Notebook fans running at full speed

[ebastler]

I had to partially disassemble my notebook computer, an ASUS Zenbook UX303LB, to replace the LCD back cover. (Which contains a plastic frame to which the hinges are attached, and which had crumbled after 5 years of use.) Put it back together and it works nicely, except for a peculiar sporadic faut it has now developed: The fans sometimes decide to switch to full speed for no apparent reason.

• In normal operation, the fan control does work correctly. Fans start very quietly, and get gradually faster if the CPU load increases, then slow down again when the CPU has less load.
• When the problem occurs, both fans (for CPU and GPU cooling) turn on at full speed, and always both at once.
• While this happens, CPU load is low according to the Windows task manager, as are all temperatures (CPU, GPU, SSD) according to the HWMonitor utility.
• The computer is running Windows 10, but I don't think this is a Windows issue. When the system decides that it is time to let the fans blast, they do so as soon as I turn the computer on, and continue if I reboot -- so the issue seems to be on the BIOS or hardware level.
  
• The transition from normal, low-speed operation to full blast can occur in the middle of using the computer, or I might put the machine into hibernate mode while the fans are quiet and when I wake it up later, the fans go full speed right away. I don't think I have seen it revert to normal in mid-operation; it takes putting the machine to sleep and waking it up again to get things back to normal. (But there is no guarantee that it will come back quietly then, sometimes it "remembers" the full speed state. Have not figured out any pattern.)

I struggle to even come up with a hypothesis what is causing this. When all temperature sensors read "harmless" values, what would cause the fan control to switch to full blast? Are there other sensors which HWMonitor does not display? Why would they affect both fans at once? What else goes into the control loop for the fan speed?

Also -- what could have gone wrong in putting the machine back together? I had to disconnect the LCD, webcam and Wifi antenna connections and transplant those components into the new display lid. And, of course, unscrew and reattach the hinges. All screws sit in brass thread inserts, without danger of piercing anything, and I checked for pinched cables. No active components were replaced, and nothing that has an obvious connection to the fan control circuit.

I am stumped... Thanks for any ideas you might have!

[Le_Bassiste]

i had a similar fault on a desktop mobo (asus x79 iirc) where the temps were fine but the fans went full speed. fan controller had a bad soldering joint. hth.

[ebastler]

Thank you! A cold solder joint is a plausible cause. I might have "triggered" a pre-existing poor joint by putting mechanical strain on the main board when I reattached the plugs and screws.

I could not identify the fan-related circuitry on the main board so far. Given the fact that the problem affects both fans at the same time, I'm assuming there must be some shared component(s) to blame. But the two fans' connectors to the main board are spaced a fair distance apart, and there are quite a few small ICs in that larger neighborhood...

Inspection under a stereo microscope showed reasonably-looking solder joints throughout. Also, I could not trigger the problem by poking at various chips with a plastic screwdriver. I will wait for the issue to recur and then try the opposite -- can I stop it by prodding the right chip?
Edit: Should stock up on some cold spray, I guess...

[Rasz]

on laptops fans are usually controlled by KBC, ite IT8585VG here

[ebastler]

on laptops fans are usually controlled by KBC, ite IT8585VG here

Thanks, I didn't know that! But unfortunately I cannot identify a KBC chip on my mainboard: The only chips with plausible integration density (package size and pin count comparable to the IT8585VG) turn out to be the memory controller, USB controller, and battery charger. And there is an unpopulated footprint of plausible size right next to the connector for the keyboard's flex cable -- but where is the keyboard controller?!

This problem continues to puzzle (and annoy) me: I had the computer running with its bottom plate removed all day, and it never went into "full fan blast" mode. Couldn't provoke the fault by warming various areas of the main board either. But as soon as I put the bottom back on and sat down on the couch with the computer for 10 minutes, the fans are back at full power -- grmpf... I can't see how the bottom plate would directly put mechanical strain on the board, so what gives?!

I'll wait for the cold spray to be delivered and see whether that helps to locate the problem.

[Rasz]

on laptops fans are usually controlled by KBC, ite IT8585VG here

Thanks, I didn't know that! But unfortunately I cannot identify a KBC chip on my mainboard: The only chips with plausible integration density (package size and pin count comparable to the IT8585VG) turn out to be the memory controller,

no such thing on this laptop, or any computer for over 10 years now

USB controller, and battery charger. And there is an unpopulated footprint of plausible size right next to the connector for the keyboard's flex cable -- but where is the keyboard controller?!

its the only chip with huge ITE on it, lower right corner when looking on the pcb from the top
https://ae01.alicdn.com/kf/HTB1F_NWG1SSBuNjy0Flq6zBpVXap/UX303LB-With-i5-5200-CPU-GT840M-2GB-VRAM-HD4400-4GB-RAM-mainboard-For-ASUS-UX303LB-UX303LN.jpg

boardview https://drive.google.com/file/d/1Ai9y0XaXoBqPOp0CchrBuKlR7h6pmiRI/view
https://drive.google.com/open?id=15I605RzlURiPgdNvJON5L2rycCKwesFz
https://drive.google.com/open?id=1JIq_oDYp_U22F0B8yaaK--knHih41aLD

[ebastler]

Duh... 

