Graphics Card Coil Whine; An Investigation.

They make noise, too.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Spoon Boy
Posts: 168
Joined: Sun Jan 23, 2005 11:44 pm
Location: London

Graphics Card Coil Whine; An Investigation.

Post by Spoon Boy » Fri Nov 14, 2014 11:45 am

Not had a chance to read through it yet, but when i read the heading i instantly thought of SPCR.
http://www.hardwarecanucks.com/forum/ha ... ation.html

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Re: Graphics Card Coil Whine; An Investigation.

Post by MikeC » Fri Nov 14, 2014 2:22 pm

Nice find. :) Posted as front page news.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Re: Graphics Card Coil Whine; An Investigation.

Post by Das_Saunamies » Sat Nov 15, 2014 2:40 am

Definitely a worthwhile aspect of computer silence: component selection goes deeper than fans or HDDs/SSDs. I've even had whine from a mouse, not to mention several power supplies or converters for monitors and laptops.

Good to see manufacturers recognising the issue, at least as far as RMA policy is concerned. Eliminating the noise altogether would of course be the most desirable course of action...

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Re: Graphics Card Coil Whine; An Investigation.

Post by edh » Sat Nov 15, 2014 4:52 am

I wonder if someone with a deeper understanding of electronics at a component level (and disposable hardware) might be able to try a DIY fix? Carefully disolder the components at fault and replace with known good components? Don't know if this is an oversimplification or not.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Re: Graphics Card Coil Whine; An Investigation.

Post by Das_Saunamies » Sat Nov 15, 2014 5:19 am

I wonder if someone with a deeper understanding of electronics at a component level (and disposable hardware) might be able to try a DIY fix? Carefully disolder the components at fault and replace with known good components? Don't know if this is an oversimplification or not.
IIRC there have been positive results with a glue gun (and blu-tac?), as far as DIY fixes go. As you more than likely know, hence prompting the question, replacing delicate electronics like this is definitely doable, and if the parts have identical specs, it should also work 100% (assuming you don't wreck the components yourself while tinkering, it's high-precision work). The only problem I can see is sourcing appropriate replacement parts, but the circuitry involved sounds fairly common.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Graphics Card Coil Whine; An Investigation.

Post by quest_for_silence » Sat Nov 15, 2014 7:14 am

MikeC wrote:Nice find. :) Posted as front page news.

Code: Select all

...reference R9 290X and R9 290 should be avoided if you’re at all concerned about squealing or any other odd noise a GPU can make...
[/size] :?:

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Re: Graphics Card Coil Whine; An Investigation.

Post by Das_Saunamies » Sat Nov 15, 2014 7:20 am

quest_for_silence wrote:
MikeC wrote:Nice find. :) Posted as front page news.

Code: Select all

...reference R9 290X and R9 290 should be avoided if you’re at all concerned about squealing or any other odd noise a GPU can make...
[/size] :?:
Check the table above the quote.

bbalex
Posts: 33
Joined: Wed Aug 20, 2014 2:04 am

Re: Graphics Card Coil Whine; An Investigation.

Post by bbalex » Sat Nov 15, 2014 8:01 am

Computerbase (a german review magazine) also made some records of the coil whine on the GTX 980 and GTX 970 cards they tested: 11 GeForce GTX 970 im Test (4/6), 10 GeForce GTX 980 im Test (5/6). "Spulenfiepen" is the German term for the buzzing/squealing. I don't speak German, but google translate gives a good enough gist of the content. There are sound recordings on those webpages as well, from each card, and at different frame rates. There are also sound recordings of the cards at idle and load, to compare the noise their fans make. Some cards are really loud compared to the others - when I see on the web the people that have those cards saying they are "silent", I keep on remembering how loud they were in this test compared to others...

It's interesting that the results from Computerbase are not quite the same as from HardwareCanucks. Just one more proof of how great the unit-to-unit difference in noise level, from the same manufacturer and same model, is between these cards.

I have issues with both tests, though. Computerbase uses an open case environment (which is good, as at least it means the case fans and hard-disks are not going to spoil the results), but they only measure from the side of the card where the fans are. My experience with the Strixes I have(had), is that the buzzing is less perceivable from that position, as if the air turbulence of the fans smooths out that buzzing. However, listen from the back of the card, and be ready to pull your hair out... crackling and buzzing galore. I also did not see mentioned if the recording sound level really was the same in all the recording sessions, so that when one card seems louder than the other, it really is louder, and it's not the recording level that is louder. They don't seem to mention what games/tests they used when they tested this. Again, my experience is that the cards can be really silent in some tests/benchmarks, while they go summoning devils in others. The results are therefore not "definitive" in any way.

