De-Dusting & Northbridge temp reduction in Dual CPU cMP 4,1 & 5,1

[MIKX]

( NOTE : I feel that both Northbridge temp reduction AND de-dusting/de-clogging the cheesegrater DUAL CPU cMPs deserves a separate thread to embrace the concepts. Northbridge cooling requires that all that hot air be efficiently extracted from the CPU case.
====================================================
Since Mid March I have been experimenting with auxiliary methods to cool the Northbridge chip in DUAL CPU 4,1 & 5.1 cMPs. To date I have been contributing to the . . “Cooler Northbridge on the 5,1 DUAL CPU” thread started by Drolson ( many thanks ! ) https://forums.macrumors.com/threads/cooler-northbridge-on-the-mac-pro-5-1-dual-cpu.2173519/page-2 My better contributions start from Post #39

I initially installed a USB powered 50 x 50 x 10 11CFM fan powered from a spare USB 3.0 PCIe socket angled slightly downwards to the Northbridge heatsink.
I have since installed a 2nd. Larger generic USB fan to cool the 1 ~ 4 RAM modules.
======================================================
Now I have moved on to general cMP case cooling with the focus on DUST !

DUST & The Cheesegrater “ case HOLES ”​

Two days ago I removed my dual cpu tray & my PSU and then proceeded to thoroughly DE-DUST ALL the fans in my DUAL cPU 4,1 case. I adapted my vacuum cleaner with a very slim nozzle capable of getting in between fan blades. I also have thin brushes with non-conductive bristles.

I was particularly stunned at the amount of accumulated dust that was clogging the CPU tray EXHAUST FAN rear ‘cheese grater’ dot-holes grid. I’m certain that hot air extraction was down by at least 50%. Got rid of all of that dust with the stiff bristle brushes & vacuum cleaner.

Further, I went on to ( using slim stiff bristle brushes ) & vacuum cleaner to de-clog 99% of ALL the cheese grater case holes. I cannot stress enough how effective this is in increasing general airflow.

If you undertake this “ hole de-dusting “ I recommend that when finished that you shine a flashlight through the holes - you will then be able to see which ones are still blocked by accumulated dust.

As I now only use an external USB DVD drive I also opted to tape OPEN the two DVD slot popup/down doors to allow larger openings for the PSU fan to assist airflow to the PSU fan.

I have removed ALL internal ( hot ) spinner HDDs. I boot from an M.2 PCIe Samsung 970 EVO blade. ( I have USB 3.0 & USB 2.0 external docks )

CONCLUSION​

The combined de-dusting - cheese grater hole cleaning + Aux USB fans in the CPU tray area have resulted in a more robust airflow & dramatically reduced case/diode temps.

Our 4,1 ~ 5,1 Mac Pros are now between 7 to 10 years old now and need more tender loving care.

I live in Japan and with Summer peak temps of up to 40°C / 104°F rapidly approaching I'm looking forward to a cooler cMP. In Summer 2018 my 4,1 dual was virtually a room heater that I didn’t need.
NOTE 1 : My Intake & Exhaust fan settings may seem a bit high - these settings are what I imagine Summer peak will require.
NOTE 2: This thread is specifically for DUAL CPU 4,1 & 5,1s
======================================================
This iStat menus grab was done today April 30th.in Japan mid-Spring.
Ambient cMP room temperature 25°C = 77°F.

( EDIT 3: I just noticed that my PSU temps are down 3~4°C since the de-dusting/clogging ! )

 

[h9826790]

I am with the single processor cMP, but I can confirm that just a thin layer of dust can make the NB run about 5C warmer.

This is my NB temperature record to shows the difference after I blow the dust away from the heatsink.

It's not super dusty, just a thin layer, but still about 5C difference.

 

[MIKX]

h9826790

Probably a lot warmer when Hong Kong's summer gets going. The same in Japan.

 

[h9826790]

h9826790

Probably a lot warmer when Hong Kong's summer gets going. The same in Japan.

Yeah, but for single processor model, the NB isn't that hot. As long as I keep the fan / heatsink clean, and make the booster fan auto spool up when the NB warm up. My NB usually can stay at around 60C in summer time with Booster spin up to around 1100.

