High pitched whistling noise related to kernel version?

ysetdng · 2014-07-11 10:56:47

Hi, I posted this Q also on the Acer C720 thread but I guess the topic is not restricted to this singular model.

I recently noticed a very high pitched noise on my laptop, suspecting it to stem from the CPU as it gets louder when I type or scroll. It's a really high frequency, probably not even audible to everyone (I'm 25 and they say you stop hearing certain frequencies around that age). I may be mistaken but can't recall noticing this before upgrading to a custom 3.16 kernel. Do you think it might indeed has something to do with the kernel or does this seem unlikely?

Thanks, Y

brebs · 2014-07-11 11:14:29

Probably the characteristics of CPU power-saving. E.g. cstate, and pstate (frequency).

Unplugging the power supply might help, if it gets in a state of electrical interference reasonance (or whatever it's called).

I believe it's fairly common - most people just ignore it.

emeres · 2014-07-12 17:48:53

This could have several reasons.
To rule out kernel aspects, either downgrade or just boot a live medium with a previous/other kernel on it. Has the laptop intel chipsets or other intel hardware by any chance? Do you notice a smell coming from the laptop? It could be coming from internal or external power supply circuit. Do you clean the laptop [inside] often?

Brebs suggestion about operational states is the most probable cause. Try another power governor.

Side note: It is called EMI and resonance, just like in audio.

R00KIE · 2014-07-13 10:32:25

Could be interesting to compare the output of 'zgrep ^CONFIG_HZ /proc/config.gz' between and "good" and "bad" kernel.

Trilby · 2014-07-13 11:24:55

FWIW I've frequently been annoyed by this on my laptop. Though, interestingly it seems to have dissapeared for me recently. When it was happening, it seemed it was only happening when running on battery while plugging in stopped it.

I have very good high-pitch hearing ... so ignoring it was difficult, scrolling a page in firefox was horrible.

mich41 · 2014-07-13 15:37:31

Sounds like "coil noise", i.e. the noise cause by coils vibrating in accordance with fluctuations in the amount of conducted current.

It's perfectly possible that this appeared as a result of kernel upgrade or reconfiguration. Check if your custom kernel configuration doesn't use a different CPUFreq governor, different configuration of CPU power saving drivers or something like that.

ewaller · 2014-07-13 16:07:23

mich41 wrote:

Sounds like "coil noise", i.e. the noise cause by coils vibrating in accordance with fluctuations in the amount of conducted current.

It is also called microphonics. The link article has to do with pickup caused by vibrations, but the same term applies when electrical stimulation causes components to vibrate (making noise).

I am a bit skeptical that a kernel would cause changes in the microphonic effects of a power supply or logic. Has the OP tried disabling or unplugging external speakers? Any chance that there is a microphone open? Any chance there is a new audio source that is enabled and making noise that is then coming out of the sound system?

Trilby · 2014-07-13 16:14:14

ewaller wrote:

I am a bit skeptical that a kernel would cause changes in the microphonic effects

I have only annecdotal evidence, but when I had this sound, I did all those checks you suggest, and then some. I even tried to localize the source of the sound and I tentatively concluded it was GPU related. While this thread has referred to kernel upgrades, it's not clear that this has been tied to a kernel upgrade, but merely coincidental with it. It could just as easily be any other package updated in the same time period. So - given my almost complete ignorance of how any of this could happen - I wonder, would you'd be as skeptical about changes in video drivers or other driver components affecting these sounds?

ewaller · 2014-07-14 06:36:51

Sorry I have been away most of the weekend. Software behavior can cause some circuits to physically make sounds, but it usually has to do with electro-mechanical things. When I was a kid, I saw some specially crafted software that made a chain printer create a recognizable tune (She'l be a comin around the mountain). I've seen switching power supplies "sing" when the switching frequency is near a mechanical resonance of a magnetic component - the mechanical resonance is excited by the Lorentz force when the current happens to be switching at the resonant frequency. In general, unrelated software things don't fart around with settings like switching frequency. Now, if the software does something to significantly impact the current drawn from some power supply someplace, and the frequency, or pulse widths change a bunch, then, yes, a fundamental or a harmonic could excite something that had not been excited before. I have, however, been fooled many times when shaving with Occam's razor

You said GPU related. Were you using a CRT display? CRT's are chock full of those electromagnet thingies, and there all kinds of frequencies and harmonics. I could definitely see a CRT sing in the 10 to 20 kHz band if you changed horizontal frequency, number of lines, etc. LCDs, not so much.

