Project: TJ07 Revived

Harpss1ngh · 31 Jul 2016 at 20:17

Yeah true, The Aquacompuiter D5 is connected via Aquabus and so is the Fill Level Sensor in the Aqualis D5 Top.

I'm also going to set up an alarm for the Flow Sensor, I'm just waiting for the Flow Sensor cable as I didn't realise it has 4 wires otherwise I'd have used the RPM cable as I have two of those for some reason!

Harpss1ngh · 31 Jul 2016 at 20:45

I'm a little stuck with the tubing as you can see from the pics below there are a lot of tight angles. I'm running short on 45 and 90 degree fittings and trying not to use too many of them wherever I can but it's looking a bit difficult.

Any suggestions? Photo's might work better than that horrible diagram I tried to draw earlier, hopefully the pics show things in better perspective...

After shortening the bottom rad's hose I managed to get the bottom rad further in the case about an inch so that drain port shouldn't stick out as much:

I've connected the GPU WB Outlet to the CPU WB Inlet and trying to plan how the rest will go, as you can see things are a little tight. I used 90 degree fittings in my old water loop and I thought that was too restrictive but looking at this again I might have to resort to it again or try and do bigger loops to avoid kinking the tube as it's quite tight:

Ignore the mess for now, I'm going to sort that out right at the end

. I'm going to leave those USB cables connected to the board and kept hidden away so when I need to connect the Aquabus devices via USB for firmware or calibration or something I can do that easily (Unless of course Aquacomputer finally allow us to do all that via Aquabus instead!)

That's not the only issue I need to solve, the Aqualis res/top's outlet (the left one) needs to go to the Koolance's Inlet (The left one behind the Res) I can't seen to find a easy way to tube it. I also have the Mesh Filter, the 240mm rad in the floor of the midplate as well as the tube coming up from 480mm that's under the GPU's, not to mention the flow sensor (not in the pic atm).

Arghhhhhhhhhhhhhhhhhhh :eek:

Cenedd · 1 Aug 2016 at 09:23

Harpss1ngh said:
I'm just waiting for the Flow Sensor cable as I didn't realise it has 4 wires otherwise I'd have used the RPM cable as I have two of those for some reason!

If it's the high flow non-usb version, it's only 3 pin on the sensor end but they're a lot bigger than fan pins. Not sure why they don't supply the required cable with it!

Cenedd · 1 Aug 2016 at 10:02

For the GPU, can you not put the inlet where you have the temp probe? Alternatively come in from the bottom from the 'basement'?

I'll try to have a look at the rest when a bigger screen is available (mobile not optimal

) but the first thing that occurs is that if your Aqualis res is feeding the Koolance res AND it's below it, doesn't that negate all/most of its benefit?

Harpss1ngh · 1 Aug 2016 at 11:59

Cenedd said:
If it's the high flow non-usb version, it's only 3 pin on the sensor end but they're a lot bigger than fan pins. Not sure why they don't supply the required cable with it!

Correct for some reason I thought the RPM wire would work (perhaps I read that somewhere...). All these different cables required (and not included must generate Aquacomputer some nice revenue lol...)

Cenedd said:
For the GPU, can you not put the inlet where you have the temp probe? Alternatively come in from the bottom from the 'basement'?

I'll try to have a look at the rest when a bigger screen is available (mobile not optimal ) but the first thing that occurs is that if your Aqualis res is feeding the Koolance res AND it's below it, doesn't that negate all/most of its benefit?

There's no bottom inlet (afaik) otherwise that would be ideal. Yeah I can go from the side, the reason I didn't was because of the tight proximity to the 8 pin connectors in the GPUs and the compression fittings are quite large but I might try and do it, maybe with G1/4 extenders.

I clearly didn't plan the reservoirs properly lol. I didn't put the koolance res lower because the aquacomputer d5 top gets in the way of the tubing as I have to bring it behind the reservoir (and without the tubing pushing against it) so I had to raise it higher. The mesh filter and the steel plate the aqualis is mounted on also stop me going any lower.