HardwareCanucks - I'm sorry to say, but I don't trust what they say there. I have no idea how they did the tests (open benches? no fans or hard-disks? were all these tests done at the same time, or do they just write there what they remember from some time in the past?). Much more critical is that there are no recordings, no measurements. I am not an expert, but from the bit of knowledge I have in the field of psychology, two of the things I know is to never trust memories, and that if you want to know the difference between two things (and don't have measurement instruments to use), compare them side-by-side, and don't rely on what your memory of the qualities of one product were. This stems from the fact that our brains don't measure and record "absolute" values, so our memory of how something was, and the current impression we have of something, might very well not be relative to the same "ground level". Our brains are much better however at sensing relative differences between two things, when the things are both available at the same time. That's why I'll always take side-by-side comparisons, over the subjective rankings of products reviewed at different moments in time. Rankings based on objective measurements done with calibrated instruments however, are a different matter. I'm also having issues with the game and test they used. In Furmark, my STRIXes were silent. Using Furmark as the coil whine test is not the "final test". Running the menu screen in TombRaider though, or its benhcmark, with frame rate capped at 60fps, makes so much noise it's no fun. And I doubt the TombRaider menu screen is more intensive than Furmark.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Re: Graphics Card Coil Whine; An Investigation.

Post by Das_Saunamies » Sat Nov 15, 2014 10:26 am

It's good to take the doubting approach. Maybe by 2020 we'll have enough people questioning the hardware sites to make them change how they have always gone about their testing - though I doubt it.

I trust HardwareCanucks not to lie, so it's more like a whine/no-whine roundup for me.

When it comes to position, I could not agree more. From the side my S12II was as quiet as a mouse, from behind it was a buzzing beehive. :idea: Post-RMA I have been with Nexus ever since, although sadly they discontinued their PSUs.

Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Re: Graphics Card Coil Whine; An Investigation.

Post by Devonavar » Sat Nov 15, 2014 8:35 pm

It's been years since I bought a graphics card, so I must be a bit out of date, but a question comes to mind:

Is this coil whine really so bad that it is audible above any fan and game noise? That's pretty bad, especially given that the issue only seems to show up under heavy load.

Sorry for the n00bish question, but I'm pretty out of date these days.

Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Re: Graphics Card Coil Whine; An Investigation.

Post by Devonavar » Sat Nov 15, 2014 8:46 pm

bbalex wrote:My experience with the Strixes I have(had), is that the buzzing is less perceivable from that position, as if the air turbulence of the fans smooths out that buzzing. However, listen from the back of the card, and be ready to pull your hair out... crackling and buzzing galore. I also did not see mentioned if the recording sound level really was the same in all the recording sessions, so that when one card seems louder than the other, it really is louder, and it's not the recording level that is louder. They don't seem to mention what games/tests they used when they tested this. Again, my experience is that the cards can be really silent in some tests/benchmarks, while they go summoning devils in others. The results are therefore not "definitive" in any way.
I actually think there's reason to question almost any "objective" results with coil whine. I came across coil whine quite a few times when I was actively reviewing at SPCR, most commonly in power supplies. Coil whine is *really* hard to measure and / or record in an accurate and consistent way. The most likely reason for some of the huge variances you are hearing is simply mic position relative to the noise source. It's a matter of millimetres, and every whine is slightly different. I remember seeing measurements vary by as much as 10 dB just from moving the mic.

The problem comes down to the character of the noise: It's very high frequency and very directional. This means you get large peaks of noise when the microphone is on axis or in a standing wave. The high-frequency nature of the noise means you get a lot of standing waves and null zones very close together, which is why moving the mic a millimetre or two or changing its angle can have such a large effect.

I never did find a good way of producing consistent results, my assumption is that any measurements of coil whine would have a margin of error that is large enough to make any number meaningless for comparison. I would use it to get a ballpark estimate of the order of magnitude, but I wouldn't consider it trustworthy.

I think the best you could do would be to move the mic around a bit and report the peak, but this isn't really a very repeatable method...

bbalex
Posts: 33
Joined: Wed Aug 20, 2014 2:04 am

Re: Graphics Card Coil Whine; An Investigation.

Post by bbalex » Sun Nov 16, 2014 2:07 am

There are ways to visualize and measure the sound field around an object. This way, you can see the intensity, direction, and spectral composition of the sound in different points around an object. This is probably the best way to visualize coil whine.

