26 replies to this topic

[Nick Kaufman]

Ok, this is more out of curiosity because right now, it seems to be working, but god damn it, this is strange.

So, about a month ago, I bought a microsoft wireless mouse. It seemed to be working fine and I was happy.

Then, 3-4 days ago, I opened the case and worked on some stuff.

After that, the mouse would freeze intermittently throughout the day.

I moved its dongle to a new USB port, uninstalled and reinstalled the software, no dice.

Then after some googling, I came across the idea that case fans and fluorescent lights can interfere with mouse signal.

Lo and behold, one of the things I ve done on the case was chaning my front case fan with an old one I had with LED lights.

I unplugged the fan and the mouse worked properly for many hours.

I then tried a new case fan and I got the mouse freezing but not as much as before.

I started powering the case fan through a power cable instead of the 3-pinned header and the freezing went away.

So explain this to me:

How can an led-case fan or a 3-pin header cause freezing to a wireless mouse plugged in the back of the case? It's dumbfounding to me.

[jayhoz]

Electronics by nature produce electromagnetic radiation. Your fan must be producing a wavelength similar to that used by the wireless mouse.

[Blacken]

Nick: which 3-pin header did you plug it into? Are you sure you plugged it back into the same one? Many boards have more than one, and the difference can sometimes be subtle. I've seen boards where plugging a case fan into the wrong header can cause the board to raise a ton of interrupts (which will screw with your USB devices). Board model would also probably help.

Electronics by nature produce electromagnetic radiation. Your fan must be producing a wavelength similar to that used by the wireless mouse.

Must, huh? Must?

Do you know what the "electromagnetic radiation" produced by LEDs is? (I'll give you a hint: the "LE" part of LED is the overwhelming, overwhelming majority of it, and LEDs don't draw enough current to possibly cause a problem with modern electronics, ever. Math is your friend.)

[mt8thsw9th]

My parents had a stationary bike that would cause their 1970s-era home stereo system to increase its volume when you'd apply resistance to the wheel. That was always fun.

[zenter]

Could this be an old IRQ thing? Am I wrong in presuming that LED lights pull power via internal USB plug on the board?

[Blacken]

Could this be an old IRQ thing? Am I wrong in presuming that LED lights pull power via internal USB plug on the board?

IRQ yes, USB probably not. I've never seen a three-pin header that exposed itself as USB, but it's certainly possible. More likely is that he plugged the fan into a header that normally accepts "smart" fans that relay information about themselves. His fan likely isn't a smart fan, so it may be wired to twig the interrupt handler for the fan header over and over and over.

[jayhoz]

Nick: which 3-pin header did you plug it into? Are you sure you plugged it back into the same one? Many boards have more than one, and the difference can sometimes be subtle. I've seen boards where plugging a case fan into the wrong header can cause the board to raise a ton of interrupts (which will screw with your USB devices). Board model would also probably help.



Must, huh? Must?

Do you know what the "electromagnetic radiation" produced by LEDs is? (I'll give you a hint: the "LE" part of LED is the overwhelming, overwhelming majority of it, and LEDs don't draw enough current to possibly cause a problem with modern electronics, ever. Math is your friend.)

Why are you focusing on the LED and not the mechanical fan?

[zenter]

IRQ yes, USB probably not. I've never seen a three-pin header that exposed itself as USB, but it's certainly possible. More likely is that he plugged the fan into a header that normally accepts "smart" fans that relay information about themselves. His fan likely isn't a smart fan, so it may be wired to twig the interrupt handler for the fan header over and over and over.

IRQs used to consistently ruin me. They still mess me up occasionally.

[zenter]

Why are you focusing on the LED and not the mechanical fan?

Because the net EM effect of replacing one fan with another is essentially zero.

[smastroyin]

If an LED produces mostly light, then how does he get cooling? Please do that math and show your work.

Isn't it just a tiny bit possible that the actual source of the EMF interference is the motor within that fan, and that for whatever reason, the fan with the fancy diodes emits different EMF than his other fans?

[Blacken]

Isn't it just a tiny bit possible that the actual source of the EMF interference is the motor within that fan, and that for whatever reason, the fan with the fancy diodes emits different EMF than his other fans?

