1st loop leaking : seeking for advice

Associate
Joined
31 Mar 2019
Posts
29
hi guys !

I’m reaching you on a friends advice. We started building my 1st custom loop on yesterday.

Our two challenges were : 1 - the case is small (Phanteks p350X) 2- I chose PETG tubing.


We spent all the day swapping blocks, installing everything and bending the tubes and while I was happy with the results, we had 2 leaking points : 1 at the CPU intake, 1 at the Radiator exit. I think that the problem was we were using very very short angles before the straight fittings, which therefore weren't sealed correctly.

02BD986C-177E-4693-AB7E-002BF6092F0D.thumb.jpeg.243986b865807af5e11fdbfb773b46ed.jpeg


So I came up with a few new ideas and I'd love to read your guys advices, improvement suggestions and other ideas.
Please note that I really like how the other tubes look like and I'm not wishing to change the loop order.


For the Res to CPU tube, I had 2 ideas :
1-A : switching the cpu fitting with a 90° fitting, and using the same path as the 1st attempt, deleting the short 90° before the fitting for a safer insertion. Drawback : Still a 180 + shift to manage.
EA571B30-D51B-48CF-B439-CE480F3613BF.thumb.jpeg.9c8c46c2e331d165d3b34324838db50a.jpeg


1-B : using 2 90° fittings and going above the other tube for an easier (IMO) one. Drawback : going to be close to the fan.
C36DD6DB-C536-4B6F-941D-4C31E471DBE6.thumb.jpeg.227895ee52281492314ee189158b9b0d.jpeg


1-C : using 2 90° fittings to swap 1-B and get an easier bend
350CB5CA-A35C-45B6-84AF-DA1843C37F26.thumb.jpeg.3f499d7e43e6fafd97332d1bc3cc88c7.jpeg


2 - For the Rad to res tube, I was planning to swap the rad fitting to a 90° one and go for something like this :
2A40CA00-1C3D-4D70-BF2C-A5D37A3EE448.thumb.jpeg.5137e422eca8fdccfd74f3ca0d9e4d6e.jpeg




Thanks guys for your time !
 
Last edited:
Yeah, tight 90 just at the fitting is hard! the biggest issue is usually the the tube becomes distorted where the fitting o-rings are, leading to issues.

I've had a little success getting rid of some 90 degree fittings in similar situations (this is 14mm XSPC PETG) using the forming tools and carefully applying heat away from the fitting end to keep the tubing from distorting and checking it carefully, but did waste quite a lot of pipe (even if fairly cheap) just to remove the necessity of having to use 90 degree fittings.

My own opinion if I was doing your setup for the first time:
- 1-A for sure, 90 degree fitting on the CPU and res Outlet, Having both tubes vertical on the CPU is my aesthetic preference.
- 2 x 90 degree fittings to make a 180 at the reservoir inlet, then a 90 on the rad. I've tried a few different things and aesthetically I find fittings mid tube a bit distracting.

For my own style, the only different routing I would make in your case would be to: (And just IMO of course!)
1. use 2 x 90's on the res inlet to put the inlet tube behind the reservoir and use the back rad fitting for that, this makes the GPU outlet to RAD line a bit straighter and requiring less 90 degree bends.
2. Move the T Piece right against the rad (use an inline m-m slim fitting) this keeps the drain at the rad itself.
 
Yeah, tight 90 just at the fitting is hard! the biggest issue is usually the the tube becomes distorted where the fitting o-rings are, leading to issues.

I've had a little success getting rid of some 90 degree fittings in similar situations (this is 14mm XSPC PETG) using the forming tools and carefully applying heat away from the fitting end to keep the tubing from distorting and checking it carefully, but did waste quite a lot of pipe (even if fairly cheap) just to remove the necessity of having to use 90 degree fittings.