To do this, you can use 3D acoustic cameras. For something cheaper (well, cheaper by "industry" standards) there's something like the Soundbrush you can see in this article.

Like I said in another thread, I'm actually working for the company mentioned above, which does a lot of NVH consultancy. I don't think I'm revealing any industry secrets when I say that car manufacturers are very interested in this. Often, they have to identify and eliminate bad sounds caused by vibrations. Imagine trying to sell a luxury car with some awful sounds in the interior of the vehicle...

I've already seen pretty professional reviews when it comes to power consumption measurements and thermal imaging. Just noticed on Tom's Hardware an article of how they made those measurements on the GTX 970 and 980, revealing the huge power spikes that those type of cards have compared to their quoted average power consumption. And there are also people doing thermal imaging on video cards tested, like guru3d. Maybe we'll see something similar in the future when doing sound testing as well. I can only hope, right? :)

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Re: Graphics Card Coil Whine; An Investigation.

Post by MikeC » Sun Nov 16, 2014 8:52 am

LMS Soundbrush looks very cool. What is the price? Not that I think this is necessary for computer noise assessment.

The THG article you linked is excellent -- the best I've read there in a long time! Thanks for that, bbalex. Interestingly, it's the result of a whole year's research by the writer. Makes me feel more justified about how long some of our reviews take -- quality reviews do take more time. :wink:

btw, I have to remind Devon that he stopped working with SPCR before the anechoic chamber and current audio tools were introduced. Measuring the faintest electronic squeal is not an issue with a mic/system whose bandwidth is is a flat 20-20k, w/ noise floor of 7 dBA. If it's quieter than the 10 dBA noise floor of our chamber, the odds of being bothered by such a noise is pretty low anyway. Directionality may be a bit more problematic, but it is just a matter of moving the mic around the object.

cerbie
Posts: 46
Joined: Sat Dec 28, 2013 3:21 pm

Re: Graphics Card Coil Whine; An Investigation.

Post by cerbie » Sun Nov 16, 2014 11:05 am

Devonavar wrote:It's been years since I bought a graphics card, so I must be a bit out of date, but a question comes to mind:

Is this coil whine really so bad that it is audible above any fan and game noise? That's pretty bad, especially given that the issue only seems to show up under heavy load.

Sorry for the n00bish question, but I'm pretty out of date these days.
How loud are the fans? I was hearing it under fairly low loads, but I cannot hear my fans if my CPU or GPU temp does not rise quite a bit.

Many modern games, unlike modern music, have as much dynamic range as movies do, and it won't be broad spectrum noise, so hearing it over game noise would not be a problem.
Last edited by cerbie on Mon Nov 17, 2014 12:26 pm, edited 1 time in total.

nagi
Posts: 125
Joined: Tue Aug 31, 2010 6:48 pm
Location: Outside the box

Re: Graphics Card Coil Whine; An Investigation.

Post by nagi » Mon Nov 17, 2014 4:20 am

Another two examples for the immense sample variation:

As a heavily modded "reference" 290X owner, luckily my card does not produce whine. It was a Sapphire tri-cooler, which uses a reference design plus a manufacturer-custom 3-fan cooler. I simply replaced that with an even better one.

My previous ASUS 6850 directCU had, however, some intense whine issues under heavy load and framerate. Yes, the same one that SPCR reviewed and found to be great and silent. My experience with it was less than stellar unfortunately, and I could hear it even though the PC case was under my desk.

cloneman
Posts: 448
Joined: Sat May 21, 2005 9:48 am

Re: Graphics Card Coil Whine; An Investigation.

Post by cloneman » Sat Nov 29, 2014 4:48 am

thanks for this. I'm a bit of a coil whine enthusiast , this was a good read.

While all coil whine is bad, I don't have too much trouble with graphics cards because of whining is usually load triggered, like the article says, only in some very high load situations. I had this with my GTX260 "mild" - but with headphones on, it didn't matter.

On a scale of how annoyed I get can get with a GPU, load noise is maybe around a 4-5, whereas idle noise is a 7-8, and the worst is "low load / activity influenced" noise 10 or 11 (e.g. webpage scrolling or window moving noise - basically unbearable)

Thankfully coil whine is getting better. There was a time where 7 out of 10 laptops that I used had bad coil whine, now it seems to be getting more rare.

pod03
Posts: 77
Joined: Mon Mar 24, 2003 12:36 pm
Location: UK

Re: Graphics Card Coil Whine; An Investigation.

