XPS 8940, Better 3rd party cooler

Also updated to Noctua fans (NH-U9S). Huge improvement. The noise of GeForce 1650 Super is not good / not bad. Unfortunately the fan in the PSU is super noise. Can somebody confirm that? Any recommended replacements for PSU fan?

For information :

I just updated to BIOS version 2.0.11 from march, 2nd a Dell XPS8940 with 10700 (non-K)

Typical behavior on heavy load before was :

4.7 GHz -> hitting 100°C in 5 seconds -> throtlling progessively to 4.1GHz @ 95°C

After 2.0.11 :

4.7 GHz -> hitting 100°C in 5 seconds -> throtlling aggressively to 3.8GHz then stabilizing to 3.9GHz @ 80-85°C

Temperatures look better this way, at the cost of 5% frequency loss

This is with the stock cooler and fans, replacing with NHU-9S would probably improve temperature further or allow higher frequencies.

Edit : that was not long, back to hitting 95° regularly.... maybe the case had the time to freeze a little before the first test.

Hello, I installed NH-9DL fan. I was trying to see if I can change fan control in BIOS. According to user manual, there should be a "Fan control Override" under "Power Management" in BIOS. But I don't see "Fan Control Override". Does everyone see the same thing under BIOS?

@sck90  Not yet, I'm planning to fit it this weekend.  I'll definitely report back.

Thanks, installed and all good. One question has to do with screwing in the heatsink with the 2 screws supplied. I was worried how tight to go. So as soon as I go a decent amount of resistance, I stopped tightening. Should I go back and tighten those 2 screws for the heatsink until they stop moving?

Thank you!

Mike

Did you need to remove the motherboard and use their backplate?

I completed my cooler journey today.  I've previously fitted an Arctic F9 PWM fan on the rear panel and an Arctic F12 PWM on a 3D printed frame at the front in the HDD position.  Today, I fitted an Arctic Freezer i13 X CO replacing the Dell stock CPU cooler.

The computer is so quiet now and I'm currently running at core temperatures between 27°C and 29°C with an ambient of approx. 20°C; I've got a dozen pages open in Chrome, plus two mail programs, file manager, Excel, Artweaver Plus, MS Teams and CPUID HWMonitor open.  The lowest core temperatures I've measured in the last few hours since rebooting have been between 24°C and 27°C.

Arctic F9 PWM replacement for Dell stock rear panel fanArctic F12 PWM on 3D printed frameArctic Freezer i13 X CO on CPU, plus cables tidied

For those dealing with the Dell XPS 8940 CPU cooling issue, I have a i9-10900 (non K) processor, so it shipped from Dell with the smaller CPU heatsink (not the VWD01 shown in the adverts or service manual).   When put it through CPU stress tests, it would almost immediately enter a thermal throttle at 100 degrees (celsius),  It benchmarked at a routine peak of 96 degrees.   In other words, any sustained processor demand would result in in a mid to high 90 degree temperature (and thermal throttling).  To solve this, I chose to go the stock K processor Dell route (vs the other good solutions shown in this thread).   The Dell Parts line was exceptional and for minimal cost I purchased the VWD01 (cpu heatsink) along with the 612F7 (VR Heatsink).   The 612F7 is helpful due to the change in fan orientation on the CPU Heatsink.  The parts arrived more rapidly than promised and installation literally took 10 minutes.   For good measure, I removed the front hard drive enclosure (and relocated the drive to the upper enclosure) and installed a Noctua NF-A9 through a SATA power connection at a fixed speed of 1250 RPM (for low noise).   The 8940 no longer thermal throttles at all during a stress test (heat is never an issue for me now) and measures a peak of 82 to 83 degrees (13 to 14 degrees lower).   The maximum processor speed has increased and other scores have improved as well.  This is the way it should have been shipped from the factory (marketing should talk with the cost accountants).   By the way, the Noctua accounts for one to two degrees of the peak CPU improvement, and slightly improves the overall score of the box.

@Jon-62 

Nice work.   I have always thought that adding the 120mm fan to the upper intake position is a good mod.   Did you trim/cut away part of the divider on the inside of the front bezel so air can actually get to the upper intake position?

The way Dell designed that front bezel they created a solid piece of plastic on the upper part of the front bezel rather than put slots there like on the Inspiron model.   Just trim way or cut some slots on the upper/lower divider on the inside of the bezel and the temps should improve even more.  There is no need to cut any openings on the bezel itself, the air from the lower grill should reach that upper position once you create a path.

The air coolers being used seem to be adequate in managing temps..  With the upper 120mm intake fan it should even be more effective.   Being able to mount the 120mm fan there also opens up the possibility of using the Corsair H60 (2018) CPU Liquid Cooler which should be plug and play on this motherboard like on the XPS 8930 and Alienware Rx models.  Would need to measure the clearance of the RAM sticks and SATA connectors on the motherboard at that location for the radiator + 15mm/25mm wide fan + 3d frame.

Are you planning on making those 3D printed frames for the 120mm fan install available for interested parties?

@Jon-62 

My mistake, looks like the 3D printed fan bracket was from @purpledrillmonkey from this post.

Plans look like they are here for anyone who has a 3D printer.

Good stuff!!

@HanoverB  Yes, I had the frame manufactured to @purpledrillmonkey 's design here in UK.  No, I've not cut away any of the plastic on the inside of the front panel yet, but I'll consider it if I need additional cooling.

thx all of you

I have done some very rough testing on a few fan configurations on the XPS 8940 for anyone looking at ordering/modding things. The XPS 8940 I have is the i7-10700 (non K) and the RTX 3070.

I used Cinebench R20 for CPU testing and Furmark for GPU testing but did not run them simultaneously. Each test was only a few minutes long as I did not loop cinebench and I only ran furmark until the temp plot basically levelled out. Case side panel obviously installed, and ambient was around 20-21°C. 

Worst score I saw in CB20 was 4182 at stock, with the best score being 4421 for test 5. Definitely take all results with a grain of salt on accuracy and validity - I'm no Gamers Nexus lol.

My conclusions in no particular order:

• CPU cooler upgrade is by far the most substantial improvement - if you can only do one thing, this is it.
• Rear fan upgrade is helpful but probably moreso for noise.
• Front fan seems to be a minor upgrade - either due to restricted intake, not enough RPM due to whatever fan curve/logic is applied, or a combination of the two. Sidenote, I tried swapping the Y connector so that the PWM signal was going to the front fan and this didn't do anything I could measure.
• Nothing seemed to really improve GPU temps - I think the variance shown in these results is well within error. The intake of the GPU just cannot be fed cool air without a fan at the lower front so I will be looking to tackle this still with some sort of 3D printed bracket.

Anyway, fun experiments. I'm also presently working on temps for an Alienware R11 w/ RTX 3090 which is much more finnicky for fan selection and chassis options, but hey I've got another front fan bracket done for it! lol

Well done.  This is good empirical evidence.  Your tests are helpful confirmation of the key conclusions regarding the need for owners to seriously consider an upgrade to the stock CPU heatsink in a non-K 8940.