 

[MIKX]

h9826790

Yes, the single CPU cMPs have a lot more open space in the cpu tray.

The dual CPU cMP's NB heatsink could have been designed a lot better.

 

[AnimeFunTv]

Honestly, I use Mac Fan Control and I bump up the speed of the fans when it reaches close to its Tcase temp. My room gets pretty dusty as well and I normally use an air compressor to blow off the dust from the inside my MacPro, for the hard to reach places I use a 1 inch nylon paint brush to break off the hard dusty sports. I do this about once every two months or so. Its helped my bought new 2006 1,1 MacPro survive all these years, but I have since retired my 1,1 for my 4,1>5,1 MacPro and I still keep that scheduled cleaning.

I think adding those extra fans might be overkill, you can always use Mac Fan Control to bump the speed up on the back fan to 'suck' out the excess heat in the cpu tray area.

 

[h9826790]

Honestly, I use Mac Fan Control and I bump up the speed of the fans when it reaches close to its Tcase temp. My room gets pretty dusty as well and I normally use an air compressor to blow off the dust from the inside my MacPro, for the hard to reach places I use a 1 inch nylon paint brush to break off the hard dusty sports. I do this about once every two months or so. Its helped my bought new 2006 1,1 MacPro survive all these years, but I have since retired my 1,1 for my 4,1>5,1 MacPro and I still keep that scheduled cleaning.

I think adding those extra fans might be overkill, you can always use Mac Fan Control to bump the speed up on the back fan to 'suck' out the excess heat in the cpu tray area.

In general, spin up the intake fan is better than spin up the exhaust fan.

Suck out the heat sounds better for cooling, however, this fan setting will create a negative pressure, makes air "leak" into the case in all the gaps, which also means, all the dust will leak in as well.

Spin up the intake fan will create a positive pressure inside the case, which makes the warm air leak out from the gap, and avoid dust getting in.

Besides, as general rule, NEVER use constant fan speed mode when using MacsFanControl (unless you assign them to run at 100%). MacsFanControl alter the target fan speed, which means will completely disable all system protection. No matter how hot the CPU / NB is, the fan won't further spin up but maintain at your assigned RPM.

Of course, if you set a fan profile that base on proper sensor's reading, then it's safe to use MacsFanControl.

Last but not least, for others info, there is no T-case reported in cMP. And there can be a big gap between T-diode and T-case. I've seen some guys confused these two figures, and believe their CPU was overheating when T-diode reach the T-case max. But actually their CPU still running at full turbo boost speed (which means well within limit).

 

[MIKX]

My current testing phase will run until late September. Last year my 4.1 2009 CPU tray failed.
The refurbished one I bought cost around US$580 + shipping. I do not want to go through that again and if keeping the CPU tray area 23°C / 73.4°F cooler ( presently ) using a combination of Macs Fan Control plus two USB fans + regular. So be it. I do not regard the fans as overkill, in fact I cannot hear them at all.

For a start CPU A gets downstream hot air from both CPU B and the rear 5 ~ 8 RAM modules.
If clogged case grid holes are clogged and the Intake & Exhaust fans are caked with dust that only compounds the problem.

Of course the Northbridge chip can run perfectly well at 70°C / 158°F but what do those temperatures do to the Northbridge Heatsink rivets over a prolonged period.

Removing 99% of my 4,1's accumulated dust + de-dusting all fans + de-dusting clogged case grid air holes + the addition of two fans -one for the NB & the other for 1~4 RAM modules + removing all internal spinner HDDs has resulted in reduced ( some dramatically) overall temperatures. Additionally my 4,1 is now strategically placed to allow vented hot air to be dissipated.

Both my 4,1 & the room where it is situated are now noticeably cooler.

Current iStat Menus readouts : Currently my Northbridge Tdiode is at 47°C / 116°F. For most of the day since 6am. it has only fluctuated by +/- 1° C.

( Since de-dusting my PSU temps are down by 3°C ~ 5°C / 37.4°F ~ 41°F. )

 

[MIKX]

Had our first ' hot 'day where I am in Japan yesterday, 29°C / 84.2 F. Hotter indoors in the Mac Pro room.