Post by pod03 » Sat Nov 29, 2014 3:37 pm

The Overclockers.co.uk/Manli GTX 970 just got a German review as well http://translate.google.co.uk/translate ... rev=search. This is one designed to minimise coil whine - as far as I can tell the whine is low under their testing - sample variance will still need to be considered.

loimlo
Posts: 762
Joined: Thu May 03, 2007 3:58 am
Location: Formosa

Re: Graphics Card Coil Whine; An Investigation.

Post by loimlo » Sun Nov 30, 2014 4:10 am

I encountered 3 times of coil whine during past 15 DIY years. :lol:

1. Gigabyte 2600XT whine. Solution : replace it with ASUS 3650 display card.
2. Seasonic S12 430W (1st gen) whine after changing MSI AM2+ board to ASRock AM3 board. Solution : Swap Seasonic with Superflower 350W gold.
3. Powercolor 7790 whine. Solution : Swap Seasonic S12 330 (1st gen) with Enermax Eco 400W.

To make matters complicated, I also found the whine disappear if assembling suspicious Seasonic PSU for other PC builds. Not sure the root cause, maybe it's the incompatibility between components ?? Anyway, for now I usually keep a quiet PSU to swap in the event of coil whine.

bonomork
Posts: 1
Joined: Thu Dec 04, 2014 3:14 am

Re: Graphics Card Coil Whine; An Investigation.

Post by bonomork » Thu Dec 04, 2014 3:27 am

I do not know if it makes a sense.
Just two days ago, coming back from an overclock setting the bios default, I noticed a graphic card coil whine (Asus R9 290 DCU2).
I checked all the bios option to find a solution (ASUS HERO VI).

Setting "Extreme" on DIGI+ CPU power phase control the noise disappear completely.

dr_traktor
Posts: 7
Joined: Tue Jan 06, 2015 11:59 am

Re: Graphics Card Coil Whine; An Investigation.

Post by dr_traktor » Tue Jan 06, 2015 1:58 pm

I am interested in buying a silent-at-idle graphics card that remains reasonably quiet under load.

With the introduction of the GTX 970 based cards I though I found just the thing, but it turned out before my purchase, that all semi-passive cards suffer from coil whine. Luckily it became widely known and strongly critizied, something which didn't happen before top-tier GPUs were marketed with the misleading 0 dB feature.

Currently I am on the lookout for a truly silent 970. For your review and feedback I have made a list of the originaly promosing but disappointing cards, which might receive an update in the future, and video cards that already address the issue.

Right now my bet is on the EVGA 04G-P4-3975-KR, which according to internet gossip, addresses the problem of coil whine, but it is not yet available in stores.

0 dB cards with occasionally obnoxious coil whine
GTX 970 GAMING 4G
ASUS STRIX-GTX970-DC2OC-4GD5
EVGA 04G-P4-2974-KR
EVGA 04G-P4-2978-KR

Quiet cards without coil whine
Asus GTX970-DC2OC-4GD5-BLACK - affordable, has the same power delivery as the 980
(explicit coil whine warranty)
GALAX GeForce GTX 970 OC Silent "Infinity Black Edition"
OcUK GeForce GTX 970 "NVIDIA 970 Cooler Edition"

Loud cards without coil whine
Gigabyte GV-N970G1 GAMING-4GD (loud fan)

Silence marketed, yet to be tested cards
EVGA 04G-P4-3975-KR Appears to be whiny as well according to first customer reviews

Sad news, even the 04G-P4-3975-KR is reported to have coil whine under certain conditions. But how much? We really need a proper noise conscious test of these cards, because coil whine is one of those things that may just be acceptable or could drive someone mad in an otherwise silent environment.

[Mod: removed link to sales page]
Last edited by dr_traktor on Sun Jul 12, 2015 6:30 am, edited 1 time in total.

NeilBlanchard
Moderator
Posts: 7681
Joined: Mon Dec 09, 2002 7:11 pm
Location: Maynard, MA, Eaarth
Contact:

Re: Graphics Card Coil Whine; An Investigation.

Post by NeilBlanchard » Wed Jan 07, 2015 8:43 pm

My new Zotac GTX 970 and/or the SeaSonic power supply is squealing when it gets close to 80C on the card. :( It seems to be in the 10kHz area? Not as high as the old TV's used to be, but plenty annoying.

I will put a Kill-A-Watt on the system to see if the old power supply can safely be used, and at least test whether the squeal is coming from the PS, or from an interaction.

Post Reply