Trilby · 2014-07-14 11:30:03

In my case it was an LCD - Lenovo X200 laptop.

mich41 · 2014-07-14 16:38:04

ewaller wrote:

I am a bit skeptical that a kernel would cause changes in the microphonic effects of a power supply or logic.

The kernel can periodically order some devices to power up and down, thereby creating a periodic variation in current consumption.
The kernel can decide whether something turns on/off periodically, possibly creating microphonic effects, or whether it runs constantly but at lower performance level.
The kernel can accidentally order devices to consume the exact amount of power which causes some voltage regulator to output PWM containing low frequency oscillations.

And probably few other completely insane things. FWIW, I have a computer supported by two CPUFreq drivers - K8 and ACPI. With K8 the CPU voltage regulators produce lots of noise when the CPU is ran at full speed but underutilized, while with ACPI under the exact same conditions they are quiet. Something must have been configured differently, go figure.

emeres · 2014-07-14 17:08:06

ewaller wrote:

electromagnet thingies

You mean inductors with ferromagnetic cores?

@Trilby could you test one of the kernels you were running back then? It would certainly be no proof, only an indicator, but still better than speculation. The OP could do the same, test another kernel that is, but I think she/he gave up [j].

I assume the PSUs in laptops for the CPU, GPU and the LCD, have at least one discrete part that is controlled by frequency (it has to be the case, because of energy efficiency). Depending on the switching frequency and the whole physical structure of the laptop, as well as other components characteristics (that may have not been considered during design in this particular scenario) the PSU could very well produce high pitch sounds. Choke coils could be responsible also, but from what I saw [or heard] it usually is one of the discrete switching components (mosfets, hexfets or igbts or whatever is used currently). In order to cut costs the parameters are only as good as necessary, which usually means lower switching frequencies fulfilling requirements for the proper operation, which is probably reduced to one or two scenarios like performance and power saving mode. And remember that there are high density currents in those laptops, which would be also an argument for mechanical vibrations actually.

Best bet would be to ask an engineer working at Intel, Lenovo, Acer or similar.

Some reference from Intel about operational states and voltage-frequency control: 1, 2.

Edit: @mich41 did not see your post in time.

Last edited by emeres (2014-07-14 17:14:09)

ewaller · 2014-07-14 17:21:48

emeres wrote:

You mean inductors with ferromagnetic cores?

Yeah, basically. Deflection coils, flyback transformers, focus coils, degaussing coils, acceleration voltage power supply. Most of them have ferromagnetic cores. The deflection and focus coils are mostly around a vacuum. It is moot because I don't think the issue in this case has anything to do with a CRT.

ewaller · 2014-07-14 17:28:22

mich41 wrote:

The kernel can periodically order some devices to power up and down, thereby creating a periodic variation in current consumption.
The kernel can decide whether something turns on/off periodically, possibly creating microphonic effects, or whether it runs constantly but at lower performance level.
The kernel can accidentally order devices to consume the exact amount of power which causes some voltage regulator to output PWM containing low frequency oscillations.
And probably few other completely insane things. FWIW, I have a computer supported by two CPUFreq drivers - K8 and ACPI. With K8 the CPU voltage regulators produce lots of noise when the CPU is ran at full speed but underutilized, while with ACPI under the exact same conditions they are quiet. Something must have been configured differently, go figure.

