Mini-Splash, about a Gallon water cooled SFF gaming PC
- Thread starter Goatee
- Start date
Hopefully!
************ Update 5 incoming ************
So this update is a bit different, no pictures of hardboard models, I will be taking the Pump / Block apart to figure out how I could do cable routing. Now I could just have taken the picture from the instruction manual to figure things out, but where is the fun in that?
This is the block turned upside down, screws removed.
Next you see the inner pump workings to figure out where the water flows in. Red in, Blue out.
This is important as it shows there are two ports and both feed the central reservoir.
I've added the flow plate back in now and you can see where the pump pushes the water against the copper plate (not pictures) which sits against the CPU. The pumps drives the water out in a spiral pattern to fill the microchannels. The fluid is then driven back out the exit of the pump block.
Based on that I now know I can put a fill port at the bottom of the loop directly into the pump block and the inlets and outlets I previously identified are correct.
I was able to remove the scratched Fractal pump logo.
I quite like the plain look.
Already existed, I wanted to make sure it fed the same locations as the other in port, which it does. It’s possible to have the pump setup with feeds on opposite ends too if required. It’s really nicely made.
***** Update 6.5 ******
A load of parts have arrived or are due imminently!
1) A shorter x4 pice powered riser has arrived, unfortunately its just about 15mm too short! Doh! So I will need to use a 90 degree riser in there, which should arrive shortly.
2) The special waterblock for the 1070 katana has arrived from China. Its nicely made and I took some photos in the garage before I fitted it to the block but they didn't come out that well. The big brucie bonus is that due to the design of the Katana (it has VRM's at the IO end of the card, with the GPU towards the back near the 8 pin power connector) the water flow is the opposite way round to flow over the GPU chip first. This means I will have a simpler loop and no modding or complicated tubing routing required. Apologies for the lack of front money shot of the block, that came out as blurred as a photocopier print after a Christmas party.
3) Just got a dispatch note from the manufacturer of the case components, only 2 1/2 weeks after I ordered it on a 3 day service
This is where I'm up to.
Got some interesting hard tubing routings to complete and figure out why the mosfet to switch the hdplex is playing silly buggers.
My amateur wiring diagram I am using for anyone interested.
A load of parts have arrived or are due imminently!
1) A shorter x4 pice powered riser has arrived, unfortunately its just about 15mm too short! Doh! So I will need to use a 90 degree riser in there, which should arrive shortly.
2) The special waterblock for the 1070 katana has arrived from China. Its nicely made and I took some photos in the garage before I fitted it to the block but they didn't come out that well. The big brucie bonus is that due to the design of the Katana (it has VRM's at the IO end of the card, with the GPU towards the back near the 8 pin power connector) the water flow is the opposite way round to flow over the GPU chip first. This means I will have a simpler loop and no modding or complicated tubing routing required. Apologies for the lack of front money shot of the block, that came out as blurred as a photocopier print after a Christmas party.
3) Just got a dispatch note from the manufacturer of the case components, only 2 1/2 weeks after I ordered it on a 3 day service
This is where I'm up to.
Got some interesting hard tubing routings to complete and figure out why the mosfet to switch the hdplex is playing silly buggers.
My amateur wiring diagram I am using for anyone interested.
I have been busy over the past couple of weeks, despite the lack of updates I have been doing alot of experimenting.
One of the major challenges I have encountered was with the water cooling loop. I was set on using hard tubing (cause you know reasons) and this set me back a good week while I played with different configurations, tried bending about a million different ways.
I managed to finally get the loop to just about fit but the way it needed to go meant I couldn't get a good seal on the fittings and my pressure tests kept failing. Here is half the rejected hardline (the other half went in the bin before I thought I should document my epic failures)
So I reluctantly moved over the some barbs which give the added benefit of allowing a res and making the loop take about 30 seconds to fit! Woohoo.
My second issue was that as I started to fully assemble the case, the 3mm thick acrylic panels bowed due to the weight of the densely packed parts. I decided to move over to Aluminium panels. These were orders and arrived yesterday. So I have now started to cut my back plate,
After a reasonable amount of action from a selection of needle and big boy files they were complete. I appear to have quite a strong right arm as it was all done my hand. Who's have thought!