Possible as in p > 0, yes. Possible as in "is worth considering, ever," no, because something something inverse square law something something. (EDIT: And the amount of power necessary to get into the bands a USB device transmits at--I mean, fucking seriously? The fan is emitting radio frequencies as leakage? Come on. Math.)

Meanwhile, what I am describing is a known problem among DIYers. Interrupt devices fight for contention all the time. This was why new hardware in the DOS era sucked to install.

Edited by Blacken, 16 November 2012 - 03:09 PM.

[zenter]

Possible as in p > 0, yes. Possible as in "is worth considering, ever," no, because something something inverse square law something something.

Meanwhile, what I am describing is a known problem among DIYers. Interrupt devices fight for contention all the time. This was why new hardware in the DOS era sucked to install.

What do I win for being right about the IRQ/interrupt thing? An LED case fan?

[Nick Kaufman]

Nick: which 3-pin header did you plug it into? Are you sure you plugged it back into the same one? Many boards have more than one, and the difference can sometimes be subtle. I've seen boards where plugging a case fan into the wrong header can cause the board to raise a ton of interrupts (which will screw with your USB devices). Board model would also probably help.

The old fan was connected to a power cable ( i don't know how that particular plug is called)

There is only one 3-pin header in that part of the motherboard.

As an experiment, after I wrote the OP, I plugged the Led case fan into the power cable. My mouse is working properly and the cursor doesn't freeze.

So, it's probably the 3 pin header.* Why?

* Let's also say there's a 5-10% chance that it's something else and I haven't had my problem solved but I am experiencing a temporary remission.

Edited by Nick Kaufman, 16 November 2012 - 03:12 PM.

[Nick Kaufman]

Oh and this is my motherboard:

http://www.newegg.co...N82E16813128400

The header in question is the one behind the sata ports:

Edited by Nick Kaufman, 16 November 2012 - 03:19 PM.

[Blacken]

Short answer: some, or all of your fan headers may be responsive to pulse-width modulation (PWM) to allow for variable fan speed. This is done by pulsing a pin on the fan header. That pulse fires a CPU interrupt (unless the board has a separate fan control, which it may or may not). A fan that isn't PWM-friendly might just leave that pin hot at all times, causing CPU interrupts to pile up. USB input devices communicate with the CPU with interrupts as well, and are probably getting drowned out (the CPU can only handle one interrupt at a time, and it is possible by design to have some interrupts dropped on the floor under heavy load).

[Blacken]

It's hard to tell on that picture - is the header a 3-pin as well? (There are 3-pin PWM fan headers, but I've only ever used 4.)

Reminds me why I'm glad I don't build computers anymore.

[zenter]

So, it's probably the 3 pin header.* Why?

Speaking as a non-computer scientist, I can only use layman's terms. Most headers have very basic two-way communication - there's some very talk between the device and the motherboard - things like voltage, etc. - via "interrupts". Those LED fans send more information to the board and if the board is incapable of understanding it, it causes issues. Things like USB, etc., also use a similar process to talk to the motherboard. OTOH, when you plug the fan into the power supply, it's pretty much all one-way "communication" (power), so none of the feedback affects the motherboard.

Once upon a time, I knew more about things like this - northbridge and southbridge are tickling my brain - but in essence, if a peripheral is "smarter" than the motherboard, things can go awry.

Edited by zenter, 16 November 2012 - 03:25 PM.

[Nick Kaufman]

Fuck me. Fuck me. I am never ceased to be amazed by the way things can go wrong. And in a way I am quite fortunate to solve this in a couple of days. This had the hallmarks for months of frustration. Thanks for the explanation Blacken and Zenter.

But basically, the curious part about this becomes why Gigabyte would put a PWM header at a spot where most people would connect a regular fan.

Edited by Nick Kaufman, 16 November 2012 - 03:27 PM.

[Blacken]

That part is weird. Especially because most PWM fans are four-pin, AFAIK (never used them myself).

[Nick Kaufman]

The four pin header rung a bell, so I would say that chances are the 3 pin header is not a PWN fan. I will check it later.

