MP 1,1-5,1 Mac Pro (4.1/5.1) aftermarket fans WITH automatic SMC control

[p11hlf]

Yes, it works and is currently running.

Read ALL of this thread.

Noctua NF-A12x25 PSU fan for 4,1 & 5,1 cMP

Today I did a test install of the Noctua NFaA12x25 120mm fan for my 4,1>5,1's PSU. The fan comes with a "low noise / LOWER RPM" adapter cable connected to the original PSU fan socket on the motherboard, NOTE 1 : I had to shave off one of this cable's 'raised locater guides" to allow the Nocuta...

forums.macrumors.com

Thanks MIKX, I prefer the method of this thread rather than controlling manually with a PWM controller.

So my question still stands, can you use a 4 pin PWM fan and simply remove the PWM pin.

similar to the table below. Note: pin 2 is swapped to pin 4 and pin 4 will not be plugged in.

Mac Pro output	Pin	Fan input
GND	1	GND
VCC (12V)	2	* empty * was 4 - PWM control signal
Sense/Tacho	3	Sense/Tacho
Control voltage	4	VCC

 

[KeesMacPro]

@amedias:
Quote:"If you're running a pair of 5690s and a big GPU then there's not much you can do under load conditions as it will be generating a lot fo heat. The Noctua will be quieter than the OEM fan but I'd be wary about running a Noctua with a max RPM of 1200 when the OEM fan often spins at 1000 RPM with your workload. You *may* run into a situation where the Noctua is maxed out under extreme load and you have insufficient cooling."


After reading your post and brainstorming for a while I took a closer look at the construction of the case of the Mac Pro.
I agree with you, and lots of others, that the air flow and heat regulation within the case is very well thought and efficient.

From the day I bought this MP 4,1 I removed the DVD drive, and along the way , any part that is not necessary like spinning HDD , obstructing the airflow ,generating more heat and slowing down boottime.
The optical bay , I think, can be seen as a tunnel in front of the PSU fan.
This tunnel should have a surface equal or bigger than the surface of the fan (circle of radius 6cm).
If the surface is smaller, this will lead to resistance , which means an increase of the airflow speed at the bottleneck, or just less throughput.

The motor of the fan is not a AC synchronous type (regulated by the frequency of the AC),neither a PWM (regulated by Pulse Frequency), the speed is regulated by voltage.
Presuming the SMC is not influenced by the current the fan motor draws, depending on the resistance of the airflow, the voltage will be adjusted too.
To put it simple: I think its save to presume that with more resistance in the airflow, the motor draws more current and will heat up more.
The power draw is compared to a big GPU marginal, but the heat might be a negative factor.

Looking at the empty optical bay, the surface is much bigger than the surface of a 120mm fan, which is good.
But the whole dvd door mechanism is definitely not, even with the flaps kept open permanently, there is still lot of material in the way I think.
Last night I “removed” the plastic mechanism with the doors for the DVD- drives completely and glued some stainless steel mesh in the slots.
A quick test showed that now the PSU fan ,depending on the load, is running significantly slower (somewhere between 100- 300 RPM less) SMC set to default for the PSU fan.
Of course, this is on my set up.
I attached a foto of the result,I hope that the purists among us won’t be shocked


BTW I tried to define the sound of my (worn) PSU fan so I ramped it up to 2800 RPM for a few seconds.
The sound is very close to a bomb alarm siren….
I’ll buy a Noctua fan this week !

 

[p11hlf]

Thanks Guys,

Just tested this myself using the noctua PWM fan plugged into the PSU fan header, I swapped pin 2 to pin 4 and removed pin 4 (the PWM signal)

IT WORKS!

this is backed up by noctua post here, and therefore safe.
https://noctua.at/en/can-i-run-4-pin-pwm-fans-on-3-pin-fan-headers

Caveat: the fan RPM shows in Mac fan control but because the range of the fan is 450-2000 rather than 600-2800 the fan shows for a brief second as 0 and then back up to 800..... this is random. When you set the fan speed manually 2000 or above, the RPM shown in Macs fan control only shows as 2000ish

Im gonna swap out my PSU fan now.

 

[KeesMacPro]

I prefer the method of this thread rather than controlling manually with a PWM controller.

