Supermicro Xeon D / FCBGA 1667 Heatsink information

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

maxu

New Member
May 29, 2017
5
0
1
43
First experience: Accelero L2 Plus
Positive:
  • low profile
Negative:
  1. some fins overlap with CMOS battery and the mini-SAS connector
  2. no match for the hole pattern (53 x 53 mm is the closest size)
These can be fixed with a cutting/drilling tool (I already own Proxxon 28481) and longer M2 screws (currently on the shopping list).
But, out of the box, it's a NO GO.

Next affordable "giunea pig" could be: Deepcool V90
Simple design, but with a better elevation, to bypass the clearance issues.
 
Last edited:

Discodave

New Member
Jun 22, 2017
2
0
1
37
Cambridge
First experience: Accelero L2 Plus
Positive:
  • low profile
Negative:
  1. some fins overlap with CMOS battery and the mini-SAS connector
  2. no match for the hole pattern (53 x 53 mm is the closest size)
These can be fixed with a cutting/drilling tool (I already own Proxxon 28481) and longer M2 screws (currently on the shopping list).
But, out of the box, it's a NO GO.

Next affordable "giunea pig" could be: Deepcool V90
Simple design, but with a better elevation, to bypass the clearance issues.
Cheers for the info. Looks like the V90 is also out of the question due to lack of 51mm mounts.

However the V400 does have 51mm mount holes...
 

maxu

New Member
May 29, 2017
5
0
1
43
Well... in my part of the world it goes to somwhere near 35$ wich is out of the "budget-friendly" zone actually.
Also, this 51mm distance might be only the width, since the only videocards with this spacing are GTX460/GTX465
And this named hole spacing is, in fact, 51 x 61mm.

Eventually, there is the "do it yourself" way.. :)
 
Last edited:

Donald Thomas

New Member
Jan 8, 2018
4
0
1
65

Hesterbest

New Member
Apr 4, 2018
1
0
1
44
Hello

Not sure I am adding anything of value, but I am just sharing my current experience in trying to silence/cool my Xeon 1541.

I created a 3D model intending to replace the stock fan-mount on my X10SDV-TLN4F REV 2.00. The stock fan was too loud, so I figured I had to go up in fan size in order to get it quieter. The stock fan was noisy, but provided good cooling. The CPU temperature sensor peaked at 68 degrees Celsius at 100% load using the stock fan, and idling at 30 degrees Celsius.

The original fan-mount had fan-mounting holes 40mm apart. My new model has the mounting holes 50mm apart. I can easily mount my Noctua fan. Also, my model has increased the radius of the hole that the air passes through in from the stock 23 mm to 27mm. Hopefully this increase allows for the increase in air to flow nicely without being "blocked". The stock fan-mount had wall thickness of 1mm, but a 3D printed model with wall thickness of 1mm tends to become fragile, so I increased it to 2mm. Also, the fan-mount mounting holes has an larger radius than the stock model in order to prevent congestion of the ASA material used.



STL-file

I did not do the printing myself, I made use of 3dhubs.com. The 3D print I received was not very smooth around the small details like the mounting holes. Some issues with the small holes that are used for mounting the fan-mount, no issues at all with the holes for mounting the fan. Overall I am happy with the print, possibly my model needs adjustment for the fan-mount mounting holes.

The Noctua fan is quieter than the stock fan. I have not made any lab measurements, so I only have my subjective opinion on this. However, the stock fan did a better job at keeping the CPU cool. The Noctua fan is idling between 40-45 degrees Celsius and peaking at 78 degrees Celsius.

I have started reasoning over why the Noctua fan did not provide as good cooling as the stock fan. My Noctua fan has a higher airflow than the stock fan, so if we simply compare the airflow, the Noctua fan should in theory provide better cooling. I think there must be some other variables than the fans themselves that affects the cooling, and the only thing that changed was the fan-mount.

This is as far as I have gotten. I find myself trying to solve a problem in a field where I have no experience and little/no knowledge, so any input is very much appreciated. I do not feel comfortable trying to remove the stock heat sink (or the back plate), so I rather continue down my current path trying to solve this problem. I have tried to reason around what causes the heat increase with the new fan-mount. From my reasoning I have come down with two things that I think can be improved:

1) The material of the fan-mount. The original fan-mount material is some type of metal (I think, this is really not my field of expertise). The material used for my print is "Standard ASA". I suspect my printed fan-mount does not conduct heat like the stock fan-mount, and that it might be the cause of the temperature heat increase. Materials and their ability to disperse of heat is NOT my field of knowledge. So in my clueless mind I have a theory that the ASA fan-mount may be preventing heat dispersion, rather than aiding in heat dispersion. And if there is any validity to my theory, having a copper version of my fan-mount should help. Anyone here with experience in materializing a 3D model in copper? Online 3D Printing Service | i.materialise seems to provide this service, but am I heading down a wrong path here?

