One topic that I DON'T see discussed much here at SPCR is the dreadful noise that seems to be so common with modern GFX cards (a.k.a. coil whine)

I have put lots of time, money, and energy into obtaining a 'silent' PC that also happens to be a pretty good gaming PC. All of my recent GFX cards (with the exception of an 8600GT) have had some form of coil whine...the noise is apparent in both the GFX card and the PSU, and is much louder than any other noise in the system. It is extremely disruptive, and makes for a miserable gaming experience.

My most recent attempt (a 4890) ended up whining even when doing something as simple as scrolling in IE.

With the release of the 5870, reports seem to indicate that there is no coil whine. I'd love to hear some feedback to back this up (preferably from someone who has experienced coil whine in the past).

Alternately, if you've got a silent setup (no coil whine) and are running an 8800GT (or better), it would be helpful if you could post what components you're using (PSU, motherboard, GPU)?

For those who are wondering, the easiest way to detect coil whine is to either:
(1) Run the spinning cube in ATITool
(2) Run FurMark

I also think this is a topic important enough for a 10-page article on SPCR, but nothing seems to be in the works?

This is in my honest opinion the largest obstacle to achieving a spcr-grade quiet but powerful modern computer nowadays.

All the other things are pretty easy to solve now as CPU max power draw hasn't increased in many years and CPU idle power draw has lately rather decreased, while the efficiency of CPU coolers, PSUs etc has constantly improved, and most components are nowadays integrated on the mobo rather than separate add on cards.

I do have the dreaded coil whine which probably sounds just like yours. 8800GTS 512MB, Corsair VX450, Gigabyte EX38-DS4

Anyone mind telling how this coil wine sounds? I believe I hear some odd noise coming from my NVIDIA as well, but not sure if it can be linked to coil whine. Also is this something we can control by modding the cards for example or is it solely in the hands of the manufacturer? Finally... why does this sound happen? Can it also come from the motherboard or other components?

Examples of coil noise below. From what I understand, the effect is different depending on the GFX card/PSU combination.

I've gone through 5 video cards in the last year and all exhibited this problem...and I tested using a couple of different brands of PSUs/motherboards.

You may also find that some people blame the PSU, some blame the video card. In my testing, the sound was clearly audible from BOTH. One of the advantages of using a P182 is the isolation between PSU and motherboard...made it easier to verify this.

Right now, I'm using an 8800GT + Corsair HX620 + Gigabyte GA-EP45-UD3P. The 4870/4890 are supposed to have "digital" VRMs which address the problem, but in my case the whine was much worse on those cards. My 8800GT does whine, but only when gaming/stressing the card.

As a pseudo-SPCRer, I'm certainly willing to pay a premium for silence. If I could shell out an extra $100 for a vid card that was guaranteed to not whine, I probably would.

What gets me is that the manufacturers are pretty much ignoring the problem, and based on the forum (non-SPCR) responses, most end-users are accepting the noise as a fact of life. I guess these cards are so loud (stock fan) that a bit of whine doesn't make a difference.


Sorry about this...I'm not allowed to post links on SPCR yet...

One more post till I can put links

Try again:

http://www.youtube.com/watch?v=UZU9JKCdCVg

http://www.youtube.com/watch?v=uZkwNXoeOeY

http://www.youtube.com/watch?v=C4d23F2jehQ

Second video link explains it the best, yes I have that noise. I did quickly disassemble the system to the point that I could analyze the source of the sound better, no not from the NVIDIA. My PSU! This PC has a bequiet! PSU. I want my money back

VSync does make a difference, however I just don't see the "link". What does VSync actually make the graphics card do to get my PSU to scream? Almost felt as if it was screaming "heeelp heelp!" since my PC runs with extremely low airflow I could feel how hot it was Not directly hot, rather very warm but you get the idea

8800GT (OEM) + bequiet! 450W

Sardius, you deserve a gold medal for starting this thread. I feel that GFX "buzzing" is rapidly becoming the hardest issue to overcome when building a quiet gaming system.

I've bought quite a few gfx cards in my days. Not all of them have buzzed. Some of the cards that have buzzed are:

-ATI 9700 Pro (low freq. buzz. Nothing helped)
-NVidia 8800GTX (high freq. buzz that could be reduced to a minimum by capping fps to v-sync)
-NVidia 280GTX (low freq. buzz. Nothing helped).
-ATI 4870x2 (low freq. buzz. Nothing helped).

My most recent card, 4870x2, sounded like a jet engine with the stock cooler. With that deafening tornado it was impossible to detect any buzzing. I then replaced the cooler to an Accelero XTREME 4870X2. An amazing cooler that reduced fan noise to a minimum. After this I could hear electric buzzing whenever the card was put to some more serious work like games and benchmarking.