Noctua quote:
Can I run 4-pin PWM fans on 3-pin fan headers?
PWM fans come with a 4-pin connectors for fully automatic speed control via your mainboard’s 4-pin PWM fan headers. Please note that 4-pin fans can also be connected to your mainboard’s 3-pin fan headers though. When connected to 3-pin fan headers, the fan will run at full speed (unless the mainboard supports voltage based speed control).



I would have bought a FLX or ULN type, and after reading the first post, install it enjoying the advantage of SMC...
No manual interference, no fixed RPM,no PWM controller, just default setting...
Maybe you can still return it to the seller and buy a 3-pin version?

EDIT:

the fan shows for a brief second as 0 and then back up to 800..... this is random

Sounds to me like the fan is alternating ramping up to 800 RPM and switching off..
Which means voltage regulation doesn't work well for this fan and/or this range of RPM...

 

[amedias]

you can use the PWN fans but I think the 3pin ones are a better option, not least because you don;t have to leave one wire dangling!

I've tested a couple of PWM fans and they've worked fine, but what isn't clear (without lots more testing) is whether they all respond to varying voltage in the same way.

Caveat: the fan RPM shows in Mac fan control but because the range of the fan is 450-2000 rather than 600-2800 the fan shows for a brief second as 0 and then back up to 800..... this is random.

The issue you describe with your fan registering zero, then jumping up to 800 sounds suspiciously like what happens when the fan stalls at the target voltage/RPM of the SMC minimum settings. Normally if you watch the RPM you'll see it still pulsing or wavering but that's fan dependant, some just wont spin until enough voltage applied, others can 'hover' between stalling and not.

The SMC min for the PSU fan in a Mac Pro 4,1/5,1 is 400RPM, normal in-use minimum though is 500RPM at low load, any load and it's up from there. The listed specs for the Noctua are 450RPM min, so it might be that it's stalling at 400, so the SMC is increasing the voltage until it starts, but this voltage corresponds to 800RPM on the Noctua.

The 3 pin Noctuas I have will spin stable well below their rated minimum, I've got them stable as low as 240RPM, so it does look like the 3 pin and 4 pin varients may respond differently to input voltage. Which is to be expected really a the 3 pin fan *expects* variable voltage where as the 4 pin PWM expects to run on 12V and varies the RPM internally based on the PWM duty cycle signal, but it's still expecting 12V. We know they'll run on less, but maybe not as low as the 3 pin.

When you set the fan speed manually 2000 or above, the RPM shown in Macs fan control only shows as 2000ish

This is expected, it can't go faster than it's max so you get the cosmetic output issue, BUT that's beacause the SMC is working off the Apple default RPM range. If you manually tell the SMC what the new fan's MIN and MAX are *before*you open your fan control software (Macs Fan Control/smc Fan Control/iStat Menus/TG Pro etc.) then they will read the new limits and display correctly.

 

[p11hlf]

Thanks all, I had this noctua fan kicking around from an older project..... I appreciate its not 100% perfect but it works and has reduced noise from my machine substantially and reduced temps at the same time!

ill be changing the rest of my fans in due course, thanks so much to @amedias for doing the hard work on this.

 

[Tomblin]

EDIT: Amediast cleared this up. The Noctua 4-pin PWM will properly work as the 92mm PCI fan when wired correctly. It will only go as slow as 1,000-1,050rpm in my experience with iStatMenu but being a noctua fan that's not bad. Software can ramp up correctly from there.

Pinout for the 6 pin connector for the PCI fan

There's only 4 pins connected as it's jut the fan in there, the other two pins are empty.

(top left) = VCC (12V)
(bottom left) = GND
(middle bottom) = V Control
(bottom right) = Sense/Tacho

I followed your pin out diagram today, using a Noctua 92mm NF-A9 PWM it shows the RPM in my iStatMenu app but I have no control over RPM. I was excited to do this because we were supposed to adjust fan speeds with software.

Black #1 GND on the ••• bottom right.
Yellow #2 12v on the • top
Green #3 Tach on the •• bottom right.
Blue #4 PWM disconnected. •••• bottom middle disconnected.

Have I done something incorrect?
Is it perhaps the app I'm using that doesn't offer control over this fan?

I'm also looking to replace the 60mm fans in my 2x CPU coolers. You've done a wonderful job illustrating pinout for the PCI fan, considering all the wires have no colour coding. Is there a simple way to tell which wires are soldered to which for the others?

