Issue with Noctua fans ramping up to 100% at certain temperatures

[Jax765]

Motherboard - Gigabyte Z170XP-SLI
CPU - i7 6700K OC'd to 4.4GHz
Cooler - Noctua NH-D15S w/single 140mm fan
Front Case Fan - Generic 140mm fan
Rear Case Fan - Noctua NF-A12x25 120mm fan
OS - Windows 10

I have an issue where the Noctua fans ramp up to 100% in certain circumstances regardless of the fan profile selected in the BIOS. The cpu fan ramps up when temperatures reach around 66 degrees, while the rear case fan occasionally ramps up for a period of time while playing a game (even if temps are below 60). The CPU fan is plugged into the CPU_FAN header, rear case fan is plugged into SYS_FAN1, and the front case fan is plugged into SYS_FAN3. Sometimes, the rear case fan has reached peaks of 2600rpm, even though the supposed max speed is 2000rpm. Where it gets especially annoying is when the CPU temps spike close to 70 degrees, and the fans ramp up to max speed before subsequently ramping back down. This can happen consecutively with temperature spikes when I'm, for instance, letting a 3D scene render or playing a CPU-heavy game like Red Dead Redemption 2.

Which leads to the question, should I update my BIOS to see if that fixes this issue? I'm currently on f6, the latest version being f22f. It mentions that the BIOS can't be downgraded after the fact back to the current version, and I'm wary of the risks of updating, especially since there's no other noticeable issues apart from this fan curve thing. And if updating the BIOS is the right move, what's the best method? Using Qflash?

The strange thing is that this ramp up problem only started occurring when I switched out the stock fans on my old Corsair h80 for 2 Noctua NF-A12x25s, since they were making grinding noises on startup. This carried over after switching the h80 out for the D15S, though it at least happens less frequently due to the improved cooling performance. Could it be a case of the current BIOS not having proper support for the Noctua fans?

Screenshot of the BIOS settings for reference:

 

[Meddling-Monk]

Just to confirm, all your fans are PWM 4 pin variety right?

The BIOS not having proper support for Noctua fans would be very unlikely.

More likely is that Gigabyte cheaped out on the fan controller and stuck some weird hybrid controller onto that board, or it needs an update via BIOS to get it to read proper sensor readings.

A quick gander at the manual gives me the impression that Gigabyte is using the same fan controls as from their Z77 chipset days, a few years earlier than yours those were (2012), and those were seriously bad at controlling the fans aswell. They either never rose as required (remaining far too low), or just jumped to max too quickly, or used a completely different sensor that was always too cold or too hot (causing attached fans to stay full or off). They supposedly worked with standard 3 Pin fans (voltage control) and 4 pin PWM fans, so likely worked off of voltage control variation rather than true PWM control. Never really found a decent way to sort that out other than trying Speedfan (which is very old now and very possibly won't work in your case) and now using an proper PWM fan controller for control.

And one final look, yep. Gigabyte have 4 pins on their non CPU fan headers, but they aren't true PWM, they're voltage control. Only the CPU is a real PWM and I think the controller is still actually using Voltage control as it did on the Z77 boards they had. That's probably why your Noctua A12x25 is playing funny, as it's expecting a PWM signal to control its speeds and not the voltage variation.

But, that might explain the funniness with the fan speeds ramping, but not the 2600rpm you've seen. Which I think is an error or bug. Which possibly needs a BIOS update to get working properly.

Looking at the BIOS available, I don't see anything about improvements to sensors and fan control, so can't say whether an update will help or not. But if I was dead set about trying to get the fans under control, that'll be something I'd try updating first, but that's just me.

 

[bremen1874]

No options in the BIOS to reconfigure the fan curves?

My current x570 board was really annoying at stock because it was ramping the fans at every little change in temperature. With the fan curves manually adjusted, it's basically silent except for when it is running with a sustained load.

 

[Meddling-Monk]

No options in the BIOS to reconfigure the fan curves?

My current x570 board was really annoying at stock because it was ramping the fans at every little change in temperature. With the fan curves manually adjusted, it's basically silent except for when it is running with a sustained load.

Looking at the manual for it, there's no manually controllable/configurable curve per-se like what we normally expect. Just a divider that's applied to simulate a curve from the temperature its using. It's really bad.

 

[Jax765]

This is the fan in the rear of the case.
https://www.amazon.co.uk/gp/product/B07C5VG64V/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1
The CPU fan is the one included with the NH-D15S.

As I mentioned, this was never an issue prior to installing the Noctua fans. So presumably it's the motherboard not working well with 4-pin fans. I just checked the old stock H80 fans that I replaced a while back, and they are indeed 3-pins. Which might explain why the Noctua fans are now having this issue, since they're 4 pins. The front case fan is also a 3 pin, which explains why that fan doesn't have the ramp up issue.

