[Ncase M1] Custom Watercooled, dual-240 mm Radiator, 8700K + 1080 Ti

[For_Science]

This is a summary thread of my experiences in the Ncase M1 that used a Intel 8700K + 1080Ti in a fully custom watercooled configuration. I also have some notes about temperatures using a 240 mm AIO (H100i v2) before and after delidding. I will be happy to answer any questions about this for those that want to try something like this in the future!

 

[For_Science]

Components
CPU: Intel i7 8700K
Motherboard: Asus Z370-i Strix
RAM: G. Skill Trident Z 2 x 16 GB, 3200 C16
VGA: MSI Geforce GTX 1080Ti Aero
Storage 1 & 2 (Windows & Linux): Samsung 960 Evo 1 TB x 2
Bulk Storage: Samsung 850 Evo 1 TB x 1
PSU: Corsair SF600
Fans: 2x NF-A12x15 PWM

PSU Installation
Obviously, nothing special about this step beyond commenting on the size of the Ncase M1, after installing the Corsair SF600 PSU, a lot of the space had already dissapeared!

 

[rfarmer]

Very nicely done, looks like a tight squeeze. How are your temps on the 8700k?

 

[For_Science]

Very nicely done, looks like a tight squeeze. How are your temps on the 8700k?

Thanks
Using a full-system stresser like Realbench, I got 75 degrees on the CPU (delidded) and 54 degrees on the GPU.

 

[axtran]

Wow, looks awesome. I ended up with a 92mm rear radiator and a 240mm bottom radiator... ended up "losing" the front ports in the process, though.

 

[Acalanata]

OMG! Please, post more pics from this project!

 

[For_Science]

Wow, looks awesome. I ended up with a 92mm rear radiator and a 240mm bottom radiator... ended up "losing" the front ports in the process, though.

Yes, thats why I had to put the bottom radiator in pull exhaust, instead of my original intended push exhaust. By putting the fans on the bottom instead, this allowed me to clear the front i/o just

 

[For_Science]

Test booting, and initial thermal tests using a 240 mm AIO

The system was test booted outside the case, using a H100i v2 as a CPU cooler.

I consciously decided to use low profile fans for the intake on the bottom despite being able to fit 25 mm fans because I thought the gap between the fan and GPU would be beneficial for airflow.

However I noticed that it was quite difficult to mount the 240 mm AIO with the CPU block in the orientation, and so had to rotate the block 90 degrees with the pipes facing the upper panel. I took the opportunity to investigate thermal paste spread, which was also good.

Using Realbench as a whole system synthetic stress test, I measured about ~76 degrees on the CPU and ~67 degrees on the GPU. These are at "out-of-the-box" setting + XMP, at the time, this would mean MCE was turned on.

 

[For_Science]

Delidding and re-try
Obviously Sky/Kaby/Coffee Lake CPUs are well known for their potential for temperature improvement with liquid metal TIM swap, and that is exactly what I did. Temperatures improved from the previous 76 degrees to about 60 degrees during Realbenching for the CPU, obviously no difference for the GPU. The delid-die-mate-2 was used and the liquid metal was thermal grizzly conductonaut.

 

[For_Science]

Waterblock installation and first iteration with bad planning
Now onto the good stuff... Waterblock installation was fairly standard except for the following quirks: I had to cut out some holes in the plastic gasket for the CPU block to accomodate the motherboard capacitor ends, and also mounted the CPU block upside down for the routing. GPU blocks were installed as standard using Thermal Grizzly Kryonaut as the TIM for all blocks.

Hardware Labs GTS Nemesis 240 mm Radiator was installed on the bottom in a pull configuration. I would have personally liked to do a push configuration but this was not possible due to some collisions with the front I/O (which by lifting the radiator up by the fans in the bottom, were cleared)

GPU block installation, thermal paste was spread as per instructions for kryonaut.

D5 pump was mounted onto the rear of the chassis using the fan grill part and the tubing holes (rubber gasket was removed), it is a tight fit, but it works.

Another pair of slim noctua fans to provide some airflow to the board as well as another 120 mm radiator to supplement the cooling performance. In attempt to balance the two exhaust fans under the GPU, these fans are in a intake configuration with the radiator on a push-pull mode.

First iterations with bad planning
Due to some poor planning with cable management, I had to put a drain port via a T-splitter just before the pump, which was a very bad idea and was eventually fixed, but this was the layout that I had used initially that works apart from the drainage system.