After reading about the problem on other forums many claimed/suggested it could be a "conflicting" issue with the PSU or motherboard even though the noise originated from the GFX card. I decided to try this. First I replaced my PSU Corsair 520HX to Antec CP 850. No change. Noise still there. I then tried a different motherboard from MSI Platinum P965 to ASUS (can't remember model). The noise from the GFX card was still there

I am certain that the source of the problem is the GFX card. Specifically the power supply part or VRM (Voltage Regulation Module) on the GFX card. The reference design from ATI or NVidia usually uses cheap low quality components that can't handle "full" power draw from the cards. Most brands (ASUS, MSI PowerColor etc.) don't change the VRM part from the reference design. This means if a certain model buzzes, brand is usually irrelevant in terms of buzzing. ATI announced this problem on the 48xx cards and said that modern gfx cards aren't designed to draw as much power as certain applications "forces" them to do. For exempel running Furmark on early Catalyst drivers caused crashes. ATI "fixed" this by capping performance in the drivers for certain gfx-patterns. They also referes to applications like Furmark as a "power virus".

In the new generation 58xx ATI has moved this power "throttling fix" from the drivers to the on-board hardware. I hope this will cause more expansive third-party brands to implement higher quality VRM which I also believe would help reduce buzzing noise.

ATI Power throttling links:
http://www.geeks3d.com/20090925/ati-cyp ... -and-occt/
http://www.anandtech.com/video/showdoc.aspx?i=3643&p=11



QFT

The people at SilentPCReview.com must start taking this "trend" seriously. If you google "gfx model + buzzing" you will most likely find a lot of threads if the gfx model is of a newer power hungry generation.

http://www.silentpcreview.com/forums/vi ... highlight=

It has been an issue before, but sometimes it's not really a coil that's making the sound.

That's a pretty old topic, but that comes to mind first about coil whine, because it's got the most solid solution using the search function here helps.

I agree that "coil" whine issues has been discussed before. But the "angle" of discussion has almost always been about faulty specimens or a specifically bad model. What we have now is rather a trend that ALL new power hungry gfx cards suffers from buzzing. Basically if you want a powerful gfx card you will get buzzing. This has got to change. The only way it will change is if reviewers start refering to it in thier reviews. I feel articles and reviews here on SPCR really do make a difference for future market trends.

I agree that it is not necessary whine from a coil we are hearing but rather electrical buzzing that seems to occur when "greater" power needs to be transformed/regulated. With the growing power usage of PC components I strongly feel that this will be a growing problem for fellow PC silencers. PSUs and gfx cards seems to be more affected than other "low" power components.

This subject really needs to be investigated in an article.

well, i absolutely agree, but from my standpoint (not really in a position to get high power components), it's a cost issue, although I did have thoughts of sending an email to AMD requesting to actually strip down the new 5850x2 attacking this (sparked by seeing this thread come up).

Considering i work with audio, i have the right equipment to actually run the card in a situation where I could visually see the high pitch frequencies the card emits. I also have this problem in my powerbook from the LCD inverter board (but i don't have the funds at the moment to buy a new one).

From what i've read here and elsewhere, even solid state coils have a whine, the ones that don't are usually shrink wrapped, and I haven't seen shrink wrapped coils on newer boards (unless you have, i don't go around looking for it). i live in an area with a relatively clean power current, so it would be easy for me to do this sort of thing.

i also haven't reviewed something for any website in a long time, and i've been looking to re-establish myself as a hardware tester (i love sitting on the bleeding edge of computer technology, even for just test purposes).

The ATI 4770 showed us that the reviewers get reference cards which are component-wise a lot different from retail. In the case of the 4770, this was so glaringly obvious, but some reviewers just gloss over the fact. Some of the manufacturers also further this deception by hiding sub-standard components under a heatsink, which most reviewers do not remove. This all has to change.

I'm glad to see discussion on this topic, and would absolutely love to see a section in most reviews dedicated to quantifying the noise...but I'm not sure that would be practical at this point in time.

Reviews should be based on consistent test methods. We have a reliable way to reproduce the noise (ATITool), but it is more difficult to come up with a way to reliably measure/quantify the noise.

Tougher still is determining exactly what causes the noise....and why it sometimes is heard only from the GPU, sometimes only from the PSU, and sometimes from both. It appears that component compatibility is part of the equation (so a given GFX card may squeal using one type of PSU and not with another). The motherboard may even play a part.

This component inter-dependibility is a nightmare for a reviewer. What works on the test setup may not work in the average PC. I suppose that a 'bad' review might put pressure on the major manufacturers to pay attention and fix the issue...but I'm not holding my breath.