Anyway, I needed to add some way of getting the air in he case at the front. Initially I went with the traditional drill a 80mm hole in the front, but 3mm aluminium didn't really like my hole saw (of my corded drill) and the finish was great.
I then remembered seeing someone who have used a template to cut a perforated pattern in the front of a case so I unashamedly copied the idea (credit would be given if I could remember where I saw it).
Its a piece of pegboard, that i used as a guide to avoid the drill bit wandering.
The finished front panel, along with res / pump / CPU block combo.
One of the major challenges I have encountered was with the water cooling loop. I was set on using hard tubing (cause you know reasons) and this set me back a good week while I played with different configurations, tried bending about a million different ways.
I managed to finally get the loop to just about fit but the way it needed to go meant I couldn't get a good seal on the fittings and my pressure tests kept failing. Here is half the rejected hardline (the other half went in the bin before I thought I should document my epic failures)
So I reluctantly moved over the some barbs which give the added benefit of allowing a res and making the loop take about 30 seconds to fit! Woohoo.
My second issue was that as I started to fully assemble the case, the 3mm thick acrylic panels bowed due to the weight of the densely packed parts. I decided to move over to Aluminium panels. These were orders and arrived yesterday. So I have now started to cut my back plate,
After a reasonable amount of action from a selection of needle and big boy files they were complete. I appear to have quite a strong right arm as it was all done my hand. Who's have thought!
Anyway, I needed to add some way of getting the air in he case at the front. Initially I went with the traditional drill a 80mm hole in the front, but 3mm aluminium didn't really like my hole saw (of my corded drill) and the finish was great.
I then remembered seeing someone who have used a template to cut a perforated pattern in the front of a case so I unashamedly copied the idea (credit would be given if I could remember where I saw it).
Its a piece of pegboard, that i used as a guide to avoid the drill bit wandering.
The finished front panel, along with res / pump / CPU block combo.
Nice project, but why you dont use a normal itx mainboard? With 45mm high will fit a normal itx board too. So you can use full pcie x16 and more options on mainboard, ram and cpu parts at the same dimension. I am very exciting of the temperature in games.
It was one of the options I thought about, but that would require running a 24 pin atx power cable and a x16 riser cable. Doing both of those would increase the size of the case. The x16 cable fouls the back of the cpu bracket so would add 3-5mm to the case width.
I have no idea how I would fit a 24 pin in.
Thin mini-ITX has no component that could block the airflow: RAM is parallel to the PCB, no protruding heatsink, very few connectors, etc.
Thanks,
I had a play bonding some of the panels using epoxy, but I don’t like how they look.
Will go back to the 10mm cubes.
All the panels need some holes drilling then I can get the sander out.
This is looking awesome!
I feel like SFF custom watercooling has been missing a reliable CPU block/pump combo. The Swiftech Apogee has been around for a while, but I've never seen anything good said about it. Other brands need to be removed from an AIO kit if you want to use them like the one your using, or the Alphacool Eisbaer, and Raijintek TRITON.
I'm sure this is just a result of low demand for such a part in the custom watercooling world, but it's still lame.
Edit: Thinking about this a bit more, while the AIO manufacturers could sell their cpu block/pump combo as a stand alone unit, they'd have to clarify it's limitations and further, they'd have to compensate for people installing them in the wrong environment due to being weaker than standard D5 or DDC pumps. Still lame, but I get it.
I dig the custom SFF design here, but I'm much more interested in:
1. how well that CPU Block/pump combo holds up to both waterblocks and the extra 90 degree turns.
2. how much heat that 2x80mm Rad can effectively dissipate.
Both questions I suspect won't be answered until this thing can be turned on and stress tested; I'll be sure to stick around until then!
I feel like SFF custom watercooling has been missing a reliable CPU block/pump combo. The Swiftech Apogee has been around for a while, but I've never seen anything good said about it. Other brands need to be removed from an AIO kit if you want to use them like the one your using, or the Alphacool Eisbaer, and Raijintek TRITON.
I'm sure this is just a result of low demand for such a part in the custom watercooling world, but it's still lame.