I agree on all counts. I just would not expect a kernel upgrade to change things so as to cause these issues in systems that had had not had the problem before. It could very well happen. and is is definitely a branch on the failure mode fishbone diagram. It just would not be the first mode I would pursue. As I said, I have been wrong in my trouble shooting before. I guarantee I will be wrong in the future

Pse · 2014-07-14 18:47:09

I have experienced this in the past with a Radeon card and mouse movement. I don't own that card anymore (for other reasons), so I couldn't say exactly when it started happening, but I distinctly remember the sound it made. I assumed it was related to power management causing certain components to vibrate on the PCB, as was mentioned in this thread above; however I am not an expert

Trilby · 2014-07-14 18:58:08

Sorry I can't downgrade to test this very well. I don't remember exactly when this was happening - and whenever it was it was long enough ago that downgrading to those packages would be quite involved.

emeres · 2014-07-14 19:49:03

@Trilby You know a simple live medium test would be enough. Yes, anything else than Arch Linux is not Arch Linux. Still, if the kernel would play a predominant role here, it may reveal itself. Worth a try, but I do understand if you do not want/have no time to.

Trilby · 2014-07-14 21:17:14

I'd have to make a live media for a gui distro, or I'd have to install a gui and firefox on to one of my arch isos. Then I could create some of the circumstances - but what would that say: if I get the sound again, that says very little. If I don't, it says even less. The odds of a randomly selected live media having exactly the kernel version that was coincident with the time when I heard this is highly unlikely.

emeres · 2014-07-14 22:07:43

I thought, that you had the issue constantly up to a certain point, ergo multiple kernel versions were involved, my bad. Like I wrote, it would only indicate a connection, but you are right, highly unlikely. I just have to disagree on the informative value however, it is still input, maybe not one of great value, but still input.

More reference on the subject 1, 2.

Dell wrote:

When a specific voltage is applied to these components, they begin to resonate producing sounds that fall within the range of human hearing (15 – 20 KHz).
The noise is normal and within the acoustic specifications of the computer.

Translation: Hard cheese, customer.

@ewaller: I can see other components doing this while maintaining normal operation, but focus and deflection coils? That should make the picture unusable (or at least blurry).

Last edited by emeres (2014-07-14 22:23:09)

Arch Linux

#1 2014-07-11 10:56:47

High pitched whistling noise related to kernel version?

#2 2014-07-11 11:14:29

Re: High pitched whistling noise related to kernel version?

#3 2014-07-12 17:48:53

Re: High pitched whistling noise related to kernel version?

#4 2014-07-13 10:32:25

Re: High pitched whistling noise related to kernel version?

#5 2014-07-13 11:24:55

Re: High pitched whistling noise related to kernel version?

#6 2014-07-13 15:37:31

Re: High pitched whistling noise related to kernel version?

#7 2014-07-13 16:07:23

Re: High pitched whistling noise related to kernel version?

#8 2014-07-13 16:14:14

Re: High pitched whistling noise related to kernel version?

#9 2014-07-14 06:36:51

Re: High pitched whistling noise related to kernel version?

#10 2014-07-14 11:30:03

Re: High pitched whistling noise related to kernel version?

#11 2014-07-14 16:38:04

Re: High pitched whistling noise related to kernel version?

#12 2014-07-14 17:08:06

Re: High pitched whistling noise related to kernel version?

#13 2014-07-14 17:21:48

Re: High pitched whistling noise related to kernel version?

#14 2014-07-14 17:28:22

Re: High pitched whistling noise related to kernel version?

#15 2014-07-14 18:47:09

Re: High pitched whistling noise related to kernel version?

#16 2014-07-14 18:58:08

Re: High pitched whistling noise related to kernel version?

#17 2014-07-14 19:49:03

Re: High pitched whistling noise related to kernel version?

#18 2014-07-14 21:17:14

Re: High pitched whistling noise related to kernel version?

#19 2014-07-14 22:07:43

Re: High pitched whistling noise related to kernel version?

Board footer