Ultimately, I think the noise is the result of power/load fluctuations. I wonder if placing additional filters on the 12V line would reduce the fluctuations? Hmmm.....

Hi, as per some other people's fixes, sometimes with nail polish, I've managed to tame my GTX260 a lot with superglue! It still has some coil whine, but mostly blocked out by the case. It's much much quieter than before. At idle it was always OK, under load you could hear it from another room! Now I kind of "sense" it more than hear it as quiet and very high pitched.

Red dot = needs cooling as stock cooler connected to it.
Blue dot = super glued
(Click pics for much larger versions)


Everything with a blue dot has been glued! I used some very thin cheap generic “super glue,â€

Awesome post Seb!

I've got a bunch of questions about this, please be patient...

How did you determine which components to superglue? Were you able to isolate the noise or were you just trying to cover all bases? When you applied the glue, were you trying to get it underneath all components? Many of those components look like sealed, solid-state units.

What PSU were you using? Dare I ask how the PSU failed? What did you replace it with? Were you able to test the GPU noise change with the old PSU?

As this fix voids your warranty, it is a big gamble...you could end up with no noise reduction or busted components. A great idea if you're willing to take the risk.

This card is one I bought from ebay so not sure if I had any warranty, certainly not now anyway
Which components Well I kind of just guessed really Anything that looked like a capacitor or that it might contain a coil as I believe they are the main culprits for the whine
I was aiming to get the glue around and under the components. Originally didn’t expect to get under but the glue thin and flowed and the components not that tight to the board. The theory is that by gluing them tight to the board they are much less able to vibrate. I actually had two goes a good few days apart that may have helped get more/thicker/stronger bond. Originally I glued the items on the power connector end only and replaced the stock cooler. The whine was much improved in my test bed PC with a Seasonic S12II-330 PSU. The PSU whined a bit too, difficult to tell how much from the card and how much from the PSU.
I read a post, supposedly from a video card engineer/designer, that the glue or nail polish can seriously shorten the components life, possibly due to overheating them by blocking air flow / holes. This seams unlikely to me given they’d have to be very small as I didn’t notice any, but what do I know. Either way if you try anything like this consider the warranty voided and imminent fiery death likely

When I fitted the Akasa cooler etc I glued up the components again and the ones at the DVI connector end too. Used quite a bit of glue but as it was a 5 bottle pack from the £1 shop not a vast expense; nothing compared to the Akasa cooler and all the Enzotech heatsinks. (~£35 + £40 + the card @ £95 )
If the glue hadn’t helped I was going to try more still and maybe something a bit thicker.

I tried it back in my main rig originally with the stock cooler, glued components and my modified NeoHE 430 PSU. It had had it’s fan swapped and extras added to the heatsinks, there’s a thread about it. The PSU whined rather a lot. It survived the first round of gluing but still whined pretty much as before. The second round I was trying gluing up the big coils with silicon sealant, I don’t believe this caused the failure. I thought the PSU was drained (had tried to power PC with AC cord removed and get flash lights/fans etc draining the PSU) but I believe the cause of death was when I was scraping excess glue off the bottom of the circuit board there was a spark Don’t know what it did but PSU completely dead now, no output at all

Swapped another spare S12II-330 in my rig and it powered up fine. Who’d have thought a 330w PSU could run a Q9400 @ 3.2GHz and an overclocked GTX260
The S12 had some coil whine like the one in the test PC but with the noise of the stock cooler not very noticeable. Surprisingly don’t think the PSU fan was noisy either.

Looked at PSU and tempted by Enermax Pro82+ 385w around £45 but put off by reports of the fans getting noisy after 6-12 months. I ended up getting Antec Signature 650w on special for £80 + shipping. With 5yr warranty I’m not going to mod it ever, probably Had to mod the case a touch to get the 7â€

Sorry to awaken a several-month fallow thread, but this topic has caused me, a long time SPRC reader, to create a forum account and weigh in.

SPCR really really needs to do an exclusive report on this and start letting the manufacturers know that WE NOTICE and CARE about this noise. I'm pretty sure the big corporations are simply betting that the majority of users don't notice or don't care enough to complain so they save a few cents by slashing the quality of their products.

I'm in the middle of an on-going 4-month battle with ASUS to get my "whine" sorted out. I have a new P7P55D-E board, a Radeon 5870, and a SeaSonic X750 all of which whines in one way or another.