Edit: Thinking about this a bit more, while the AIO manufacturers could sell their cpu block/pump combo as a stand alone unit, they'd have to clarify it's limitations and further, they'd have to compensate for people installing them in the wrong environment due to being weaker than standard D5 or DDC pumps. Still lame, but I get it.
I dig the custom SFF design here, but I'm much more interested in:
1. how well that CPU Block/pump combo holds up to both waterblocks and the extra 90 degree turns.
2. how much heat that 2x80mm Rad can effectively dissipate.
Both questions I suspect won't be answered until this thing can be turned on and stress tested; I'll be sure to stick around until then!
Last edited:
Actually, the Eisbaer block/pump is available to buy separately, as is the GPX Pro for GPUs.
Neat project btw.
I didn't know that was available. In the end a used AIO cost much less but that's good to know I could pick up something comparable if required.
Anyway, another update!
I got the new cut panels bolted together.
When reattaching the HDPLEX I noticed one of the new radiator mouths fouls the HDPLEX. Ooops, I guess I should have picked that up in my extensive 3d modelling work prior to starting the project (If I had done any!).
I added the GPU back in, you can see how tight its all getting and thats before I get to the wiring.
At least the GPU 8 pin has some clearance.
I then added the front side components.
I was really pleased with how some of the fittings lined up
Final money shot with all the tubing fitting.
Notice how the res is sitting at an angle. I think the fitting isn't engaged tight as I am also getting a slight leak when pressure testing. So will need to strip it all down and add some PTFE tape.
I got the new cut panels bolted together.
When reattaching the HDPLEX I noticed one of the new radiator mouths fouls the HDPLEX. Ooops, I guess I should have picked that up in my extensive 3d modelling work prior to starting the project (If I had done any!).
I added the GPU back in, you can see how tight its all getting and thats before I get to the wiring.
At least the GPU 8 pin has some clearance.
I then added the front side components.
I was really pleased with how some of the fittings lined up
Final money shot with all the tubing fitting.
Notice how the res is sitting at an angle. I think the fitting isn't engaged tight as I am also getting a slight leak when pressure testing. So will need to strip it all down and add some PTFE tape.
Last edited:
Nice job !
Are you using Bitspower 1/4 ID fittings ? Where did you get them ?
Regarding your leak issue, your hose is clearly forcing on your fittings.
You could relieve some tension by giving the right shape to your hose.
To do that :
- draw the bend section and print it to use it as a template.
- take a 6 mm electric cable and put your hose on it. Use some liquide soap if needed.
- bend both the hose using your template.
- put it into boiling water for a couple of minutes then in cold water for a couple of minutes.
- remove the electrical cable.
Are you using Bitspower 1/4 ID fittings ? Where did you get them ?
Regarding your leak issue, your hose is clearly forcing on your fittings.
You could relieve some tension by giving the right shape to your hose.
To do that :
- draw the bend section and print it to use it as a template.
- take a 6 mm electric cable and put your hose on it. Use some liquide soap if needed.
- bend both the hose using your template.
- put it into boiling water for a couple of minutes then in cold water for a couple of minutes.
- remove the electrical cable.
They are just come cheapie 9mm ones from amazon. Hose is 8/10mm food grade stuff.
https://www.amazon.co.uk/gp/product/B07F8SQGZ7/ref=oh_aui_detailpage_o05_s00?ie=UTF8&psc=1
I spent most of this evening trouble shooting the leak, turns out it wasn't the RES but was actually a missing seal from one of the other fittings.
I recut the tubing and secured it with black adhesive heatshrink and double ties. It dont move!
I also started playing around with a mesh to provide filtering on the twin intake fans.
Looks like this when encloded.
https://www.amazon.co.uk/gp/product/B07F8SQGZ7/ref=oh_aui_detailpage_o05_s00?ie=UTF8&psc=1
I spent most of this evening trouble shooting the leak, turns out it wasn't the RES but was actually a missing seal from one of the other fittings.
I recut the tubing and secured it with black adhesive heatshrink and double ties. It dont move!
I also started playing around with a mesh to provide filtering on the twin intake fans.
Looks like this when encloded.