The height of the pumps won't have much negative effect as having a lot of angled fittings does. The D5 has a 3.9m pressure head, thats measured by how much pressure it can generate against gravity (upwards) so should be fine feeding the other two pumps and shouldn't be too restrictive.

If I ever get this completed I'll know for sure via the flow sensor and the water temps of the impact, although having 3 pumps in the loop it probably will be negligible lol. Youtube videos from the various popular channels show it should just work fine.

I probably won't do anything further in this case after this project, the next one would have to be a giant caselabs! I would look forward to those waterblocks on Skylake that cover the CPU and the Motherboard with one single block too!! It's been a pain having 8 inlet and outlets to work with in a tight space!

Cenedd · 1 Aug 2016 at 12:24

Ah, ok. So it's parallel with in and out from the top. Looks like you could use the side as an input but you'd have to either stick something like a 90 on it or a double 45 to clear the power cables. An off-the-wall idea is you could flip the bridge upside down and have in and out at the bottom through the floor. The engraved writing would be upside down though.

The angled fittings don't have that much effect on temps - I think someone tested it and worked out that it would take about 20 right angles before it had more than a degree difference in temp. It was probably Martin's Liquid Lab - which is worth a Google anyway. With three pumps it's not going to be an issue. What I meant was more that the res wasn't having gravity feed the column of water to the pump....but I missed the pump on the bottom of the res :rolleyes:

With that pump there, you're going to be pumping the water from that res into the other res. That effectively does nothing as the water just collects in the Koolance res waiting for the pumps to pump it. Would you be better (and would it simplify tubing) to have that pump at a different point in the loop (I'd say after the double pump to avoid one starving the other) or as an alternative, make it a dual loop - the single pump and res for the CPU and the dual for the gfx?

Harpss1ngh · 1 Aug 2016 at 12:37

Cenedd said:
Ah, ok. So it's parallel with in and out from the top. Looks like you could use the side as an input but you'd have to either stick something like a 90 on it or a double 45 to clear the power cables. An off-the-wall idea is you could flip the bridge upside down and have in and out at the bottom through the floor. The engraved writing would be upside down though.

The angled fittings don't have that much effect on temps - I think someone tested it and worked out that it would take about 20 right angles before it had more than a degree difference in temp. It was probably Martin's Liquid Lab - which is worth a Google anyway. With three pumps it's not going to be an issue. What I meant was more that the res wasn't having gravity feed the column of water to the pump....but I missed the pump on the bottom of the res With that pump there, you're going to be pumping the water from that res into the other res. That effectively does nothing as the water just collects in the Koolance res waiting for the pumps to pump it. Would you be better (and would it simplify tubing) to have that pump at a different point in the loop (I'd say after the double pump to avoid one starving the other) or as an alternative, make it a dual loop - the single pump and res for the CPU and the dual for the gfx?

Yeah Martin's Liquid Lab is awesome.

If I have all 3 pumps in the loop the weaker one would need to feed the stronger ones but guess I could still stick something in between them, that would help the angle of that tubing and I've thought about that but not sure what I could put there. Maybe the visual flow indicator? 90 degree fitting from the Aqualis OUT port, (around to the right of the Aqualis instead of left and behind it), going to the visual indicator which then goes straight up to the koolance's inlet?

The aquacomputer Mesh filter is also at an annoying angle, I'd need to figure something out with that too.

I could do dual loops too (the koolance itself can handle dual loops but I would miss out on the Aquacomputer fun array of sensors), just the CPU with the top 240mm rad with the motherboard and GPU's on the 2nd loop (the mosfets dump a lot of heat into the loop so would want it with more rads) but I'm not sure whether or not that two loops would be too complicated to plumb up in the tight space.

This case is showing its age now compared with monster cases that have plenty of space to do all this while I'm here spending months on trying to cram it all in here!

