Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
@romaric17 :
Exactly!
Now by knowing V and I, you can calculate R=V /I.
The resistor X you'll need is the voltage drop (12-4,4=7,6) / I.
Example: you measure 20mA(=0,020A)>R=7,6/0,020 =380 ohms.
With a power rating of 2w for resistor X , you're on the save side.
If you can't measure it, you could also do it by testing different resistors, between 200 and 500 Ohms, they cost like €0,15...
 
Last edited:
I'm very interested in the fan speed setup.
Could you maybe post a screenshot of MFC, iStats or whatever tool you're using to adjust the T vs. fan speeds?
Thx in advance.

Here are the heats after 24 hours on a medium Mac desktop usage with 3 monitors on and CPU load around 12% for 2 hours (some background audio encoding going on) with auto settings for fan control. Unfortunately, I do not have a screenshot of the temps before the fan upgrade.

When I stress the system fully including the GPU, I need to boost PCI, exhaust and intake fans to keep temps under control. What I really like about the new fans is that they are silent even when you boost them a little bit in these situations.

My current issue in this setup is the power (and fan) hungry Radeon VII which I think makes PCI fan inadequate - Which I'm going to manage with GPU water cooling next.
 

Attachments

  • system.png
    system.png
    53.9 KB · Views: 299
  • heats.png
    heats.png
    72.6 KB · Views: 320
Hello amedias,
Thank you so much for your great write up. One thing I wanted to discuss is the option of replacing/re-using the Apple motor controller. This way, everything would stay the same. Every fan comes with a motor controller chip and a motor in one unit. It should be possible to separate the controller board from the motor and replace just the motor. This way the electrical values would not matter for the original mac pro system, just the motor would be replaced. After all the motor has only two leads, - a simple 12V motor. All you need is to carfully separate it and resolder it, assuming the the motor has roughly the same size/power/miliWatt values.

Hi there - did you actually go down this path and try swapping the fan IC?
 
Here are the heats after 24 hours on a medium Mac desktop usage with 3 monitors on and CPU load around 12% for 2 hours (some background audio encoding going on) with auto settings for fan control. Unfortunately, I do not have a screenshot of the temps before the fan upgrade.

When I stress the system fully including the GPU, I need to boost PCI, exhaust and intake fans to keep temps under control. What I really like about the new fans is that they are silent even when you boost them a little bit in these situations.

My current issue in this setup is the power (and fan) hungry Radeon VII which I think makes PCI fan inadequate - Which I'm going to manage with GPU water cooling next.

Thx for the screenshots!

I think there are lots of opinions about how to setup your fans ,so I'll try to leave that by side..

I was curious about this PSU fan for example, especially because of the Radeon VII load.
For my taste the PSU gets too warm, if this is with 12% load..
Since its a single CPU, I think adjusting the fan a bit could lower it significantly.
I guess you did the Pixlas mod? I would recommend it because of the thirsty GPU.

I agree with you: I doubt there's much you can do (with the PCI fan) about the heat the Radeon generates... water-cooling would be nice !

I noticed the Northbridge T delta is 16 degrees... I would replace the rivets etc
After reading a test about the best thermal paste for heatsinks with lower pressure,
I replaced the Arctic silver (which I always used) by Thermal Grizzly Kryonaut for the IOH and noticed an improvement of a view degrees.
The delta of the IOH in my MP 4,1>5,1 dual always stays between 5 and 9 degrees...

EDIT: I run a dual with X5690 and a GTX780Ti, so the PSU Temps are ,like in your setup, higher too,
removing the DVD drive , and later the whole door mechanism lowered the PSU Temp noticeable.
 
Last edited:
Hello amedias,
Thank you so much for your great write up. One thing I wanted to discuss is the option of replacing/re-using the Apple motor controller. This way, everything would stay the same. Every fan comes with a motor controller chip and a motor in one unit. It should be possible to separate the controller board from the motor and replace just the motor. This way the electrical values would not matter for the original mac pro system, just the motor would be replaced. After all the motor has only two leads, - a simple 12V motor. All you need is to carfully separate it and resolder it, assuming the the motor has roughly the same size/power/miliWatt values.

It' a great idea, and one I'll look into at some point (unless you beat me to it!) after I've finished testing this method and the V->PWM converter modules I've got on order.

The only stumbling block I can immediately see is that there is a LOT of variation in fan housings, in both housing diameter and fixings etc. so it might be tricky to find a replacement fan that has both the specs required, and sufficient physical room to accept the driver from an Apple/Delta fan and attach it. Another potential pitfall is that most fans are quite difficult to disassemble and re-assemble without causing damage.
[automerge]1589969023[/automerge]
I noticed the Northbridge...