 

[amedias]

As per the few posts earlier, and my original post on page 1 all my testing was done with non-PWM 3 pin fans. It appears that at least some 4 pin PWM fans do not respond in the same way.

I’m afraid the only way to work out which wire is which is to label them and start testing! Either by probing output pins on the board while the machine is running, or where this is not possible, like on the CPU heatsink fans, apply test voltages to the fan and work it out. This was the most annoying bit of the whole procedure but if you label all the wires and work methodically only a takes a few minutes.
[automerge]1588574521[/automerge]

EDIT - also, looks like you’ve wired it up wrong too.

double check my notes on page 1, I think from your description you’ve wired the 12 fan input to the 12v output in the board and NOT the Vcontrol output pin which is middle bottom according to my notes and photo[/QUOTE]

 

[amedias]

It’s all fun and games here!
Got some more testing to do now...

 

[p11hlf]

That’s a nice collection you have going there, i have also done a little bit of testing for a broader 5,1 project i am working on.

two grey MAC fans will run using a PWM splitter from the front intake header.

two thermaltake ring 3pin fans will not run from the front intake header until you crank up the RPM’s using macs fan control.

Ill need 2 fans for my project, most likely the Noctua NF-f12 as they will be on display. If i cant get them both to work from one header, ill be splicing up the PCIE plug.

 

[p11hlf]

@amedias:
Quote:"
From the day I bought this MP 4,1 I removed the DVD drive, and along the way , any part that is not necessary like spinning HDD , obstructing the airflow ,generating more heat and slowing down boottime.
The optical bay , I think, can be seen as a tunnel in front of the PSU fan.
This tunnel should have a surface equal or bigger than the surface of the fan (circle of radius 6cm).
If the surface is smaller, this will lead to resistance , which means an increase of the airflow speed at the bottleneck, or just less throughput.

Last night I “removed” the plastic mechanism with the doors for the DVD- drives completely and glued some stainless steel mesh in the slots.
A quick test showed that now the PSU fan ,depending on the load, is running significantly slower (somewhere between 100- 300 RPM less) SMC set to default for the PSU fan.
Of course, this is on my set up.
I attached a foto of the result,I hope that the purists among us won’t be shocked

Can you give me some instructions on how the dvd slot doors and frame are removed, i removed one screw but it seemed to be fixed in there.......

 

[KeesMacPro]

Can you give me some instructions on how the dvd slot doors and frame are removed, i removed one screw but it seemed to be fixed in there.......

The "removal" of the door mechanism was something closer to violence than to science...
So I'm in no way responsible to the damage you could do to your Mac Pro, by doing the same thing...

I attached a photo of the plastic mechanism after removal.
You are right, there's only one screw , but as you can see on the other side there are two latches .

The doors when "open" slide into a space between the front and another perforated aluminium strip (which looks exactly like the front).
These strips got a tiny edge which prevents the doors from sliding upwards beyond the correct position (when closed).
That's why you can't get it out:
-the latches
-the doors

I placed the MP on its side, open side facing upwards.
I don't know exactly in what order, because I was fiddling for a while, but forcing the whole plastic frame a bit away from the front with a flat screwdriver , I managed to get the doors above the edge of the strips, and then you can slide the plastic latches in the back out.
Maybe its necessary to bend these strips a bit backwards, but after removal you can easily bend them back.
Another possibility is to manage to get the doors out of the plastic frame, by breaking or cutting the hooks at the end(photo), and then remove the whole unit.

I counted on breaking the whole thing (but it didn't), so it's something that could be irreversible ...
You will need patience to get it out!

 

[handheldgames]

Looking at this from a different perspective... I've built a Hackintosh to that needed additional PWM cooling and back in 2009 I picked up 4 spare 4,1 CPU bay fans. Plugging the cMP fan into the PWM motherboard connector, the RPM's are detected and I can control the fan without issue.

 

[KeesMacPro]

Looking at this from a different perspective... I've built a Hackintosh to that needed additional PWM cooling and back in 2009 I picked up 4 spare 4,1 CPU bay fans. Plugging the cMP fan into the PWM motherboard connector, the RPM's are detected and I can control the fan without issue.