I've accepted that the Radeon will buzz faintly in the background as my 8800GTX did - at least it only does this while gaming when I have explosions to distract me. However, the MOTHERBOARD makes nice high-pitch squeaking sounds and even causes some kind of feedback to trigger the PSU to whine. I know it's the board because I've used the PSU with other boards and it's fine (actually, the X750 produces a very very feint buzz when powered on at all times, not covered in the review here and I was very upset to discover this). I mean, it's completely ridiculous. I move the mouse across the screen and the PSU goes "eeeennzzz" from the moment the cursor begins to move to the moment it stops.

This problem really needs to be addressed.

Older parts (3~4+ years) seem to be alright, but modern electronics seem to ALL have this problem. Hell, even my Thinkpad X61s has this problem and it makes me hate using it.

To the SPCR Staff: PLEASE DO A REPORT ON THIS ISSUE!

Thanks!

I would recommend to get server hardware it makes no funny noises like coil whine. This is true for server motherboards.

Also I can confirm that older GFX cards did not have this problem.
I first noticed this on some gigabyte mb, that were making coild whine while moving the mouse.

And why would server hardware not make coil noise? How is it different from desktop hardware noise-wise?
Actually, since many server motherboards go into racks, I'd expect the manufacturer to not care about noise.
Don't get me wrong, I'd be happy to find a coil-noise free motherboard for my planned Ivy Bridge build in 4 months.
My Gigabyte X58A-UD7 (a high end Skt 1366 mobo) has coil whine, and previous mobos also had some.
I don't get coil whine from my Powercolor 6850 SCS3 (with 600 rpm Slipstream), since it makes very little noise in Furmark.
Or if I ocasionally get some coil noise, I don't notice, because it's drowned by the motherboard!

What SebRad did is interesting, but before risking physical damage to a new expensive card, I'd rather like to see
a safe procedure detailed. I also believe that a high-temperature resistant glue should be used, otherwise it might
catch fire (or just smolder and generate funny smells) when applied onto the VRM components.
I doubt SPCR is going to do such a review, it would probably cost 1000$, and who is going to pay for it?

Electronic whine is the single most bothersome in laptops these days, at least from a noise-sensitive person's PoV. But ridiculously difficult to even document well. Why?
-- sample variance is bigger than you'd think
-- software/hardware interactions create noise under some circumstances in some samples
-- possible interactions with factors outside the unit, such as AC power, RF interference, etc

With desktop PCs, that AC power factor looms large. Take a PC and plug it into different AC sources, and you get different results. Go to a different house, neighborhood, city, country, etc.... I've often thought that SPCR being in Vancouver does a disservice in a way -- we have really good quality AC power, well regulated, and my house wiring is good. In NY & LA, for examples. I've heard of much worse conditions, from people I trust -- who have reviewed hardware for SPCR. I'm sure we hear less PSU and other electronic noise simply because of this.

It would be worthwhile, probably for the whole consumer electronics and computer industry for a really good study on this to be conducted, but I don't know who'd finance it. I mean, aside from noise-sensitive users (who probably number less than 20%), who would benefit? Unless, of course, we could identify this noise as a potential health or safety risk. Modern society (read: capitalist pigs) does not pay for things that are unprofitable or without high risk of costly litigation. It would probably cost manufacturers $$ & resources to keep the problem at bay, at least for a while until a reliable set of "best no-whine practices" were established. If I had the resources, I'd assemble a big range of hardware (and many samples of the same model from different production runs), employ a couple of techs, do a huge range of testing here to establish a baseline, then repeat at least core tests in older & newer homes (different wiring conditions), different cities, and perhaps different countries (including those with 100V & 220V lines). Realistic time frame -- 6 mos to a year. Cost -- at least $100,000; much more if lots of technical quantified data is needed.

On this tangent, if you haven't seen it --- http://www.storyofstuff.org/movies-all/ ... ectronics/

I wasn't thinking about a 'Test dozens of motherboards and graphics cards for coil noise' review when I estimated the 1000$ cost.
Such a review would only have a time-limited usefulness, since it can't guarantee that newer products won't get the coil noise.
But I also didn't explain what I was thinking about, so I'll do it now: how about a 'How to fix coil noise in a motherboard an gfx card'?
Take two known offenders of the latest generation hardware, a Socket 1155 mobo and a Radeon 7950 (coming soon!) card, and
try to fix the coil noise, doing what SebRad did and maybe some other tweaks. Show every step in detail, what can go wrong, etc.
I estimated 250$ for the mobo and 350$ for the gfx card (assuming you already have the rest of the system), four days work
at 100$ per day, it could be squeezed in 1000$ (probably more ). But it's still too expensive IMO, without someone (like a
manufacturer) paying for it, so that's why I think it's improbable.

That's relatively simple. I'm sure we could do that. There's probably enough gear here in the lab which make some whine noise to get us started.

FYI -- viewtopic.php?f=19&t=63610