Cenedd · 1 Aug 2016 at 13:05

The dual loop is just that it suddenly occurred to me that you had two of everything.
If you have the single pump before the double you run the risk (I suspect) of the dual pump's res emptying. If you it after, it's likely to restrict the dual pumps. This assumes that the flow rate of the two is higher and that the closing of the loop (apart from the pressure relief valve on the XT res) doesn't negate this. I'm theorising here and I could be totally wrong.

Harpss1ngh · 1 Aug 2016 at 13:39

Ah I understand what you mean.

I thought about this before too and reading forums where people had done something similar they went into some sciencey stuff (that's the technical phrase

) explaining that it would all balance out. Provided the pump before the two keeps up with the two pumps in front of it and are the same type, which it will as they are all D5's. I just need to keep them all at the same speed. They all will perform with the same flow speed, it's the head pressure that will be greater in the dual bay compared with the single bay.

The flow (pump) speed should remain the same across all 3 pumps and that will reduce any extra wear on the weaker D5 as well as ensuring the res doesn't empty.

I think the same principle applies for anyone thinking of sticking a different type of pump (DDC, etc) with a D5, that wouldn't work because they're rated differently and work differently and there'd be an inbalance

I can't remember where I read it now but it made sense. Once I get everything together I will test it out and determine the settings to keep. I won't have the Aquacomputer D5's speed change according to temp or anything, I'll just match it to the Dual D5's in the Koolance Res.

Cenedd · 1 Aug 2016 at 13:52

Worst case, you could run the doubles a bit slower than the single if you find one res emptying.

Harpss1ngh · 1 Aug 2016 at 15:27

Makes sense!

I'm going to buy a bunch of different rotaries and try and get it all fitted.

I'm thinking of those angled adaptors with 2 45 degree angles for a couple of those tight awkward bends. I have some chrome angled fittings I'm still using so might take this opportunity to replace them all with black bitspower stuff

Cenedd · 1 Aug 2016 at 15:59

Bitspower is good but pricey. Barrow is very similar but without the logo and much cheaper. Worth a look.
Koolance also do some useful fittings. Specifically a 30° fitting (ADT-DXG30-BK) a double-rotary 60° (ADT-DXG60-BK) and a 3-segment 90° fitting (ADT-DXG90-BK).

Harpss1ngh · 1 Aug 2016 at 16:41

Cenedd said:
Bitspower is good but pricey. Barrow is very similar but without the logo and much cheaper. Worth a look.
Koolance also do some useful fittings. Specifically a 30° fitting (ADT-DXG30-BK) a double-rotary 60° (ADT-DXG60-BK) and a 3-segment 90° fitting (ADT-DXG90-BK).

I just bit the bullet and spent a tiny fortune in fittings lol. I love bitspower, they look and feel quality, they never leak and the O rings they use are always nice and snug! The rotaries are also very stiff (heh) so they don't move all about loosely. I have a bunch of xspc chrome rotaries I will replace because with a bit of stress they start to very slowly leak, although they don't completely fail. I've noticed blue coolant coloured stains on my case from those dripping over time but slow enough not to notice or cause problems and not noticeable from the usual res top up when the coolant evaporates.

Koolance are very good too. I only pick EK or XSPC when I can't find Bitspower or Koolance in stock, even though they might be fine, I just don't want to take chances.

Also I've put huge strain on my X58 Asus waterblocks for 4 years with tight loops adding stress and the motherboard nor the fittings have failed!

Harpss1ngh · 3 Aug 2016 at 15:10

Ok so I've been playing around with the angled fittings, I can now make that tight fitting of the two reservoirs together without going around behind the Aqualis res (so that's one challenge completed).

I'm now trying to experiment with tubing direction with the Mesh filter, the rear 240mm rad's bulkhead, the top 240mm rad, the Out port on the Koolance res, the In port on the Aqualis res and both of the the flow meters! They're all in close tight proximity so making them all loop is a PITA! With all the screwing and unscrewing I'm doing to experiment with which loop would be best I hope I'm not wearing down the Plexi and Acrylic threads!