Quite why Apple went with such a pathetic cooling solution for the NB when the rest is so well engineered is a mystery that I cannot explain. I mean, it's a neat idea, having the NB heatsink where it is in the airflow path and it can take quite high temperatures safely but it would have been easy (and worthwhile) to have just made the NB heatsink a bit bigger! Or include it fully in the CPU heatsink with it's own heatpipe?

An opportunity lost....I have been rummaging through my "'box 'o crap" looking for old heatsinks I could butcher and modify to replace the OEM one, and there's plenty of candidates that might work on the dual CPU trays, but room is a bit tight!

No-go on a Single though as the NB is fully under/inside the CPU heatsink so not a lot can be done there, but the Single CPU NB runs a fair bit cooler anyway.
 
Last edited:
Not having a big GPU does make a considerable difference to PSU temps and fan speeds accordingly.

This Single CPU machine has been on a burn-in test for 19hrs with CPU @100% on all cores, but no Ram or GPU load, stats shown in attachment.

It's deadly silent at those fan speeds (with Noctuas) and temps are warm but OK, could be considerably cooler with a bit more RPM and would still be quiet.
 

Attachments

  • Screen Shot 2020-05-20 at 11.16.48.png
    Screen Shot 2020-05-20 at 11.16.48.png
    193 KB · Views: 223
  • Like
Reactions: KeesMacPro
Not having a big GPU does make a considerable difference to PSU temps and fan speeds accordingly.

This Single CPU machine has been on a burn-in test for 19hrs with CPU @100% on all cores, but no Ram or GPU load, stats shown in attachment.

It's deadly silent at those fan speeds (with Noctuas) and temps are warm but OK, could be considerably cooler with a bit more RPM and would still be quiet.

...still testing the NFA12X25ULN as a PSU fan, and this one works perfect and extremely silent too:)
RPM goes up to 1200 max for this version, but it seems sufficient for my setup.
I have to put my ear in front of the Optical bay to hear it, and when I do, I can even hear the inductors buzzing behind the fan...
I'll report a datasheet by the end of the week.
@p11hlf : this could be the fan you were looking for: no issues, (almost) inaudible and runs from SMC min rpm to 1200 max ...


An opportunity lost....I have been rummaging through my "'box 'o crap" looking for old heatsinks I could butcher and modify to replace the OEM one, and there's plenty of candidates that might work on the dual CPU trays, but room is a bit tight!


Same here!
In the dual you could make a larger heatsink, especially because the space directly in front of the fan(inside the CPU heatsink) is open almost up to the top..
I've been brainstorming about that a lot last year.
I even contacted a firm specialized in heatsinks:)

BTW I got my IOH mounted with nylon m3x16 nuts and bolts.
I prefer that solution;its rock solid , and you can even carefully adjust the pressure on the die.
 
Last edited:
Another two confirmed working models with pin swap

Noctua NF-P12 REDUX 12V 900RPM
Noctua NF-P12 REDUX 12V 1300RPM

Tried in cage and PSU locations, strangely despite the different max RPM both fans had a min RPM of about 300RPM at starting voltage (4.5V). These are both pressure optimised fans so was curious to compare to the airflow optimised models. I wanted to try one of these in the PSU location as I was curious if the slightly higher static pressure would help get air through the PSU.

Initial impression is that they are very slightly noisier at low RPM than the airflow fans, but slightly quieter at max RPM, so that's erm...odd? It is hard to say though as the noise is a slightly different pitch so it it's a bit subjective. Cooling performance seems to be as good as the airflow fans at the same RPM, so if there is a difference then it's probably too small to be of note.
 
Last edited:
  • Like
Reactions: hutzi20
These are both pressure optimised fans so was curious to compare to the airflow optimised models. I wanted to try one of these in the PSU location as I was curious if the slightly higher static pressure would help get air through the PSU.

Nice!
That's actually one of the things I've been thinking about too: the PSU is better cooled by a pressure or a airflow optimized fan?
Seems like somewhere in the middle..

according to the information by Noctua the fans I've been testing (NF-A) are recommended for both purposes..
With the ULN I've got here as a PSU now, used for Intake and exhaust too, you might get the lowest possible noise level.
They're rated for only 12dB at max rpm if I remember well!
My doubt is if the airflow would be sufficient for the CPU case...
Tomorrow I'll test the lowest rpm ,voltage etc.

BTW the NF-A9FLX I installed runs faster than rated:about 1680rpm max
 

Attachments

  • NF-A9FLX at max rpm.png
    NF-A9FLX at max rpm.png
    8.2 KB · Views: 199
Last edited:
Can i jump in a bit on this venturing slightly off topic to the relative clone of the 4,1 known as the 2009 xserve?
As some may or may not know there are a pile of 40mm fans in this thing behind the 3 drives and in front of the CPU/Memory.