The Northbridge Tdiode rose to 55°C / 131°F at 2:45 pm but slowly fell to 54°C / 129.2·F at 6pm. and then a rapid decline through the rest of the day with a lowest temp of 41°C / 105.8°F at 6am. this morning. ( Here in Japan in late August high evening temps often lasting through the entire night. )

I feel confident that as summer temps continue to rise from early July ~ late August that the NB Tdiode will be around 59°C / 138°F ( or lower ).

An identical Sunon 50mmx50x10 11CFM is now assisting cooling of 1 ~ 4 RAM modules in front of CPU A. 1 ~ 4 RAM module temps are way down.
=============================================

I have ordered two Noctua fans to replace my Intake & Exhaust Apple fans.

=============================================

Keeping open the two DVD tray sliding doors in the open position seems to have resulted in lower PSU temps. NOTE : Replacing the Apple PCU fan requires removing the 4,1 backplane which is something I will eventually get around to.

Bear in mind that this project is an experiment. the use of USB fans to assist in NB & RAM module temp reduction may seem like ' overkill ' to some but I would rather have my 10 year old 4,1>5,1 cMP function as a computer rather than a room heater.

NOTE : I cannot stress enough that de-clogging all the cMP case ventilation holes and de-dusting the Intake & Exhaust fans WILL result in more efficient ventilation and significanly lower temps, especially in the lower CPU tray area.

 

[AidenShaw]

but I would rather have my 10 year old 4,1>5,1 cMP function as a computer rather than a room heater.

Of course, if the cMP is pulling 800 watts from the wall, then it's putting 800w of heat into the room - regardless of the internal temperatures. A bunch of small fans may reduce peak temperatures at certain points, but they won't make your room cooler.

 

[MIKX]

AidenShaw

. . open house windows and a generous ventilation space around the cMP . . which just happens to have it's exhaust fans facing a screened but open window helps shunt a lot of that 800w out of the room.

Mind you I wouldn't have open house windows without security grids in the USA but then .. I live in Japan.

 

[MIKX]

Tonight I finally had enough time to pull apart my 4,1>5,1 to remove the PSU.
Thoroughly de-dusted the PSU fan and the PSU as well as cleaning the PSU case space.

Now my DVD drive bays area is completely empty with teh DVD drive sled doors taped OPEN to allow more airflow. This resulted in a 4~ 6°C / 38.2°F ~ 42.8 F reduction for this time of day ( 1:20 am ).

I found a 12 CFM. 60mm square x 11mm 12 CFM fan in my Mac Pro bits box and now use that to assist in cooling the CPU A 1~4 RAM modules.

Lastly having removed the CPU tray I de-dusted & cleaned the Intake & Exhaust fans and further used canned air to blow dust out of all areas.

All fans including the two USB fans ( powered by spare PCIe card’s USB 3,0 ports ).

Put everything back together and the 4,1>5,1 is quietly purring along like new.

I’m looking forward to seeing how this all performs in the height of the typically hot Japanese summer.

================================

If anyone wants to cool down their Northbridge ( and RAM perhaps ) here is the link to the DigiKey Sunon catalog - https://www.digikey.jp/products/en?FV=ffecfe96
but be warned DigiKey’s shipping charges are exorbitant eg: My first Sunon 50mm square x 10mm 11CFM fan cost about ¥400 ( US$4.5c ) shipping was over US$20 ! but I found that Amazon has most of the interesting, high CFM/ performance fans in stock.
Google “Amazon + Sunon + the Sunon fan part number.

NOTE : Do NOT buy a5v fan,​

PSU bolts 4,1 cMP

I'm keeping the DVD drive area 100% empty to record whether this helps with PSU cooling.

( Of course I de-dusted & cleaned this fan.)

The other fan is which is positioned above the Northbridge heatsink is a 12v Sunon 50mm square x 10 mm 11 CFM I adapted to USB.

