-
Save 20% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Concept MK-2 (Sub 6L Watercooled SFF)
- Thread starter Zackmd1
- Start date
System up and running! Idle temperatures are 32c for the CPU and 34c for the GPU. Downloading a few benchmarks now to get an idea of load temperatures!
Ok so load temperatures... Running the heaven benchmark for an hour at ultra settings and 1080p resolution (99% GPU usage and about 40% CPU usage) produced stable load temperatures of about 60c CPU and 73c GPU. Not the greatest in the world for water-cooling but keep in mind that this is with a non ideal tubing situation (big loops of tubing to prevent collapses and small diameter hose barb reducers) which is hurting liquid flow and airflow. When I redo the loop I will be moving to all 1/4" tubing which should increase flow as I will not be going through a lot of small diameter hose barbs and I will be flipping the fans around to allow hot air to be exhausted out the top of the case instead of exhausting it into the case.
The case itself faired well with the temperatures. There was some bowing/warping in a few weak areas that I have already addressed with the production candidate. Flipping the fans around will also greatly help reduce any warping that might occur due to temperature. Moral of the story, a 65 watt CPU and a 120 watt GPU is about the max you will be able to cool with a dual 80mm rad setup. A CPU only loop or GPU only loop will likely be the best option for water cooling.
The case itself faired well with the temperatures. There was some bowing/warping in a few weak areas that I have already addressed with the production candidate. Flipping the fans around will also greatly help reduce any warping that might occur due to temperature. Moral of the story, a 65 watt CPU and a 120 watt GPU is about the max you will be able to cool with a dual 80mm rad setup. A CPU only loop or GPU only loop will likely be the best option for water cooling.
This seems weird to me. The r9 295x2 is a 450W card cooled by a 120mm radiator. 120*120=14400 which is only slightly more than 80*80*2=12800.
Radiator surface area I don't think is the problem. The problem is with airflow over the fins. The 80 mm fans can't move enough air over the fins without sounding like a jet engine. To check to be sure the case design wasn't an issue I removed the side panels basically making it an open air test bench and temps only decreased by about 1-2c.
Plus I believe my current water loop has severely limited flow due to the tube routing and reducers. I hope to test this theory Wednesday when I redo the loop in 1/4 tubing. I'm going to give it one more shot with both the CPU and GPU and then I'll likely move to a CPU only loop if the tests don't fair well.
I don't think airflow over the fins is the problem. If you compare the airflow of the best noctua 120mm fan (non-industrial) vs two of the best noctua 80mm fans, the airflow is nearly identical.
It will be interesting to see the temperature with the improved tubing.
What I am saying is while 80mm fans can produce enough flow, they sound like a jet engine doing it. During the test I had the fans set to an RPM just below audible but ramped them up to full blast in order to stop the climb in temperatures. I sure do hope that my new tubing will improve things. There is also a possibility that there is still some air trapped in the radiator which of course wouldn't help.
So another update. As I have been thinking about this at work today I believe I have come up with a solution to potentially get better airflow, better rigidity from the case, make it cheaper, and allow for better compatibility of pc components. This unfortunately means though that I have to go back to the drawing board...
I believe though that this might be the best way to move forward. I want to go back to my original layout with the radiator mounted on its side. That way it can pull in cool air on one side of the case and exhaust the hot air out the other without ever heating up the 3D printed frame. The difference between my new concept and the original is that the motherboard and GPU will be mounted back to back. This will insulate the two from each other which is another concern I have with prototype one and will allow for a dual slot card in a case that is about 70-80mm in thickness vs the 95mm thickness of the current prototype. It will also allow for the use of standard 6.5mm motherboard standoffs vs the m3 screw approach of prototype one and hide them from view. With the reduction in thickness means a reduction in 3D printed material (in theory) which will decrease the cost of the frame. Now that I know the production capabilities of Ponoko laser cutting I can incorporate more laser cut parts into the design.
I do still plan on redoing the water loop in prototype one just to see what type of performance I can expect from a dual 80mm rad in ideal conditions. If performance still is not great then I might consider trying to incorporate a larger dual 92mm rad into concept 2. More updates to come!
I believe though that this might be the best way to move forward. I want to go back to my original layout with the radiator mounted on its side. That way it can pull in cool air on one side of the case and exhaust the hot air out the other without ever heating up the 3D printed frame. The difference between my new concept and the original is that the motherboard and GPU will be mounted back to back. This will insulate the two from each other which is another concern I have with prototype one and will allow for a dual slot card in a case that is about 70-80mm in thickness vs the 95mm thickness of the current prototype. It will also allow for the use of standard 6.5mm motherboard standoffs vs the m3 screw approach of prototype one and hide them from view. With the reduction in thickness means a reduction in 3D printed material (in theory) which will decrease the cost of the frame. Now that I know the production capabilities of Ponoko laser cutting I can incorporate more laser cut parts into the design.