I'm trying to put the Mesh filter right before the AQC pump so that it catches any crap in the tubing before it kills my pumps but it's becoming difficult to make that angle. Is having the mesh filter right before that Res the best way to do it?

I know the flow sensor should have straight connections but I have nothing in front of it that I can connect to easily, the easiest I can see is having it go straight into the mesh filter. So the bottom 480 rad's tubing that comes through the midplate below the GPU > Flow sensor > Mesh filter. But the filter is off angle, so I will have to use a tight fitting with a tiny 1-1.5" tube between them. Then from the mesh filter I need to get the tube to run into the Aqualis D5 top inlet port, again another awkward angle.

I also have the 240mm Radiator in the top I need to get into the loop. There are two ways I can seem to achieve this, either have it connected to the Koolance's Out port and then to one of the 8 Motherboard barbs, or have it connected to the bottom 240mm radiator (via the bulkhead in the front of the case) and then to the motherboard, without hitting the 5.25" blu ray drive or the koolance res.

Then there's the loops in the motherboard that are a pain too

The pic of my case layout from earlier is below and I'll mark out the points I need to tube up somehow..

Hopefully you can see my frustration here!!:

Cenedd · 3 Aug 2016 at 23:15

Spin the top rad by 180 degrees and have it between the VRM blocks. Something like GPU > top left VRM block > top rad > top right VRM block > CPU > southbridge > res
or wherever you need to get to. Can always got to one res/pump through something else and then through the other - probably preferable to be honest.

The MPS flow sensors are no moving parts, based on differential pressure. They're picky about having straight input and output or can be inaccurate. You've got the high flow with a spinning impeller haven't you? These aren't fussy as far as I know but they are best mounted flat or the bearing the impeller spins on can be a bit noisy.

Filter: that's where I've planned to put mine. Seems the most logical to me but I suspect that gunk is generated evenly across the loop so it probably makes little difference. It tends to get trapped in the pins of the blocks so there's an argument it should go before those too.

Harpss1ngh · 3 Aug 2016 at 23:20

Cenedd said:
Spin the top rad by 180 degrees and have it between the VRM blocks. Something like GPU > top left VRM block > top rad > top right VRM block > CPU > southbridge > res
or wherever you need to get to. Can always got to one res/pump through something else and then through the other - probably preferable to be honest.

The MPS flow sensors are no moving parts, based on differential pressure. They're picky about having straight input and output or can be inaccurate. You've got the high flow with a spinning impeller haven't you? These aren't fussy as far as I know but they are best mounted flat or the bearing the impeller spins on can be a bit noisy.

Filter: that's where I've planned to put mine. Seems the most logical to me but I suspect that gunk is generated evenly across the loop so it probably makes little difference. It tends to get trapped in the pins of the blocks so there's an argument it should go before those too.

I can't flip the top rad that way as in the corner there is a triangular bracket that holds the backplate to the motherboard plate. I don't think cutting it down is a good idea either with the amount of weight hanging on that thing! My motherboard with all those blocks fitted is very heavy! I'll look at it again, there may be something I can do. I moved the membrane to a different inlet on the top res so I think I can swing some tubing on a 90 degree angle towards the motherboard using a couple extensions.

Watch this space

Harpss1ngh · 4 Aug 2016 at 13:42

Cenedd said:
The MPS flow sensors are no moving parts, based on differential pressure. They're picky about having straight input and output or can be inaccurate. You've got the high flow with a spinning impeller haven't you? These aren't fussy as far as I know but they are best mounted flat or the bearing the impeller spins on can be a bit noisy.

Filter: that's where I've planned to put mine. Seems the most logical to me but I suspect that gunk is generated evenly across the loop so it probably makes little difference. It tends to get trapped in the pins of the blocks so there's an argument it should go before those too.

