I’m using Mint 20.1 – Cinnamon.
The main reason for replacing the CPU fan is, one blade has a hairline crack. It’s not flopping around, but I’m concerned about tempting fate. Cooling -wise, it’s doing well for now.

I'm not sure if it could be repaired (reliably), and not be out of balance. Maybe doing the "same thing" to a blade opposite the repaired one.

I bought a replacement, "Arctic F8 PWM PST", (80 mm, 4 wire - just like old fan) at reputable, local repair & sales shop. I'm now wondering if the fan is a suitable replacement for a Quad Core CPU?
Some specs: https://www.amazon.com/ARCTIC-F8-PWM...002KTVFPI?th=1

Plugging it into the mobo "CPU Fan" 4 pin socket (only 4 pin fan socket on MOBO), boot up, immediately says, "CPU cooling fan failed. Shutting system down in 4 seconds." And it does.

I read a lot about using lm-sensors, "sensors-detect", PWM settings & lots of related info.
I know enough now to be dangerous.

1st issue is help deciding if the Arctic fan is a suitable replacement, spec - wise.

I can only find some specs on old fan, "Delta / Cooler Master AUB0812HV, 80mm, 4 pin (no other characters in the model #). I've seen specs it has MAX RPM around 3600. I haven’t found the MIN RPMs anywhere. That may? be part of the problem. The Arctic may be defective - not sending / receiving data, or its current 300 RPMs are perceived as failing / failed.

Either installed or testing on workbench, the old fan's RPMs are much higher than the Arctic fan. I can only observe the relative speed of each fan - on a bench, applying 12V to -Blk & +Red wires. Arctic fan's specs are 300 - 2000 RPMs. On the bench or installed - for a FEW seconds before system shuts down, it spins VERY slowly (300 RPMs?) vs. the old fan.

Testing the old fan w/ 12V, it spins several times faster & according to "sensors" or Psensor, it starts at around either ~ 1650 or 1070 RPMs (depending on which Fan (1 or 2)is CPU fan. Only 2 show RPMs - assume the 2 plugged into mobo. Two other fans detected don't show RPMs. I'm "guessing" the 80 mm CPU fan would be the higher RPM vs, the other 120 mm case fan(s). The RPMs of the 2 don't vary much. Nothing I've seen in any files identify any of the fans.

When applying 12V to variable speed fans - to Blk & Red wires (or neg & pos, if all black wires), I don't know if the resulting speed is its MIN, MAX RPM (or something else). Seems clear, the Arctic has considerably lower (300) MIN RPM, and AFAIK, a lower (2000) MAX RPM than the old fan.

So, if I manage to find an 80mm PWM fan with RPM range closer to the old fan, would that greatly reduce / eliminate problems of replacing the CPU fan?

One of the four wires is feedback signal for reporting RPM. The actual RPM is not likely the issue, but rather the lack of any feedback signal. The BIOS must not be seeing any signal from that wire to cause the shutdown.

Note, as mrmazda mentions, this is a firmware (BIOS/UEFI) issue, not an linux issue. Might be an option to turn off the warning - more likely not probably.

As pointed out, this is a BIOS feature - it may be defeatable ('CPU Fan Fail Warning' or some permutation is usually the name of the setting), but this sounds like an OEM machine if it is this aggressively enforcing fan fail (in my experience usually only OEM boxes are this aggressive - most white box or otherwise self-built machines will just complain but won't shut down). Likely the BIOS expects the RPM range of the fan the machine shipped with, and sees the 'low' RPM as indicative of a failure (this is in error, but welcome to the world of OEM boxes).

Things I would consider:
- You said the original cooler uses an 80mm fan, that's a standard size, is it per chance using a standard mount such that you can just replace the fan itself?

- Can you actually disable that feature within the BIOS?

- Whether or not the Arctic heatsink is appropriate needs more information beyond 'a quad core CPU' - what specific CPU do you have? Can you share more information about the machine generally as well?

I appreciate the input. Uh-Oh. I was afraid of no consensus on this issue. Keep in mind, a BIOS from 2009 isn't like from 2019 (forgot when updated, but was last HP offered).
Before I type more, can others just give their experience replacing CPU FANS (don't think it matters if fan's separate, or fan & heat sink come together).

When you plugged fan in the "CPU Fan" socket (assuming were speed controlled (PWM) by the mobo / CPU / BIOS), did you just start surfing? Or did you have to [B]use something to "set up," or get it to recognize the new fan or other tinkering, to get fan to operate at suitable speeds?
My Mint 20.1 has LOTs of data in files, for fan PWM values, temperature, etc. I assume it's automatically set up.
Bottom line, did your system immediately recognize the fan & continue running, long enough to allow tweaking min / max RPMs (if needed)?