2) The air hole in the fan-mount is too small and not effectively letting the air from the fan flow through onto the heat sink. I can quite easily change my 3D model and increase the radius of the air hole to 28mm (from 27mm), however I am worried the wall thickness may become an issue.

I am hoping that someone in this forum may have a comment or two.

Cheers

Additional information:
* Ambient temperature was about 20-22 degrees Celsius during testing
* This needs to fit within a Silverstone DS380b
 
Last edited:

Netwerkz101

Active Member
Dec 27, 2015
308
90
28
What is the difference you have seen from the stock fan and cooler? i have the XEON D 1541 in a 1U case with three fans and at idle the CPU temp is 68 degrees C and easily goes up to 80 plus when active, i need to change the fan fast.
Sorry .. just now seeing this.

Keep in mind I had an X10SDV-TP8F which is D-1518 (4c/8t) vs. your D-1541

I saw temps at ~40 degrees Celsius at idle with any fan setup I tried:
Stock heat sink with a Noctua 40mm fan laying on top
CoolJag active cooling combo
Supermicro active cooling combo

Swapped copper heat sink from Cooljag to Supermicro.. no changes in idle temps.
 

fl0w

Active Member
Oct 28, 2015
117
47
28
38
Paris
Hi,
I tried removing the heatsink on a X10SDV-4C-7TP4F but it seems to be glued on the CPU. Do you think they use adhesive thermal paste? I didn't force to remove the heatsink, I was afraid of damaging the CPU.
 

sth

Active Member
Oct 29, 2015
379
91
28
I put some load on mine to heat the CPU up and then it twisted off without much force. From cold it wouldnt budge before applying the force that made me nervous too.
 

fl0w

Active Member
Oct 28, 2015
117
47
28
38
Paris
Thanks, with some load I was able to remove it.
The fan is very loud, any PWM fan to recommend for
SNK-C0054A4L heatsink?
 
Last edited:

fl0w

Active Member
Oct 28, 2015
117
47
28
38
Paris
The fan keeps switching between 3000 and 6000 RPM, anyone tested a Noctua or Scythe fan on this heatsink?
 

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,394
511
113
Check what the IPMI thresholds are set at - the fan cycling between high and low typically means that one of the fans connected is throwing a low rpm assert (which, in my case at least, was actually triggered by having a noctua fan).

Thread when I ran into this issue on an Atom board:
https://forums.servethehome.com/ind...ro-a2sdi-8c-hln4f-ipmi-and-fan-control.17834/

PigLover's treatise on tweaking SM fan control settings:
https://forums.servethehome.com/index.php?resources/supermicro-x9-x10-x11-fan-speed-control.20/

Once the assert stopped happening I found the fan noise to be perfectly bearable at 3000rpm.
 

fl0w

Active Member
Oct 28, 2015
117
47
28
38
Paris
Thanks, I set all the fans to 75%, keeping it a little quiet.
I guess the main issue is the fan directly in contact with my case backplane:



The best solution would be to replace the fan with a blower fan, what do you think?
 

Marsh

Moderator
May 12, 2013
2,644
1,496
113
I have 4 x X10SDV 4 cores boards.

I am cheap, so I went to my "fan junk box", picked out some Panasonic fans that fit on top of the heatsink.
The fans only have two pins but reasonably quiet in full RPM mode. I used 4 zip-ties to secure the fan on top of the HS.

Just for fun, I re-run Prime95 this morning for the past 2 hours.
The board in a tight Supermicro Superchassis CSE-721TQ-250B Mini-Tower , 4 core Xeon-d temperature is holding at 68 degree.

The other one in a chassis with a zip-tied heatsink fan Plus a 80mm blowing at the system board. CPU temp is 53 degree.

Try zip-tie a fan on top for a quick experiment before doing more surgery .

I was reasonably happy the CPU temp, so I stop.
 

fl0w

Active Member
Oct 28, 2015
117
47
28
38
Paris
This Noctua fan is too thick so I ordered a Cooljag BUF-A4 and an Evercool EC6010TH12BP fan. The fan is PWM, 60mm and 10mm thick which would be great. The Cooljag heatsink is thinner than the Supermicro SNK-C0054A4L so I should get at least 5mm free on top of the heatsink + fan. Airflow should be better than now.
 

nev_neo

Active Member
Jul 31, 2013
157
44
28
Has anyone tried slapping on an EVO T2 on these Xeon-D's ? I have it in a Corsair Carbide 100R silent - but the cpu fan running at 3300rpm is the noisiest in the box. I've swapped out the heatsink for the LSI3008 and for the Intel X540, trying to rig something bigger on the Xeon-D that would quite it down some more.
This is one of the Asrock D1541D4U-2T8R that I picked up from the datto sale earlier last month.
Coincidentally, I have one of the supermicro heatsink fan combo for the Xeon-D and that feels a lot heavier than the Asrock fan combo.
 

nev_neo

Active Member
Jul 31, 2013
157
44
28