My own opinion if I was doing your setup for the first time:
- 1-A for sure, 90 degree fitting on the CPU and res Outlet, Having both tubes vertical on the CPU is my aesthetic preference.
- 2 x 90 degree fittings to make a 180 at the reservoir inlet, then a 90 on the rad. I've tried a few different things and aesthetically I find fittings mid tube a bit distracting.

For my own style, the only different routing I would make in your case would be to: (And just IMO of course!)
1. use 2 x 90's on the res inlet to put the inlet tube behind the reservoir and use the back rad fitting for that, this makes the GPU outlet to RAD line a bit straighter and requiring less 90 degree bends.
2. Move the T Piece right against the rad (use an inline m-m slim fitting) this keeps the drain at the rad itself.


Hi !

Thanks for this useful feedback !
As for now, I totally agree with your opinion. I think I'm going to try 1-C because that's my favorite in terms of aesthetic but if I fail then I'll go with 1-A which is going to be very cool & easy to do :)
My problem for the second part is for the res to rad tube I won't be able to use a 90° because it'd be shifting the tube and make the bend much harder... :(


Again, for your last suggestions, you're totally right ! that's actually what I was initially planning. However, I then realized that I chose this res because of its space saving capacity. You can insert a DDC pump on its back, which I did, but now, I don't have enough room to drive a tube behind it.



Thanks again !
 
You could turn the cpu block through 180 degrees so the inlet is closest to the res. It won't affect temps at all. You will have to put a curve into the gpu to cpu tube but if I can manage it on my first ever PETG tubing attempt anyone can. You really need to have the tubing running straight and level going into the fittings to prevent leaks.
 
You could turn the cpu block through 180 degrees so the inlet is closest to the res. It won't affect temps at all. You will have to put a curve into the gpu to cpu tube but if I can manage it on my first ever PETG tubing attempt anyone can. You really need to have the tubing running straight and level going into the fittings to prevent leaks.

We thought it but it would have meant tilting the EK & velocity logo sideways as well which I'd have liked to avoid. However, we may end up doing this if we don't manage to solve it in other ways...

Thanks ! I think my main mistake was both applying heat too close to the tube end and cutting it too close to the turn beginning so they were not perfectly straight and therefore not correctly sealed... The 90° fittings should help A LOT :)
 
One more bit of advice:

Just double check if you get a very obvious initial leak that you've really gone to town chamferring/cleaning the tube end, lubricated all o-rings (including those in the fitting) with distilled water and that the tubing has definitely seated right in to the fitting.

You hear a lot of people complain how hard it is to get the hard tube into the fitting, and most often than not, I've found that they've either not chamferred the ends of the tube, or hardly touched it.. I use P1000 or P2000 wet/dry paper to really make the chamfer 'smooth' to the touch and get the least issues. My first build had the very first connection just pee out water everywhere, luckily I didn't even need to start the pump to realise, and that was just me not doing a good enough job.

I'd say in at least 50% of leaks I've had it's not been the bending to blame, just my lack of prep.
 
One more bit of advice:

Just double check if you get a very obvious initial leak that you've really gone to town chamferring/cleaning the tube end, lubricated all o-rings (including those in the fitting) with distilled water and that the tubing has definitely seated right in to the fitting.

You hear a lot of people complain how hard it is to get the hard tube into the fitting, and most often than not, I've found that they've either not chamferred the ends of the tube, or hardly touched it.. I use P1000 or P2000 wet/dry paper to really make the chamfer 'smooth' to the touch and get the least issues. My first build had the very first connection just pee out water everywhere, luckily I didn't even need to start the pump to realise, and that was just me not doing a good enough job.

I'd say in at least 50% of leaks I've had it's not been the bending to blame, just my lack of prep.


Thanks for the advice ! I actually use TT bending kit which includes blades for this purpose. However, I'm not a big fan of the result so I've been sanding most of the tubes. I think that the problem was here only the tube shape at the insertion. not a perfect O because of wrong heat application.
But I won't miss sanding every piece of the new tubes after cutting
 
Yeah, sanding is a thankless task, My aim is something like this (found a good image from another thread):
GQ57QvX.jpg


But as you say, distorted tubing is most likely the cause (you should be able to see it).