Note: This Arctic fan (& old Cooler Master) are JUST separate fans (not a heat sink), that screws onto a plastic "frame" around the heat sink. The heat sink came w/ the computer & HP chose the old fan model to go w/ it. Any 80 mm fan w/ PWM speed control (& suitable Min/Max RPMs) should fit. The BIOS / CPU / mobo "liking" the fan is different matter. There's a very slight recess in plastic mounting frame, to line up the fan for screws. New fan fit perfectly.

But I have bought new items (almost any category), bad out of the box. It wouldn't surprise me if the fan was partly broken, or if this fan's electronics just aren't compatible w/ (something).

Could be. Only realistic way to check, is try another new fan (same model or one w/ higher RPMs). If another identical fan worked, might show some bad electronics in this new one, rather than low RPMs being the issue.
I assume it's the BIOS, though there's NOTHING mentioned in this BIOS about "fans" at all. I wonder if the CPU / mobo could be showing the error? If the msg is from BIOS, it's hidden from me.

@syg00 - there's nothing in BIOS on fans. "Not a Linux issue" - maybe - not sure, but users can still ask for help, can't they (on hardware)? There are LOTS of data, settings in files about PWM & temp. values for different fans. That wasn't generated by BIOS. Maybe by lm-sensors, or other tools that were pre-installed.

I wonder if temporarily switching another fan to the CPU heat sink would show if the new fan wasn't reporting any speed signal? Have to be a 3 pin - & > 80 mm. I think? a 3 pin fan (if it will fit the 4 pin socket & leave 1 pin hanging out) would report the fan RPMs? The other fan reporting RPMs is a 3 pin. I'd zip tie it on, to test.
Thanks for suggestions.

I don't see why there shouldn't be consensus - 'try a new fan' seems like the best next step.

Been working on PCs for over 20 years - it 'just works' unless the RPM range is not acceptable to whatever the BIOS has programmed, which as I said is typical for OEM machines that use borked BIOSes. Adjusting 'fan speed curves' for noise mitigation is secondary, but there is *nothing* that needs to be 'installed' or 'configured' or 'programmed' in the OS here (and it doesn't matter if you're using linux, windows, macOS, etc - that's why folks have said 'this isn't a linux issue' is my guess); this isn't a complex device with drivers.

This is good information to have (it wasn't clear from your first post). I would just try a higher RPM 80mm fan - as far as whether or not the new one is suitable overall, you haven't specified what processor you actually have, or what temperatures you had observed before/after. Very generally, going to a lower RPM fan means less airflow and less static pressure, which will reduce cooling performance on the same heatsink - but 'is that a problem' is contextually dependant. In other words, it may run the CPU a few degrees warmer, but if it's still not throttling or overheating, that's perfectly fine. Different fan designs also have different airflow/SP curves vs their RPM, so it's also possible that fan A at 3000 RPM is 'just as good' as fan B at 4000 RPM, or whatever, but the HP BIOS won't know (or care) about that because it has hardcoded values locked to specific HP parts.

It's possible, but if the fan is spinning and isn't visibly broken, it's more likely that the RPM is just lower than the HP BIOS will accept. Yes I've seen cheap Chinese no-name fans with fake RPM or PWM wires that terminate to nothing, and yes I've seen stuff damaged in shipping, but Arctic is not what I'd consider a 'no-name brand' so unless you can visibly see damage to a connector, solder joint, etc it likely isn't broken. Again, this 'issue' is very common for OEM boxes (like HP, Dell, Acer, etc) when the user goes and tries to replace hardware at a later date - they aren't really designed to be mended.

And I think there's your consensous - try another fan. I would skip replacing this Arctic like-for-like and get something with higher maximum RPM, and if that works, return the Arctic or use it elsewhere.

Very typical of a Dell/HP/etc OEM BIOS - it's not meant to be serviced after all.

Usually a 3-pin fan on a 4-pin header will just run at full speed all the time (at 12V), the BIOS may or may not like this (again, thank HP for this 'feature'). It won't hurt anything to try.

Overall I don't think there's anything really 'big' going on here - OEM BIOSes are always a pain when it comes time to service the machine. If this system is really from 2009 (or earlier) it may also just be worth considering a replacement (I'm not saying 'throw it out because the fan broke' I'm just saying 'this would make it a 13+ year old computer, that's pretty old') - you haven't provided any other specs but I'm guessing it's pretty prehistoric and may not be patchable against modern vulnerabilities (like Spectre, Meltdown, etc) among other things. If you don't need a whiz-bang modern GPU, modern computing power is seriously cheap. My advice would be to try another fan, with higher rated maximum RPM, and see if it accepts that and lets the system boot. Conversely, see if you can disable 'halt on - all errors' from the BIOS and if that will let you ignore this error with the new fan, and then keep an eye on CPU tempterature under a heavy load and make sure it isn't reaching dangerous levels (usually around 100* C is 'too high' but we don't know anything about this system so it's hard to give you specific guidance).