I guess that's good news for the Hackintosh builders who use a cMP case!

 

[p11hlf]

Update from me.

4 pin Noctua fan was fluctuating from 0 to 950ish, and actually caused a shut down when in sleep.

4 pin PWM fans do not work!! Unless....

Caveat: if you run them at a constant 1100 rpm they are fine, no doubt this is something to do with the voltage and feedback but once running at 1100rpm i had not issues.

i Swapped out to a Noctua 3 pin fan, the black industrial model as the 2000 brown model was not available for quick shipping.

it also does not idle, and has to be run at 1100 rpm..... its much louder than the brown fan so will be coming out later! As i plan to use this for a project later down the road anyway.

cd rom doors

you must remove the screws on the doors themselves in order to remove this tray, and the Philips head screw that holds the assembly to the frame. It then can be pivoted towards the power supply and the two pins cans be removed.

 

[KeesMacPro]

you must remove the screws on the doors themselves in order to remove this tray, and the Philips head screw that holds the assembly to the frame. It then can be pivoted towards the power supply and the two pins cans be removed.

Cool!
You found the proper way

 

[amedias]

Update from me.

4 pin Noctua fan was fluctuating from 0 to 950ish, and actually caused a shut down when in sleep.

4 pin PWM fans do not work!! Unless....

Caveat: if you run them at a constant 1100 rpm they are fine, no doubt this is something to do with the voltage and feedback but once running at 1100rpm i had not issues.

Can you please clarify exactly which models you've tried as it is important to note this.

I have several 4 pin fans here that DO work.

- A Nanoxia Deep Silence 92mm PWM Ultra-Quiet PC Fan, 400-1400 RPM (actually runs fine down to 325RPM)
- A Zalman PWM fan from one of their CPU coolers but no model number on the fan, 800-1800 RPM (actually runs fine down to 450RPM)
- A TC (Thermal Control) DF0922512BIMN from a Foxconn CPU cooler, 500-2500 RPM

All the above operate fine when re-wired, WITH automatic control via SMC and do not stall or cause shutdown/sleep issues.

I'm also surprised at the sleep issue as you can even FULLY REMOVE Mac Pro fans and it will continue to operate normally in my testing, boot up, shutdown and sleep all work fine. Obviously it wont pass an ADT but it doesn't impact normal operation in any of my 4,1s or 5,1s.

My recommendation is still to use 3-pin fans and not 4-pin fans.

4-pin fans expect a full 12V as the supply voltage and vary the RPM with a PWM duty cycle signal. Although some PWM fans will operate <12V not all will and it's not intended that voltage be the primary control method.

3-pin fans expect 12V as supply for full RPM but variable input voltage is expected and an intentional method of RPM control for these fans.

i Swapped out to a Noctua 3 pin fan, the black industrial model as the 2000 brown model was not available for quick shipping.

it also does not idle, and has to be run at 1100 rpm..... its much louder than the brown fan so will be coming out later! As i plan to use this for a project later down the road anyway.

Again, model number please. As detailed in my earlier posts I have been using several Noctua 3 pin fans with no issues at all.
I think the Industrial ones are much higher RPM so possibly have a higher minimum RPM so will stall at the speed the Mac Pro wants to run them at by default.

A quick check though shows the 3-pin Noctua 120mm Industrial 200RPM fan model 'Noctua NF-F12 iPPC 12V 2000RPM' should have a minimum stable RPM of 450RPM, but it doesn't specify at what voltage. So it should be possible to run this fan < 1100 RPM without issues.

https://noctua.at/en/nf-f12-industrialppc-2000-pwm/specification

I'm curious as to why your results are so different... I have 3 Mac Pro chassis here. A 2009 4,1 and 2009 4,1->5,1 and a 2010 5,1 and they all behave the same.

May I ask what software you're using to control your fans, and have you done an SMC reset at any point?
Which position in the Mac Pro were you testing this fan in and exactly what model is it?
Have you tried it on other headers to see if the behaviour is the same or specific to one header?
Are you able to check the output voltages on the header to confirm what range you're seeing and if it increases properly as you manually change RPM?

I measured 0.2-12.3V on the V Control pin, When I have some more time I'll tabulate voltage against RPM for a few of my fans as an example.