I do still plan on redoing the water loop in prototype one just to see what type of performance I can expect from a dual 80mm rad in ideal conditions. If performance still is not great then I might consider trying to incorporate a larger dual 92mm rad into concept 2. More updates to come!
Well here is prototype two... The idea at least. Same width, thickness, but had to increase height by 1cm pushing volume to 5.3L.
I like this layout much better since you can now see and service all components without having to disassemble the PC. Only the top face and bottom face will be 3D printed which should drastically reduce printing costs. The question is whether or not laser cutting more parts will be cheaper then the 3D printing. I am betting that it is by a significant amount.
I like this layout much better since you can now see and service all components without having to disassemble the PC. Only the top face and bottom face will be 3D printed which should drastically reduce printing costs. The question is whether or not laser cutting more parts will be cheaper then the 3D printing. I am betting that it is by a significant amount.
Any updates?
Edit: Is that a wrap around bezel in the new design?
Edit: Is that a wrap around bezel in the new design?
Was this printed in nylon or polyamide? They have super high heat resistance
Not much to show at the moment as I haven't had time to create a new CAD model but the wrap around bezel is the new design direction I am moving towards. The new concept is to reduce the 3D printed parts to just the corners and then have 4.5mm thick acrylic panels for the side panels and bezel. This should solve the issue of the case being weak.
The test with 1/4" tubing in prototype one was unfortunately a failure as I simply could not bleed the system. So moving forward, prototype 2 will be getting larger in order to support off the shelf AIOs and off the shelf pump/res combos. The idea right now is to shrink the footprint to a square with dimensions around the 230mm mark and have a thickness of about 140-150mm and have a max volume under the 8L mark. This will allow the GPU to be traditionally mounted in the case and would not require a GPU riser. The increase in thickness also allows for 92mm AIO support and possibly 120mm AIO support. The increase thickness also allows for standard height CPU air coolers and coolers up to 120mm tall (more or less). I want to target a volume of around 6.7L but it could go as high as 7.5L depending on what I find during the design process.
It was printed in SLS nylon. Shapeways lists this material as heat resistant up to 165F which is not much considering what the air temp likely was coming off of the radiator.
Im seeing 163 c/325 f on the data sheet....
are we lookin at the same thing?
are we lookin at the same thing?
The material data sheets do specify a softening temperature of 325F but if you read the description of the material they state this
"Strong & Flexible plastics are heatproof to 80℃ / 176℉ degrees. Higher temperatures may significantly change material properties."
I have found this to be the case though my own testing and tinkering with the material. I usually try to keep any design max temperature around 165F to ensure no softening will occur.
Might sound silly but what was your loop order? What was the highest/lowest point in the loop?
So no more sub 5L?
Could you try doing sub 7L?
Id be fine with a 6-7L design if it allows for better compatibility, but I'm not very comfortable with an 8L case.
And is the prototype design above a no-go as well? You said it was ~5.3L and that seemed like an optimal design IMO.
Not at all! The highest point was the T line which fed to the lowest point which was the pump. From the pump it went through the GPU and then to the radiator. The pump just seemed to be to weak to properly bleed the system and the T line was to small to allow much any air out. You would likely have to use a syringe in order to force fluid through the system without the aid of the pump in order to properly bleed it. I came to the conclusion that it simply needs a reservoir and larger tubing in order to be practical.
I am targeting a 6.5-6.7L design for sure but as I said I have yet to build it in CAD so some dimensions might have to be increased in order to work. The concept 2 I posted could work for straight water-cooling with a dual 80mm rad only. However, I would like to open up the case to those that do not want to mess with water-cooling and simply want a decent air cooled experience which means fitting more then low profile coolers into the case. Plus having the GPU mounted to the motherboard removes the need for any additional support due to the flexible PCIE riser.
I also did a bit of digging through the forum and found the H2O project where there was an attempt at water-cooling a sub 3L build. While he had custom water blocks to allow for low profile operation, the radiator he used was about the same 80mm rad that I am using. He had temperatures comparable to my first test with the horizontal orientation of the radiator that I have for concept 2. This tells me that there isn't much to improve in design over prototype 1 in terms of temperatures which is why I want to move to supporting 92mm and possibly 120mm radiators.
How much bigger? could you go and grab some 7mm ID ptfe tubing and just work up from there, maybe getting a stronger pump to force everything through? The 5.3 case was so nice and I will be sad to see it disappear
I'd suggest finding a way to use a stronger pump before going to larger tubing. Your maximum flow-rate will always be limited by having g1/4 fittings.
Similar threads