+1 to everything obobskivich wrote. In 30+ years working on PC compatibles I've never had a shutdown at startup due to a too slow CPU fan speed. The worst that happened to me is complaining about no fan detected that went away on striking the anykey.

Thanks for all the input. It should be helpful.
Because I couldn't find right away a different 80mm, 4 pin, PWM fan (that I'd trust), I tried my idea of plugging a 3 pin, PWM case fan into the CPU fan 4 pin socket.

This did prevent BIOS from saying the CPU cooling fan had failed. The test replacement fan has much higher MIN RPMs than the new Arctic. So the Arctic fan probably had problems on its PCB that didn't send a signal that a fan was plugged in & turning. I did find a discussion about the same Arctic fan.

Someone claimed it didn't start spinning, until the OS login screen appeared (I think). They claimed that caused some BIOS to send a fan failed message. I don't know if that's true, or if it is, is delayed start by design, or a bug in the fan's circuitry?

NOTE: IF the 3-pin "test" fan has the same order of wires (if they're colored) as the fan being temporarily replaced, then everything will work. Hopefully the test fan has the same kind of connector, which will guide you on which one of the 4 pins will be unused.

Since the Arctic F8 only had a MAX RPM of 2000, & it appeared the old CPU fan started off @ ~ 1650 RPM, that's near the max RPM of this Arctic fan. I wanted to find one w/ a (bit) higher low speed & higher max - just in case. So I didn't try a another Arctic fan - same model.

Anyway, ordered a Noctua fan with a (bit) higher MIN RPM & higher top end than the Arctic F8, plus the Noctua's noise specs are lower. Sometimes specs are inaccurate. For now, have a used 120mm fan fixed to the plastic fan-mounting frame on the CPU heat sink. I salvaged different hardware from some desktop PCs someone set on the curb. It's turning much faster than the orig. CPU fan, but it's gettin er done.

Hi,

You might find this document informational; https://www.silentpcreview.com/forum...ic.php?t=36548
Dated but applicable!
Plus the specs here;https://www.intel.com/content/www/us...sort=title:asc

You can dig deeper if you want but the 3 pin substitution for a 4 pin will result in full on for the 3 pin. So if you want to hear the fan noise when using a 3 pin then your CPU temps should be OK.
Hope this helps!
Have fun & enjoy Gnu/Linux.

1 members found this post helpful.

@ onebuck - Thanks for the links with details. Though some of these fans have -blk, +red, yel & blu. Some I've seen substitute a white wire. Maybe quite old fans. Now, many new fans have all black wires, "so it looks better / matches other wiring in the case." That's what some fan brands listed under features.

No, I never intended to permanently use a 3 pin (120mm) fan on an (80mm) heat sink's "fan mounting frame."
That was some fine engineering, there. The new Noctua is almost silent @ 800 - 850 RPMs, under light CPU load.

I couldn't risk using the orig. fan & the cracked blade fall off, or sag enough to stop turning.
Hopefully, the system would shut down before the CPU fried.

Long ago, had a P3 with the Intel heat sink + fan combo. It had 2 odd, plastic clips (no screws) holding the fan on the heat sink, that broke on their own in < 3 yrs. Good design.
Don't remember if one or both clips broke, but the BIOS on a good mobo didn't shut the system down fast enough. Much diff technology then.

It was still under warranty, so Intel sent me a new CPU, heatsink + fan. Once a CPU fan falls partly away, the CPU may fry very fast.

https://www.youtube.com/watch?v=NxNUK3U73SI

Note this video is ancient (probably turns 20 this year), but see how the Pentium 4 suffers no damage in this scenario? (note that nobody recommends testing this with hardware you care about) Newer CPUs (from the Pentium 4 forward) will behave in this manner, as opposed to cooking off like Pentium 3 (and similarly old AMD Athlon) because of onboard thermal protections. Again I'm not saying 'run it like this, it will be fine' - instead just trying to assuage fears that the CPU will imminently fail with the fan out. That doesn't mean OEMs have gotten the message, so you still have hold-over (from the last century) things like 'fan fail warnings' or 'fan fail halts' because at one time it really did mean catastrophic failure (which could involve an actual fire - look at the temperatures those CPUs reach before dying!) was possible.