Should we expect that 3 pin 40mm fans could be swapped in and have proper SMC control?
the fans are pairs of different blade and depth fans likely to get the CFM and turbulence needed to get the cooling in a datacenter.

With that in mind does anyone know of a good source for 3 pin 'quiet' 40mm fans that can run in the right RPM range?

MFC claims they spin idle at 1500 RPM but this seems to be off by a factor of 3 compared to what the IPMI reports which is around 5000 rpm at idle and a lot higher at full blast.
 
I'm wondering what a good replacement for the CPU fans would be - both are capable of up to 5200 RPM, but no Noctua seems capable of reaching that speed.
 
I'm wondering what a good replacement for the CPU fans would be - both are capable of up to 5200 RPM, but no Noctua seems capable of reaching that speed.

There aren't many off-the-shelf fans that have the range of RPM of the Apple fans. There are plenty of high speed fans, and low speed fans, but few with the overall range, either 3-pin or 4-pin PWM. But there are plenty that can provide adequate range, it's jsut a case of working out what your requirements are. If you don't already have a good understanding of what your fans are up to with your workload you may need to do some extended testing to find out.

If your environment and use case requires the full RPM range (ie: your Mac runs with the CPU fans regularly at high speed due to your environment and workload) then you're probably better off sticking with the Apple fans as NO fan is going to be quiet at high RPM.

If your environment and workloads allow you to get away with lower RPMs then you may be fine with replacement fans. For example I rarely see speeds >1200 RPM in normal use, and a 24x7 torture tests don't even see my ans exeeding 1700-1800RPM with a pair of 130W TDP CPUs fitted, lower TDP or single CPU and RPM is even less so I have no issues using lower fan speeds.

Under normal use the 60mm* CPU fans in a dual Mac Pro are actually pretty quiet below 1200RPM, the big 120mm cage fans and PSU fans however can contribute significant noise (even at lower RPM) and only get worse as they spin up, so they're normally my first point of call for replacing with quieter ones as there are plenty out there will enough cooling capability at lower noise levels.

* Weirdly the 80mm fan in a single CPU Mac Pro is noisy, and has an annoying tone so it's much more noticeable (to me) than the dual CPU fans under normal use.
 
  • Like
Reactions: atonaldenim
Had an hour or so spar this morning, made a couple of fan adapter cables up for pin swap method to make my testing easier and reduce stress on the LoBo connectors from all the plugging and unplugging. So now this adapter cable does the pin swap and no need to mod each fan to test.

Also.... my Voltage -> PWM converter boards arrived from China, had a preliminary play and.... SUCCESS!! I can use 4-pin PWM fans with SMC control. And you get the full RPM range of whatever fan you fit, so you can use fans with a min RPM above the SMC min default. Need to test more fan models obviously as I only hd 4 PWM fans here to play with but they all worked fine on initial testing. Need to do some more sleep/wake tests etc. to make sure it all still works as expected there.

The converter boards aren't big or difficult to wire up but they are bulkier than a pin swap so I'll probably keep using the pin-swapped 3-pin fans on my own machines but this is an option for people who want to use un-modified 4-pin PWM fans, the boards cost under $5 each. These could be the solution for those of you wanting to use the higher RPM Noctuas.

I'll do some more testing once I get my hands on some of those fans, just waiting till I have some spare cash as they're not cheap fans!
 

Attachments

  • IMG_6286.JPG
    IMG_6286.JPG
    360.5 KB · Views: 369
  • IMG_6285.JPG
    IMG_6285.JPG
    341 KB · Views: 362
  • IMG_6284.JPG
    IMG_6284.JPG
    382.6 KB · Views: 359
Make it fit the PCI fan and all your camels will bathe in milk, I promise.
Well wiring it up to work is easy, the issue is where to put the board... it might be easier to break it out with an extension cable to the DVD bay (which is the obvious place to hide the board for PSU fan), and the take it back to the PCI housing, then it’d just be a few wires and fairly unobtrusive. It’d only need a tiny mod to the PCI fan housing to ‘escape’ the cabling, a small hole drilled in the plastic should do it.
 
  • Like
Reactions: hwojtek
Had an hour or so spar this morning, made a couple of fan adapter cables up for pin swap method to make my testing easier and reduce stress on the LoBo connectors from all the plugging and unplugging. So now this adapter cable does the pin swap and no need to mod each fan to test.