Temperatures were quite interesting in that while the GPU went from 76 degrees to 54 degrees (Air --> Water) the CPU went up from the 50 degrees to 75 degrees. It is clear to me that in this case, 360 mm is really the bare minimum to keep a 8700K and 1080 Ti under control. The added heat from the GPU and the power delivery of the GPU makes the system much warmer for the CPU than when it had a dedicated 240 mm AIO. Having said that these are not noise normalized metrics and so the system is much quieter under water.

 

[For_Science]

Fixing the drainage system and final tube runs
As mentioned in the post above, the drain port was put in a really stupid place due to bad planning on my behalf, so I had to change a few things to move it to its original intended location. Hindsight does say that this would have been done from the very beginning ideally.

Firstly, excess SATA and molex power cables were moved to the front of the case

This freed "just" enough space to move the drain port to the PSU area via a splitter. This would be improved if I had custom cables to length which at the time did not have.

This freed up some space in the main compartment as well and a few runs were cleaned up with the use of fittings

and when closed, almost every bit of space is taken up

Very unassuming when everything is closed off

Apart from the optional RGB, of course (no, its not always rainbow mode)

Overall the system is very quiet and I quite like knowing that there is an absolute beast of watercooling inside this sleek case. Hope this gives ideas for others and I am willing to answer any questions you may have!

 

[Engr62]

This is an amazing build. I bet it weighs quite a bit with all of that packed in there.

 

[For_Science]

This is an amazing build. I bet it weighs quite a bit with all of that packed in there.

Thanks
Yes, its definitely not what I would call light, but it still fits into a hand-carry for an aircraft (this has travelled from Sweden to the UK in my suitcase, drained of course).

 

[riktanius]

Thanks
Yes, its definitely not what I would call light, but it still fits into a hand-carry for an aircraft (this has travelled from Sweden to the UK in my suitcase, drained of course).

i didnt drain mine before traveling from the US to Italy... I dont think it is required.

 

[For_Science]

i didnt drain mine before traveling from the US to Italy... I dont think it is required.

Perhaps, but I didn't want to risk any excuse for my glorious system to be taken away from me.....

 

[Supercluster]

That's some pretty nice packing you've done with this build!

You might also be the person to unriddle the question for myself and, at least, a few other people: What cools better; a radiator with a standard (25mm thick) fan (in a pull orientation for example), or a radiator + two slim (12-15mm fans), noise normalized?

 

[For_Science]

That's some pretty nice packing you've done with this build!

You might also be the person to unriddle the question for myself and, at least, a few other people: What cools better; a radiator with a standard (25mm thick) fan (in a pull orientation for example), or a radiator + two slim (12-15mm fans), noise normalized?

I think that's a fantastic question, and I'm afraid the answer is not so clear cut anymore. At the time of their release the 15 mm Noctua fans had shown similar performance to their thicker brethren such as the NF-F12 in a noise normalized scenario, and so at the time it was clear-ish to me that I would go push-pull on this particular intake radiator.

However with the newer NF-A12x25 I think you may be better off with one of those instead. Either way, the difference is probably neglible in this case since the majority of the heavy lifting is still done by the 240 mm radiator down below which can only have a 15 mm fan on it anyway.

 

[hyp36rmax]

Nice! Great job!

 

[T4iga]

Firstly thanks for posting the various angles of your impressive build.
I am currently half done the process of cramming hardware into my M1 v6. I was wondering if the bottom rad assembly is the only way to retain the front IO. Would you go with a 30mm rad for the bottom again or would you pick a thinner 25mm instead? Do you have any data on bottom fan orientation?

 

[For_Science]

Firstly thanks for posting the various angles of your impressive build.
I am currently half done the process of cramming hardware into my M1 v6. I was wondering if the bottom rad assembly is the only way to retain the front IO. Would you go with a 30mm rad for the bottom again or would you pick a thinner 25mm instead? Do you have any data on bottom fan orientation?

I would still go for this 30 mm radiator for performance reasons given that the 15 mm Noctua fans do have quite a lot of static pressure. The only thing I would do differently perhaps is to mount the fans in push since I do get the impression that these fans perform better in a push orientation. The only reason why this wasn't done the first time was I think the holes for the radiators would not line up fully without compromising the front I/O. However If you accept the radiator will be pretty solidly grounded even with only 4-6 screws, I think I will try this in my next scheduled maintenance.

I would also consider replacing the push/pull 120 mm fan with a single A12x25 Noctua, as that wasn't available at the time of the build. Also to consider would be to replace the D5 pump for either the Eisbaer LT or Apogee drive and accomodate a 2nd 240 mm radiator instead of the 120 mm for even better cooling.