I should have realized that there is a lot going on on the other side of the PCB... I guess I just did not want to think about that, since getting access to that side is a pain in the neck: Remove the display hinges, CPU and GPU cooling systems, all connections, and the main board. And no chance of any functional tests while I look at that side of the board; just need to take a stab at resoldering stuff on sight. 

Anyway, thank you for your help! I can obviously use it, since I don't seem to be thinking straight with this repair. Probably I am just too annoyed... This is my "daily driver" notebook, and while it is 5.5 years old already, I loathe the idea of potentially having to replace it only due to a noisy fan which is driving me nuts...

I will still wait for the cold spray before I move any further. Maybe blasting the PCB from the bottom side will tell me something about the fault location, thanks to your picture of the top side.

[ebastler]

Quick update, although a frustating one:

It still seems likely that the fan problem is triggered by (mildly elevated) temperature of the main board, or by mechanical stress on the board: I have only ever seen the fans crank up when the notebook's bottom plate was installed.

But I could not localize the problem using cold spray -- could neither provoke the problem that way, nor stop the rapid fans once they had spontaneously cranked up. And likewise, I could neither provoke nor stop the problem by pushing and prodding the PCB and components in various places.

I then took the whole computer apart to get to the other side of the PCB. Visual inspection was uneventful -- some dust, but no rogue pieces of metal flying around, and the solder joints looked good. Cleaned the PCB and took a stab at reflowing some parts, including the KBC chip (a small BGA) with flux + hot air. Put it all back together, but sadly the fans still show the same behavior.

I am about to give up on this machine -- but hate the idea of replacing a computer which is otherwise perfectly fine, just because its fans are acting up. (But I can't use it in this state; the fans really get on my nerves.) Any other ideas would be most welcome. Thanks!

[jamesglanville]

Slightly left field ideas:

- Can you replace the fan with a quiet version that won't annoy you even at full blast?
- Can you put an inline resistor on the fan to limit the max speed and therefore noise?

The latter is a bit of bold move, but if you're giving up you don't have much to lose.

[Rasz]

I didnt see you trace or measure anything. reasons for fans to go crazy:

- computer loses ability to measure temperature, damaged thermal probe or its connection. For example long time ago Radeons had the tendency of blowing its thermal diodes resulting in working cards with no temp feedback and Fans blasting full speed.
- EC detects some other problem, hmmm maybe erroneous fan feedback?
- CPU asserts PROCHOT#? signaling its throttling, this signal goes to EC THRO_CPU# pin. There is also a connection to PWRLIMIT#_CPU and P_PL_IBAT_THROTTLE_10 suggesting it might be triggered when something is wrong with battery connection/charging?
- in the same vein NCT7717U https://www.nuvoton.com/resource-files/Nuvoton_NCT7717U_Datasheet_V111.pdf asserts ALERT# pin when it detects something weird, and afaik doesnt let go after things return to normal
- GPU has GPU_THERM#_L going to EC
- software problem

try if unplugging one of the fans triggers the problem
setup motherboard outside the case with external monitor and use hair dryer/hotair instead of cold spray. If you cant trigger crazy fans on demand then repair is impossible.

There is always the ghetto way of building ATtiny85 temperature controller driving fans PWM pins directly with two external temp sensors near CPU/GPU.

[ebastler]

That's right, I have not really measured or traced anything. One reason being that I struggled to figure out where to start (thank you for the pointers!). But more importantly, I have not been able to "force" the problem in an opened system.

While I have not tried to plug things back together in a working configuration outside of the case (when I have everything apart this weekend), I have heated the PCB from below while the notebook's bottom lid was removed. I am sure that things got much warmer than in normal operation, with the case closed -- but never was able to trigger the issue.

I think a software (Windows) issue can be ruled out, since sometimes the fans run fast as soon as I switch on the notebook, i.e. under BIOS control. Bad sensor readings from the CPU or GPU can probably be excluded as well, since HWMonitor shows normal values. I can also rule out a charging-related cause, since the problem has occured with and without the power supply connected. I will try to follow your other hints - thank you!

[ebastler]

Slightly left field ideas:

- Can you replace the fan with a quiet version that won't annoy you even at full blast?
- Can you put an inline resistor on the fan to limit the max speed and therefore noise?

The latter is a bit of bold move, but if you're giving up you don't have much to lose.

Finding a quieter fan that fits seems very unlikely, unfortunately. These are custom, low profile radial fans which fit the tight space constraints of the notebook -- see the picture below. Also, the noise I hear seems to stem mainly from the airflow through the radiators at the end of the heat pipes, and maybe through the ventialtion slits in the case. The fans themselves seem well-designed and not excessively noisy.

Limiting the fans' speed via series resistors could be a last resort, although it would obviously put a lot of thermal stress on the CPU and GPU (and/or cause them to throttle) when they really have to work hard. I have toyed with the idea of a manual switch...

[Rasz]

I linked you boardfiles, trace fan connectors signals back to EC checking all the resistors in the way for damage/cracks/marginal solder. Same for the thermal readout lines between GPU/CPU/sensor chip and the EC. FAN_TACH_GPU/FAN0_TACH etc...