So is it likely that this 3-pin/4-pin thing also applies to the cpu header? Since the fan ramping up happens far more frequently with the cpu fan than the rear case fan.

 

[Meddling-Monk]

Yeah, thanks for confirming. In which case, it's not just Noctua fans, there's a variety of fans that use proper PWM (4 pin) controls that'll experience similar issues on a variety of points. Some will be OK with the Voltage Control your motherboard fan controller users, but a fair few will experience hiccups with a Voltage Control setup. You may experience clicking from fans as they hover between voltages that cause them to stall and normal operation (I've experienced that, that's what lead me to get a multi-channel PWM fan controller), and a variety of other issues.

 

[Jax765]

Can't say I've experienced clicking or stalling.

So with regards to solving this, the long-term plan is to build a new pc next year with a new motherboard. Until then, would it be a good idea to swap out the 140mm fan on the NH-D15S with either of these?
https://www.amazon.co.uk/Noctua-NF-A14-ULN-Ultra-Silent/dp/B00AA89BN4
https://www.amazon.co.uk/Noctua-NF-A14-FLX-Premium-Quiet/dp/B00AA875NC

 

[Meddling-Monk]

The clicking from the fan part way stalling only happens when it tries to lower the voltage down too much. Usually around 45%.

If you're going to be upgrading soon to a new system, I'd just save the money and bare with the noise for now instead. Noctua fans are not cheap and you won't be reusing the 3pin fans going forward in all likelihood, meaning you're paying for something now which has no real function in the near future. Especially since those fans are likely going to be lower perfomance than the fan on your CPU cooler right now. So it's like paying up to slow your CPU (or any other) fan down, and because a new system won't be reusing these fans because of noise/performance, their lifespan of use is essentially less than 13 months (if you get a new system by end of next year or earlier). Not really worthwhile cost wise, but if you're OK with it. I don't see why not.

In which case, the fan you'd be after is the A14 FLX model for your CPU cooler. Not the ULN version.

 

[Jax765]

It's going to be a good long while before I upgrade, and the ramp up is incredibly annoying. Been putting up with it since I first installed the Noctua fans last year, and at least now I know the actual reason. So I might just go ahead and get the FLX to fix the issue until then.

On the topic of my upgrade btw, the plan is to replace my case, a Carbide 300R, with a Lancool II Mesh performance. I've got 2 questions regarding it:

1) Is the Lancool shorter than the Carbide depth-wise? Right now, the Carbide just about fits on my desk, with the front panel hanging off. Ideally, the new case would be a bit shorter to fit better. I've seen conflicting numbers between the official site and various retailers/review articles.

2) What sort of fan setup would be ideal with that case combined with the NH-D15S? 2 front 140mm fans, 1 140mm NH-D15S fan, 1 120mm rear case fan?

 

[Jax765]

Got a reply from Gigabyte on the issue, which at least provides a definitive statement on the fan headers.

 

[Meddling-Monk]

1) Is the Lancool shorter than the Carbide depth-wise? Right now, the Carbide just about fits on my desk, with the front panel hanging off. Ideally, the new case would be a bit shorter to fit better. I've seen conflicting numbers between the official site and various retailers/review articles.

Looking at the official dimensions for both, the Carbide 300r should be longer than the Lancool II Mesh, so it should still sit on your desk fine. Note however, sometimes the specifications listed aren't accurate, so I can't guarantee those numbers will transalte accurately for you; there's a thread a bit further down, someone had an issue with a CPU cooler and their case not fitting and judging from their response, they checked the numbers first, and still in the end didn't quite match up right. So yeah, the numbers suggest the Lancool is shorter is all I can say with certainty.

2) What sort of fan setup would be ideal with that case combined with the NH-D15S? 2 front 140mm fans, 1 140mm NH-D15S fan, 1 120mm rear case fan?

I think that's the classic setup and in theory should be ideal. But what you'll often find, is that sometimes, certain configurations can get better results than expected, depending on what particular goals you're after. But that'll need time and effort testing, something which you may not want to do.

Got a reply from Gigabyte on the issue, which at least provides a definitive statement on the fan headers.

Not surprised, their manual for your board already showed that the 4 pin fan header for the CPU had PWM control (12v going down the power line, and PWM on the 4th line that controls the speed), whilst the other fan headers had VCC (Voltage Control) instead on the power line and nothing on the 4th line that normally controls speed.

 

[Mcthommo]

Might be complete different but I have a noctua fan on a Corsair cp and and have to set the fan port as a 3 pin rather than 4 pin otherwise I get massive humming noise