 

[MIKX]

Currently I power the two USB fans by routing the USB cables out from the CPU tray through a 5mm wide slot I cut into a 'spare' cMP 2,1 side cover ( it fits perfectly but won't lock ). I then run the two USB fan cables under the 4,1's bottom and up to two spare USB 3.0 ports on my SATA powered Inatek 4 port KTU3FR-4 USB 3 PCIe card.

Not completely satisfied with the above . . I have ordered an Inateck SATA powered 7 port KTU3FR-5O2U PCIe USB 30 card. ( US$ 27.99 ) https://www.inateck.com/inateck-ktu3fr-5o2u-7-port-usb-3-0-pci-express-card.html
Two of the Inatek's 7 ports are internal which will allow me to power the two USB fans internally. Using this card I can replace my 4,1's original side cover.

 

[MIKX]

Today : Japan, mid Spring May 6th. 2019.

( no airconditioning in my tradional Japanese "Machiya" house . )

Time of day : 2pm in Japan.

Outside temp. 31°C / 87.8°F
Cmp room temp. 30°C / 86°F
Cmp iStats Menu :Ambient temp. 34°C / 93.2°F
Northbridge Tdiode temp. 54°C /129.2°F

It would be helpful, nice to see other 4,1 / 5,1 cMP owners temps.

 

[MIKX]

( NOTE : Bear in mind that this project is an experiment in reducing temps for the Northbridge Tdiode, RAM and the CPU Tray case area in general using ..

1. Two small, high CFM rating, USB fans 2. Slightly increased Intake Exhaust fan speeds.

================================================

Today's results :May 7th. 2019. Japan, mid Spring

Best 24 hour cycle results so far.​

I have learned a lot about cooling the Northbridge chip.

Just ordered 2 Sunon MB50151V1-000U-A99
50mm x 50mm x 15mm x 17 CFM fans which I will adapt to USB ( easy to do if you have moderate soldering skills ).

These two fans should be able to maintain still lower temps Northbridge Tdiode and RAM modules 1 ~ 4 ( in front of CPU A ).

DigiKey ref: https://www.digikey.jp/products/en/...e96&quantity=&ColumnSort=0&page=1&pageSize=25

This fan MB50151V1-000U-A99 is available on Amazon ( DigiKey's shipping is very expensive )

 

[MIKX]

May 12th, 2019 JAPAN

It's definitely getting hotter everyday here in Japan. Currently 32°C / 89.6°F outside in my garden at 4pm. I got hold of an Omega Typhoon 60mm x 60 x 10mm 12v 16.4 CFM fan for the Northbridge which I converted to USB.
cMP room ambient temp is 33°C / 91.4 ( no air conditioning ).
Last year, 2018 at the same time/date with similar outside temps the Northbridge was constantly at 70°C / 158°F so it appears that Northbridge fan cooling is viable.

I got hold of an Ainex Omega Typhoon 60mm x 60 x 10mm 12v 16.4 CFM fan for the Northbridge which I converted to USB.

This fan's 60mm x 60 x 10mm size seems to be the optimum as it can be angled downwards towards the NB heatsink.

This fan is apparently sold only in Japan for ¥869. https://www.amazon.co.jp/AINEX-OMEGA-TYPHOON-薄型・超静音タイプ-CFZ-6010SA/dp/B00YH79H60
I now have two them, the other one will be changed to USB and used to cool the my Samsung 970 EVO & it's adapter.

Ainex also sells a SATA to TWO pin cable which I used to set up the spare Omega Typhoon to cool my Samsung 970 EVO M.2

Common sense will tell you which SATA plug locking TAB to 'snip off.

The two 'support erasers ' firmly keep the fan in place but I will come up with a better solution.

The GPU is not actually as dusty as it looks.

 

[MIKX]

Best consistent overall results to date - getting hotter here in Japan every week but with today's results I feel that I'm finally ready for Japan's hot, humid, dusty summer.

If you are in the Northern hemisphere I highly recommend de-dusting your CPU Tray area every two weeks ; no need to completely remove the CPU Tray, just shut down, remove the power cord, slide the CPU Tray out just enough to allow access for canned air.