Yeah I have the high flow with the impeller! I've heard the noise on YouTube. If I hear it i'll try coating it in sound dampening foam lol. I have a ton of the stuff left over. I have 3 D5's I can crank up to block it out as a last resort

Yeah that's the problem with the filter placement, everything can gunk up and everything can get destroyed by it, depends on what you cherish the most I suppose lol. 3 expensive D5's and two reservoirs. Toothbrush and vinegar for the rest

I'm using X1 coolant instead of pastel anyway so hopefully I have less gunk and staining problems than I did last time.

Also, this is my planned loop now! It looks to be achievable having had a look last night:

That blue box is the visual flow meter. I've got it screwed into the GPU via a bitspower rotary extension plugged into a male to male xspc extender. It's a bit wobbly for some reason, I'm worried the fitting might fail over time and leak everywhere, the stiffness could damage the plexi threads too, what do you think?

It just seems like the perfect place to put it.

Cenedd · 4 Aug 2016 at 14:03

It concerns me that pumping into the Koolance res is going to completely waste the power of that pump. Could you go GPU > VRM > Southbridge > CPU > VRM > Top rad > top centre inlet of the Aqualis (assuming you want fountain) then go out of that through your lower rads? The filter and flow meter either on the way up or down from the Koolance.

The spinner it depends how much force it's applying to the sli bridge. From earlier tests, it's quite hard to break the thread in acrylic by pulling it straight out, but reasonably easy by pulling it sideways - like a lever. The more weight the spinner and the tubing above it exerts, the more risk...but you'd have to judge whether it's an issue by feeling it to be honest - could be fine.

There is a male to male rotary (without the rotary, it would be guesswork at to which way the spinner faced) by Bitspower: https://www.overclockers.co.uk/bits...ng-male-to-male-rotary-adapter-wc-141-bp.html
XSPC do one that's probably cheaper but OCUK don't seem to have it.

Harpss1ngh · 4 Aug 2016 at 14:23

Cenedd said:
It concerns me that pumping into the Koolance res is going to completely waste the power of that pump. Could you go GPU > VRM > Southbridge > CPU > VRM > Top rad > top centre inlet of the Aqualis (assuming you want fountain) then go out of that through your lower rads? The filter and flow meter either on the way up or down from the Koolance.

The spinner it depends how much force it's applying to the sli bridge. From earlier tests, it's quite hard to break the thread in acrylic by pulling it straight out, but reasonably easy by pulling it sideways - like a lever. The more weight the spinner and the tubing above it exerts, the more risk...but you'd have to judge whether it's an issue by feeling it to be honest - could be fine.

There is a male to male rotary (without the rotary, it would be guesswork at to which way the spinner faced) by Bitspower: https://www.overclockers.co.uk/bits...ng-male-to-male-rotary-adapter-wc-141-bp.html
XSPC do one that's probably cheaper but OCUK don't seem to have it.

I was just looking for one in black. I have that one in non-rotary. Rotary in that looks good and it's bitspower so it will not fail me! I have some old BP fittings and even being about 5 or 6 years old they look in mint condition.

I can use the top center inlet of the res? I didn't know that was an option! I thought for the fountain to work I can only use the D5 top inlet and outlet as I thought the coolant goes up inside the inlet and comes out through the holes to the side of the nano borosilicate glass. I might just be misinterpreting the german to english translation on the manual.

EDIT: Just saw a YouTube video where someone had done exactly that - going through the top and it still works. What's the middle shaft for then? I guess that's if you want to use the bottom inlet so the water goes up the middle? Had another look, I suppose what happens is, fi you use the top, the middle shaft fills up and the water spills over the top through the holes.

I might just end up creating a temporary loop, fill it with DI Water to test how it works! In fact that's an awesome idea, could test the Koolance Res with it while I'm at it and just make sure that the pumps are running fine.

Harpss1ngh · 4 Aug 2016 at 16:38

I think I might just add a second flow sensor! I'm itching to add another aquabus device! I have the rtc module which gives me an extra aquabus high port. I'm guessing it'll have to be an mps as there's only one flow port on the aquaero. I will take your suggestions as it will work well in my setup and with a 2nd flow sensor i can measure flow on both res/pumps.