This is highlighting why it is essential to pick fans that have a compatible RPM operating range, it's not vital to have the high max RPM but they must be able to run at the SMC minimum RPMs without stalling if you want to be able to use SMC automatic control.

for ref: the default SMC minimums in a 4,1/5.1 Mac Pro are:

PSU: 400 RPM
PCI: 800 RPM
INTAKE: 500 RPM
EXHAUST: 600 RPM

SINGLE BOOSTER: 800 RPM

DUAL BOOSTA: 600 RPM
DUAL BOOSTB: 600 RPM

They can be set lower manually if you fans can run at lower speed without stalling but those are the SMC defaults

 

[p11hlf]

Hi amedias,

Conclusive response, hopefully i can capture all of your questions

the pwm 4 pin fan i refer to is the following noctua

https://www.amazon.co.uk/gp/product/B07C5VG64V/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1

I found that when left on auto, the fan would occasionally surge. This left a gap on istats logging graph for all sensors. The fan would register from 0 to 950 rpm. I found that this irregularity was consistent until i left it running at 1100rpm.

before i remedied the irregularity i found that the sudden surge would cause the system to switch off when sleeping, i was not happy with this and once applying the 1100rpm it no longer happened.

not happy that a 4 pin would work correctly, i sought out the 3pin variant. Because i wanted to fix this quickly i ended up buy the following 3 pin fan.

https://www.amazon.co.uk/gp/product/B00KEST8PQ/ref=ppx_yo_dt_b_asin_title_o00_s01?ie=UTF8&psc=1

The noctua specs don’t include a minimum for this fan, however as explained above...... it has the same characteristics in terms of irregularity.

https://noctua.at/en/nf-f12-industrialppc-2000/specification

i have not reset the SMC, and i used macs fan control and istat. Could these two be conflicting? My reasoning for the two is to show the historical graphs but prefer the control that macs fan control gives me.

i note your comments regarding the ability to unplug a fan, I don’t believe your statement to be correct because if you remove the front intake fan then the exhaust and/or booster fun at full speed.

Lastly i have been using these fans in the PSU slot.

again, great work...... its people like you that keep these things running well.

 

[amedias]

The fan would register from 0 to 950 rpm. I found that this irregularity was consistent until i left it running at 1100rpm.

This is a symptom of either a fan that will not run at the SMC target speed, OR a misbehaving SMC. Basically the voltage output on the control pin is not enough to start/stabilise the fan, so it stalls. In response to this the SMC increases the voltage until the fan starts, at which point it then starts decreasing it again aiming for it's target RPM....and then the fan stalls again, repeat on and on....

the pwm 4 pin fan i refer to is the following noctua

Well that is very curious as I was pretty sure someone else had reported the NF-A12-25 working for them using the Mac Pro headers and the pin swap, although I can't remember if they were referring to the 3 pin version or not. I will see if I can dig out the reference.
It's not a model I've tested myself personally.

https://www.amazon.co.uk/gp/product/B00KEST8PQ/ref=ppx_yo_dt_b_asin_title_o00_s01?ie=UTF8&psc=1

The noctua specs don’t include a minimum for this fan, however as explained above...... it has the same characteristics in terms of irregularity.

I linked to the specs to on Noctua website above for the PWM version of that fan. Although the control driver will be different the airflow, pressure and RPM ratings are identical which would suggest the motor is capable of spinning without stalling at 450RPM when controlled via PWM signal. But with no listed min RPM and without one of those fans in front of me I can't check the starting voltage and min RPMs required when controlled via Voltage alone.

As you and another have noticed there are difference between Noctuas PWM and non PWM fans even when they otherwise appear to be identical in spec!

NF-F12 industrialPPC-2000

120x120x25

2000

29,7

121,8

3,94

3-pin

NF-F12 industrialPPC-2000 PWM

120x120x25

2000

29,7

121,8

3,94

4-pin PWM

i note your comments regarding the ability to unplug a fan, I don’t believe your statement to be correct because if you remove the front intake fan then the exhaust and/or booster fun at full speed.

Lastly i have been using these fans in the PSU slot.

This is why I asked about whether you'd tried any other headers or done an SMC reset and your sleep issue as I can confirm that on all 3 of the Mac Pros in the room with me right now I can plug and unplug the PSU (and other) fans without there being any immediate impact on the others. They do not spin immediately spin up, no fault (other than zero RPM on the header) is registered and there are no issues sleeping or waking. I've done this on the PCI and PSU fan headers and on the front Intake header numerous times while testing different fans before fitting them.