 

[MIKX]

( Bear in mind that this thread is an experiment to see just how much extra cooling can be achieved particularly for the Northbridge chip which heats up the whole area near CPU A. Additionally cooling for RAM + the PSU + any M.2 NVMe ( or AHCI ) blades present. )

With the usual hot & humid summer rapidly approaching here in Japan, this year I want to be prepared !
====================================================

Today I got a new toy ! An HDE infrared laser thermal imaging unit fro m Amazon for US$13.99 + shipping https://www.amazon.com/HDE-Accuracy-Non-Contact-Temperature-Thermometer/dp/B009RUPQGW

First I tested it and it seems to be accurate. It is rated for +/- 1°C accuracy depending on the distance from the target

At the time I scanned this RAM module iStats Menu showed 23°C

So . . . this unit will come in handy for monitoring case temps.

==========================================
Yesterday's iStat Menus.

====================================

I've now turned my attention to my Samsung 970 EVO with it's "Owltech " M.2 adapter which comes with a heatsink. I've installed a SATA powered 12v Omega Typhoon 16.14 CFM fan ( only seem to be available in Japan ) with two rubber erasers to hold it firmly in place.
The effect was immediate with a 9°C / 48.2 reduction in iStats Menu.

No perceptible noise nor vibration.
=================================================

From now until June I will be looking at RAM temp reductions for DUAL CPU trays.

I'm starting to have doubts as to whether RAM module heatsinks actually DO achieve as much heat dissipation as touted. In my case, i feel that a good hi-CFM fan strategically positioned might achieve more.

In any case a hot Northbridge chip plus hot 1 ~ 4 RAM modules most certainly make the entire CPU A area hotter than it needs to be.


NOTE : Those with SINGLE CPU trays needn't worry too much as there is a lot more space for the fans to dissipate heat.

 

[MIKX]

Yesterday in my part of Japan the outside ambient temp was 36°C / 75.9°F with low humidity.

From 11:45 am. to 4:45pm the Northbridge chip hovered at 56 ~ 58°C. 132.8 ~ 136.4 °F

Didn't have time to do any iStat Menus screen grabs.

 

[h9826790]

Yesterday in my part of Japan the outside ambient temp was 36°C / 75.9°F with low humidity.

From 11:45 am. to 4:45pm the Northbridge chip hovered at 56 ~ 58°C. 132.8 ~ 136.4 °F

Didn't have time to do any iStat Menus screen grabs.

Nice work.

I think even on the hottest very humid day. Your NB should still able to stay around 60C. Which IMO is a nice temperature for cMP.

 

[MIKX]

h9826790

I did a " no extra fans " test last week on a hot day. All fans were set to automatic.
The NB Tdiode rocketed up to 62°C straight from bootup and then went to 70°C +.

My cMP sits in a well aired space - not close to walls in a room with all windows fully open.

I fully agree that 60°C is well within the Northbridge chip's thermal limits but that 60°C heats everything close to it and surely is not good for the NB rivets or any applied thermal compound.

Here in Japan where I am situated ( mountains to the North, East & West ) it often gets to 38°C in July & August and even 40°C in the shade occasionally.

2018 late August.

My Japanese home is a traditional " Machiya " style with Tatami flooring which allows airflow
I do not have any electrical air-conditioning beside a couple of room fans which I have running during the peak time on the hottest days.

I'm quite happy with the 56 ~ 58°C NB temps I'm getting presently using the ( now ) 16.4 CFM fans which in my mind are showing ( to me at least ) that the DUAL cpu trays cooling could have been designed better. The NB Tdiode's heatsink could have been copper and in it's current format gets warm air from CPU B's fan ( & the intake fan ) plus it gets HOT air from the 4 RAM modules directly behind CPU B. I suspect that this contributes to why CPU A always runs hotter than CPU B.

The current NB heatsink fans could have been extended up & around to the front of CPU A and above the CPU A's 4 RAM modules - that would have allows a little more airflow from the existing intake & CPU B's fans.

Now that I have the HDE infrared laser temp gun . . . .

. . . . I have discovered that the 4,1's side cover surface gets up to 5°C hotter than the rest of the case ... perhaps an "icepack" over a drip tray & strapped to side cover . . . . .

