riding the PSU's DC output voltages that it creates enough physical movement in those chokes to be so audible as to be annoying, that PSU is faulty and needs to be replaced.

I note the ATX standard dictates that ripple and noise from 10Hz up to 20MHz must not exceed 120 mV p-p (millivolts peak to peak) on the 12V rails and only 50 mV p-p on the +5V and +3.3V rails. That is not very much at all. And if you look at most PSU reviews, ripple and noise suppression typically is much better with values over 30 mV p-p typically considered sub-par. I note for example, even this budget Corsair CX650M review pointed out the PSU's good ripple suppression, with it averaging below 30 mV p-p across the board.

So while technically you are correct, it should not be seen in the real world - at least not with a properly functioning supply.

 

[Shrek]

What really matters is how with ferrite core compares to without.

No need to ponder or conjecture; we have someone who can do the experiment.

 

[Bill_Bright]

0.12V AC is large in my opinion and could run quite a speaker

As I noted, 0.12V or 120mV) is the maximum allowed, and not typically seen in any decent, properly functioning, supply. Plus, it all depends on the frequency whether or not it could be heard.

Clearly, it would not be 50/60Hz as that would be heard as low frequency hum, not a high pitched whine. There is no industry standard but from what I can find, it seems most computer switching PSUs have a switching frequency up in the low 100 - 200KHz range, well above the ability of human ears. Even if from a harmonic, it would have to be 3rd and 4th and then the amplitude would be much much lower than 120mV.

So this again in my view points to EMI from another source, or a faulty component.

No need to ponder or conjecture; we have someone who can do the experiment.

Yes, but that would be a one-off anecdotal example and not conclusive, except for that specific unit. So it seem pondering and conjecturing is all we can do.

 

[Shrek]

I note the ATX standard dictates that ripple and noise from 10Hz up to 20MHz must not exceed 120 mV p-p (millivolts peak to peak) on the 12V rails

You already noted that the 0.12V covered the audio range (and more).

it seems most computer switching PSUs have a switching frequency up in the low 100 - 200KHz range, well above the ability of human ears.

It is 100KHz modulated, so frequencies less than 100KHz are present (independent of 60Hz)

 

[Bill_Bright]

You already noted that the 0.12V covered the audio range (and more).

Huh? Where? Volts does not equate to audio.

It is 100KHz modulated, so frequencies less than 100KHz are present (independent of 60Hz)

So what? That does not mean they are audible.

 

[bobogdan625]

Hello guys, I tried the glue method(used a ethyl 2-CA glue, ethyl is not that strong of a solvent so it does not affect the PCB) and had no improvement. I left the glue to cure overnight (about 12-14h) and this morning I added a bit more glue, fired up RDR2 to test it and the whining still persists. I can attach a picture if you guys want. What should I do next? I tried this on a Powercolor Fighter 6700XT. I heard that using 2x8 pins cables instead of a split 1x8 like I have may drown the whine a little. What do you guys think? Undervolting the AMD cards is an absolute nightmare, in fact, AMD cards cannot be undervolted, you push the voltage slider down and the frequency goes down with it, you cannot undervolt without losing some performance, plus the voltage does not stay locked like it does with nVidia. I had an EVGA SC2 GTX1080 and that thing was a dream, super easy undervolting and did not lost any performance, only the power draw went down, the performance stood the same, like it should be.

 

[ad1]

Hey guys.

My "GeForce RTX 4070 SUPER 12G VENTUS 2X OC" was always making a high-pitched noise at idle, but I have greatly improved it by installing the ferrite beads mentioned in this thread.
The original volume was unbearable, but now it's drowned out by the quiet fan noise at idle.
Thanks for the great information you provided. I created an account to say a word of thanks.

I have a couple of notes related to this.

My 4090 screams like a banshee under heavy load. I have a high quality power supply and I've tried with 2 power supplies, even with one powering only the GPU and another one powering the rest of the PC. No change in noise under load. It's perfectly silent at idle. I'm pretty sure my noise is not PSU related.

I just tried adding a ferrite bead to my 12vhpwr cable and it didn't reduce the noise. It did however introduce new sounds from the bead itself vibrating, probably both against the plastic casing and the 2 halves against one another.

I'm glad to hear that your issue was resolved with a couple of cheap ferrite beads - definitely more cost effective than changing the PSU. Personally though, I would probably try a different PSU and if it's silent without the beads, I'd swap.

 

[KLiKzg]

I have a couple of notes related to this.

My 4090 screams like a banshee under heavy load. I have a high quality power supply and I've tried with 2 power supplies, even with one powering only the GPU and another one powering the rest of the PC. No change in noise under load. It's perfectly silent at idle. I'm pretty sure my noise is not PSU related.

I just tried adding a ferrite bead to my 12vhpwr cable and it didn't reduce the noise. It did however introduce new sounds from the bead itself vibrating, probably both against the plastic casing and the 2 halves against one another.

I'm glad to hear that your issue was resolved with a couple of cheap ferrite beads - definitely more cost effective than changing the PSU. Personally though, I would probably try a different PSU and if it's silent without the beads, I'd swap.

Take some "isolator" (do not do it with hands) & touch your card under load on different places. You will find which part of it is making a noise & then decide how to fix it!

I would not plug-unplug so many times 4090, as the connectors have finite number of connecting.