I routinely run two of my Mac Pros with no PCI fan fitted at all with no ill effects which is why I was asking, as I do not think your sleep issue/behaviour is normal so I wonder if there is something else going on?

Is yours a 2009/2010/2012?

i have not reset the SMC, and i used macs fan control and istat. Could these two be conflicting? My reasoning for the two is to show the historical graphs but prefer the control that macs fan control gives me.

They do conflict when both are running, but do not normally produce the behaviour you describe. how they conflict depends on which you start first and how you configure them to control/not control the fans.

You can check and control the SMC manually using the open source smc comman budled with smcFanControl

https://github.com/hholtmann/smcFanControl/tree/master/smc-command

Or if you use Linux using the AppleSMC module, either of these will give you direct access to control fan speeds manually, or automatically with modified min/max limits. (They also allow you to check temps on any sensor as well).

I have done the bulk of my testing using these as they allow a level of control that is not possible using Macs Fan Control, smcFanControl (gui), TG Pro or iStat Menus.

But I do often keep TG Pro or iStat Menus running to collect the historical data while I'm testing (but not using them for fan control so they don't override my test settings)

 

[KeesMacPro]

@amedias :

Noctua NF-A 9 FLX confirmed as a possible PCI fan!

Last week I ordered 2 fans which arrived today :
NF-A9FLX
NF-A12x25FLX

I tested minimum voltages to spin :
NF-A 9 FLX : V=4,5v
NF-A 12x25 FLX : V=5,0v
With a little lower V they start "pulsing".


I couldn't resist to do a quick swap for the PCI fan, it was like a walk in the park
Thank you so much for the pin-layout, it saved me a lot of time!

Unfortunately the PSU fan has to wait a few days, because of things I have to do on the same Mac Pro...
Since I only got installed a Mac OS I can't test the minimum RPM for these fans, but at least the NF-A9 runs perfectly smooth at default min. (800rpm).
No issues with wake/sleep , and normal response to SMC.

Both models come with new silicone grommets, rivets inclusive.
I put them "reversed" because I didn't see the possibility to stretch them when you mount the fan.

 

[p11hlf]

Ill reset the smc anyhoot, its a 2010 5,1

i have also removed the PCI fan without others running at full speed but when testing the intake header with fans i noted that another fan went full speed.

i will be buying the 3 pin nf-a12-25 and trying that later down the track.

 

[amedias]

@KeesMacPro

excellent! Thanks for the update and keep us posted on any more fans you replace, whether successful or not. All data is good data on a project like this

@p11hlf
Also keep us updated please, I’m especially curious about your machine as it seems to exhibit behaviour I have not seen and am unable to replicate.
I’d like to know if it’s specific fans we need to put on an ‘avoid’ list or if there is something else different about your machine.

 

[MIKX]

Ill reset the smc anyhoot, its a 2010 5,1

I advise doing the "TRIPLE Happy Mac CHIME NVRAM" reset fro m shutdown then old boot.

 

[p11hlf]

ok I have reset SMC and done a NVRAM/PRAM reset.

current NF-F12 industrialPPC-2000 still surges unless at constant 1000rpm or above

for clarity, I have not tried the NF-A12x25 4pin PWM fan since doing the SMC and PRAM reset. and since swapping over to the NF-F12 industrialPPC-2000, I have not experienced any power offs during sleep. This is a 3 pin fan.

ill await the feedback on the NF-A12x25FLX from @KeesMacPro before I order.

 

[amedias]

current NF-F12 industrialPPC-2000 still surges unless at constant 1000rpm or above

I think this is almost certainly because at the target default SMC RPM it will be outputting a voltage below that fans starting voltage.
Only way to be sure would be multimeter probes on the pins and test the minimum starting voltage of the fan and it's corresponding RPM. Will await further info on the A12x25 FLX as it may be more voltage tolerant than the industrial models.

I am going to be buying some more fans to test soon but can't do it all at once as this project has already sucked a fair bit of money out of me since I've already bought nearly a dozen fans, an additional CPU tray and a bunch of CPUs for testing just to satisfy my curiosity!