[Myt1]

Possible as in p > 0, yes. Possible as in "is worth considering, ever," no, because something something inverse square law something something. (EDIT: And the amount of power necessary to get into the bands a USB device transmits at--I mean, fucking seriously? The fan is emitting radio frequencies as leakage? Come on. Math.)

Meanwhile, what I am describing is a known problem among DIYers. Interrupt devices fight for contention all the time. This was why new hardware in the DOS era sucked to install.

I have no idea what anything in this thread means, but I am laughing my ass off in my office.

Dude, you must have faced an absolute holocaust of fists to the face growing up.

[kneemoe]

I have no idea what anything in this thread means, but I am laughing my ass off in my office.

Dude, you must have faced an absolute holocaust of fists to the face growing up.

Says the guy who would know

[smastroyin]

Nick's described problem is not necessarily with his USB. It's with his WIRELESS mouse. You know, wireless as in radio frequency. Wireless with low power because why waste power when operating a device with a battery. And by the same inverse square law that you feel so snarky about, the signal from the mouse (transmitter) to the dongle (receiver) necessarily gets weaker as it gets closer to the dongle, and what do you know, closer to the fan or other possible sources of EMI in the case. So yes, if the fan is emitting EMF in the radio frequency range then yes it is actually possible that the interference can happen. It is, as they say, a possible explanation for Nick's troubles.

As to the odds that this is the problem given the information we have. It's not likely but we also have evidence that other devices have worked and you know, I get 414000 hits on a simple google search of "wireless mouse EMI" so it actually isn't so rare as you think, and Nick basically did also ask the question of how the fan change could possibly interfere with his mouse.

How can an led-case fan or a 3-pin header cause freezing to a wireless mouse plugged in the back of the case? It's dumbfounding to me.

Perhaps it's the word "or" that threw you for such a loop?

Now, I realize, jayhoz used the word "must" which is certainly grounds for crucifixion from the knowitall gallery, though since we tend to speak colloquially here I would use contextual clues to suggest that he wasn't actually ruling out other possibilities, merely pointing out that given the information presented, it seemed a good possibility. You know, like, when Myt1 says

Dude, you must have faced an absolute holocaust of fists to the face growing up

he's not actually saying that it is certain you've been hit in the face.

[jayhoz]

Unless my Super Awesome Computer Dork badge for cracking my computer case open a couple times got lost in the mail I have no delusions of being an expert. My suggestion was just that, and the use of the word must was ill advised given all the cheerful and supportive techies willing to pass on knowledge without so much as a hint of condescension we have in our midst.

Edited by jayhoz, 16 November 2012 - 11:06 PM.

[Blacken]

Nick's described problem is not necessarily with his USB. It's with his WIRELESS mouse. You know, wireless as in radio frequency. Wireless with low power because why waste power when operating a device with a battery. And by the same inverse square law that you feel so snarky about, the signal from the mouse (transmitter) to the dongle (receiver) necessarily gets weaker as it gets closer to the dongle, and what do you know, closer to the fan or other possible sources of EMI in the case.

Well, this is surely a plausible explanation to retrofit jayhoz's aggressively uninformed but definitively stated comment into something that makes OH WAIT

I then tried a new case fan and I got the mouse freezing but not as much as before.

I started powering the case fan through a power cable instead of the 3-pinned header and the freezing went away.

Apparently motherboard pins have special electricity in them. Who knew?

Edited by Blacken, 17 November 2012 - 12:29 AM.

[Nick Kaufman]

Just to makes sure that we ve found an answer. I checked the manual and the 3-pin header is for a PCH fan not a PWM fan. Does this make a difference?

www.manualowl.com/m/Gigabyte/GA-P55-UD4P/Manual/207519

Edited by Nick Kaufman, 17 November 2012 - 01:46 AM.

[Blacken]

"PCH" is the name of the fan. PWM is how the system modulates the fan speed. I've never wired up that header so I can't say for certain, but I assume it's for (optionally) cooling the Platform Controller Hub--the facility on the motherboard for talking to audio, USB, and other low-bandwidth peripherals. It's the descendant of the southbridge chip, renamed after the northbridge, which used to talk to RAM and, at the time, the AGP graphics bus. In some configurations that chip can run fairly hot, which can affect performance, but to my knowledge I've never encountered a system that required it.