This current " Northbridge Cooling " experiment has very revealing and has now been extended to frequent " regular Summer de-dusting = ( weekly ) cooling the PSU + PCIe slots area where my Samsung M.2 970 EVO lives + removing ALL, hot spinner HDDs from the cMP case

Everything is running much cooler than standard & just as silent as before . . . . and that cannot be a bad thing.

========================================

NOTE : Those with SINGLE CPU 4,1 / 5,1 CMPs needn't worry too much as there is much more airspace in those trays.

 

[h9826790]

Just my opinion. I think 60C should be OK for the surrounding components.

No scientific data to support, but that's a very common NB temperature for the single processor cMP in hot Asia. Since single processor cMP rarely has that rivet issue. So, it seems that extra 15-20C NB temperature make a big difference in this area.

For the heatsink design. I think it's very effective already. And the reason to use aluminium most likely because it does better job than copper on transferring the heat to air.

 

[MIKX]

The 2009 4,1 Northbridge heatsink

by comparison

The 2008 3,1 Northbridge heatsink

I actually am not. sure what the 4,1 heatsink blades are made of they are certainly not aluminium .. an alloy ? stainless steel ?

Copper Vs Aluminium

Copper is more expensive than aluminium = logical choice for Apple

I have two heatsinks on one of my M.2 PCIe adaptors; the original is a nylon spring rivet aluminium one - I attached a copper one with slightly adhesive, good brand thermal tape - I also tested using just the aluminium one and 20 minutes later with the copper one ( both are approx the same size & height ). The copper one only ( in the same position as the aluminium one ) showed better results with both iStats menus and with my infrared laser thermal gun.

( You can see both heatsinks attached in post #18 in this thread. )

Copper is faster at transferring heat but you are correct in saying that copper is slower to dissipate copper-heat to the air. . but that is where my extra 16 CFM fan comes into play !

I have run one of the 16 CFM fans on the CPU A front RAM modules on a warm, hottish day and the iStat Menus comparison between CPU B RAM and CPU A RAM was dramatic. CPU A's RAM was up to 5°C cooler; of course the RAM heatsinks are not copper.

I have access to a CGI milling machine; gonna fool around with it this summer.
I think that it is possible to replace the top fins on the Northbridge heatsink with re-designed copper ones. No free time for this at present but definitely on my " to do " list.

In any case this is a productive, fun project.

 

[h9826790]

The 2009 4,1 Northbridge heatsink

View attachment 839203

by comparison

The 2008 3,1 Northbridge heatsink

View attachment 839202

I actually am not. sure what the 4,1 heatsink blades are made of they are certainly not aluminium .. an alloy ? stainless steel ?

Copper Vs Aluminium

Copper is more expensive than aluminium = logical choice for Apple

I have two heatsinks on one of my M.2 PCIe adaptors; the original is a nylon spring rivet aluminium one - I attached a copper one with slightly adhesive, good brand thermal tape - I also tested using just the aluminium one and 20 minutes later with the copper one ( both are approx the same size & height ). The copper one only ( in the same position as the aluminium one ) showed better results with both iStats menus and with my infrared laser thermal gun.

( You can see both heatsinks attached in post #18 in this thread. )

Copper is faster at transferring heat but you are correct in saying that copper is slower to dissipate copper-heat to the air. . but that is where my extra 16 CFM fan comes into play !

I have run one of the 16 CFM fans on the CPU A front RAM modules on a warm, hottish day and the iStat Menus comparison between CPU B RAM and CPU RAM was dramatic. CPU A's RAM was up to 5°C cooler; of course the RAM heatsinks are not copper.

I have access to a CGI milling machine; gonna fool around with it this summer.
I think that it is possible to replace the top fins on the Northbridge heatsink with re-designed copper ones. No free time for this at present but definitely on my " to do " list.

In any case this is a productive, fun project.

Oh, I actually forget to finish what I want to say in the last post.

I want to say that's effective already, but really too small

 

[adam9c1]

Is cleaning with compressed air (from a 110v compressor) OK?

I would hold the fan blades still and wait several hours for any possible moisture to dry up.