Also.... my Voltage -> PWM converter boards arrived from China, had a preliminary play and.... SUCCESS!! I can use 4-pin PWM fans with SMC control. And you get the full RPM range of whatever fan you fit, so you can use fans with a min RPM above the SMC min default. Need to test more fan models obviously as I only hd 4 PWM fans here to play with but they all worked fine on initial testing. Need to do some more sleep/wake tests etc. to make sure it all still works as expected there.

The converter boards aren't big or difficult to wire up but they are bulkier than a pin swap so I'll probably keep using the pin-swapped 3-pin fans on my own machines but this is an option for people who want to use un-modified 4-pin PWM fans, the boards cost under $5 each. These could be the solution for those of you wanting to use the higher RPM Noctuas.

I'll do some more testing once I get my hands on some of those fans, just waiting till I have some spare cash as they're not cheap fans!

Thanks for posting the wiring. I ordered/received a different controller board(DC 12v 3a 4pin PWM PC CPU Fan Temperature Control Thermostat Speed Controller):


My question is - could I possibly wire up the Control Voltage connector into the thermostat pin?

2nd Question - I am specifically working on the PowerSupply fan. The wires inside fan controller loom are they color coded? To wire up the PWM controller board I ordered, I would have to identify VCC(#2), Ground(#1) on the system board side . I would wire the Techometer wire from the FAN directly to the #3. I know you identified the wires on an earlier post and all the wires were black. I could only find Noctua PWM fan that ran over 2500+RPM, so that's the reason for not going down the 3wire fan. The connector for the PowerSupply fan can be viewed here:


I do have a cheap/crappy multi-meter and I will admit I am not savvy with it. So I could probably identify the wires with some guidance.

Appreciate your hard work!
 
@amedias tested lots of fans, including different Noctuas.
I've tested a few Noctua's too and these I'm still using with satisfaction :
PCI: NF-A 9 FLX no issues whatsoever , runs officially up to 1600rpm, but in real life reaches ~1700 rpm.
PSU:NF-A 12x25 ULN no issues, runs up to ~ 1250 rpm.

The Intake/exhaust and PSU fans are the same, so I'm planning on replace the Intake/exhaust by the NF-A 12 ULN too.
They're extremely silent (12dB! )

Note that these fans can not run the max rpm of the original Delta fans, so if the rpm range is sufficient for your setup, depends on the load , GPU, single or dual CPU etc...

I've tested the NF-A 12x25 FLX too, but the min rpm leads to issues (not recommended ).
 
Last edited:
  • Like
Reactions: smallwonder
@amedias tested lots of fans, including different Noctuas.
I've tested a few Noctua's too and these I'm still using with satisfaction :
PCI: NF-A 9 FLX no issues whatsoever , runs officially up to 1600rpm, but in real life reaches ~1700 rpm.
PSU:NF-A 12x25 ULN no issues, runs up to ~ 1250 rpm.

The Intake/exhaust and PSU fans are the same, so I'm planning on replace the Intake/exhaust by the NF-A 12 ULN too.
They're extremely silent (12dB! )

Note that these fans can not run the max rpm of the original Delta fans, so if the rpm range is sufficient for your setup, depends on the load , GPU, single or dual CPU etc...

I've tested the NF-A 12x25 FLX too, but the min rpm leads to issues (not recommended ).
Oh the NF-A 12 ULN works? That's fantastic. My PSU literally never exceeds 1,100 RPM and often runs at 900-1,000 and is noisy so I'll certainly give that one a go.

I'm curious is this would also be sufficient as intake and exhaust. I made the mistake of replacing my intake with the one you mentioned and I'm in the boat of a minimum of 1,000-1,100 for intake and it's less noise than before but I'm sure something that idles better would be ideal for that location. I might order a couple ULN one for the PSU and try swapping my intake for the ULN and see if it can accommodate the idle speeds there better.

I've also 3D printed a new bracket to hold an SSD and a 120mm fan in front of the PCI fan I would like to test the ULN version in.
 
Since early May 2020 I have been using a Noctua 120mm NF 12x25 with a Noctua NF FC1 PWN fan speed controller as my PSU fan. it is silent at just about any speed, Works very well.

 
Well wiring it up to work is easy, the issue is where to put the board... it might be easier to break it out with an extension cable to the DVD bay (which is the obvious place to hide the board for PSU fan), and the take it back to the PCI housing, then it’d just be a few wires and fairly unobtrusive. It’d only need a tiny mod to the PCI fan housing to ‘escape’ the cabling, a small hole drilled in the plastic should do it.

I have a question regarding fans..

Ideally I would like to have 120mm fans with white LED's, for pretty much every fan Ive looked at, it shows an RPM of around 1400 +/- 10%. Range is 7v-12v. In a normal environment with no temp spikes, would these be fine?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.