Saying that, it might be worth persevering, I decided to try it coming in/out the GPU block on the last build with no leaks:
7jnCLv4.jpg
 
Last edited:
Yeah, sanding is a thankless task, My aim is something like this (found a good image from another thread):
GQ57QvX.jpg


But as you say, distorted tubing is most likely the cause (you should be able to see it).

Saying that, it might be worth persevering, I decided to try it coming in/out the GPU block on the last build with no leaks:
7jnCLv4.jpg


That’s beautiful definitely worth the try !
Regarding the tip shape, that’s almost what you get with TT blade tool but it’s not that clean (might be a misuse) I’ll have to sand it again afterwards
 
Ok so this is how the first tube try looks like ! It seems to be perfectly sealed. When i pull nothing comes off.

IMG_0437.thumb.jpg.e049435c7d94c0ae5d4f5b34a9dbe510.jpg


Any thoughts ?



I'm probably going to do the same thing with the other tube because I enjoy how it looks like :)
 
If it doesn't leak, then in principle that's perfect!

Aesthetically, if had to be critical (and the camera might be distorting it) but it does look like the vertical section from Rad is not quite vertical and the top horizontal section is not quite horizontal.

Everyone has their own methods/processes for this and I'm very much at the beginning a bit myself (only 4 hard-lined systems to date),
My approach would have been based on this process:
uuuPBxg.jpg


1. I'd start with the black section, using a 180 degree form for a perfect even radius and a perfectly horizontal registration. I'd leave way more of a stub on the reservoir fitting part (to be cut to length later).
2. Inside the case, I'd offer up the 180 bend, see how far over the RAM I wanted it to give a nice straight entry to the reservoir fitting, then measured from the left most part of the 180 bend to a vertical line above the rad fitting (Cobbled together plumb line used to dangle over the rad fitting), this only needs to be rough
3. I'd then use that Measurement and offer up the 90 form tool to work out where that would need to start, and mark that point with some tape on the tube.(Start of the red bend)
4. I'd then have to approximate the angle between rad upright and the 30-45 degree angle of the 180 bend, with that I'd then bend the red 90 degree bend.
5. Offering the whole piece up, I would then but the gray section to length first, then trim the black section to length and fit..

TO be fair, sometimes with other peoples builds I've wasted loads of pipe going for perfection, but some people are super-fussy and I've had to do things like this to get it all perfect. However, my own build has a couple of imperfections that I just live with as I ran out of tubing (only just had enough for one attempt!)
 
It was definitely not a correct angle, I modified it after shooting the picture :)


I let it ran for an hour using demineralized water : no leaks, and I drained the bubbles.
Then I plugged all in and booted, then felt a bit concerned.

Side note : CPU is an i7-9700k, not set to be overclocked

1st : my pump didn't start when plugged to the mobo. It thermal throttled and shut down automatically. That's "fine" : I plugged it using MOLEX, and it now runs just fine.

2d : I went to the bios first and gosh ! The CPU frequency is very high ! So are the temps which kept increasing to 82C in 5mn approx. I then shut it down by safety. Are those frequencies normal while just in bios ? I would have expected the CPU to be idle at low fréq and temps.

IMG_0441.jpg.ab4478b74727c61cf03fe9d990199de6.jpg



What troubleshooting process should I follow ?





Thank you !
 
1. All the pumps I've had must be powered from a MOLEX / SATA Power and the connection to the Mobo is only for PWM speed control, so that might explain this, at some point you might want to have a profile that lowers pump speed at lower coolant temps, but most pumps spin at 75-100% with just the MOLEX/SATA power connected.

2. I'd boot in to windows, use any of the diagnostic tools (e.g. HWINFO) to get more detailed stats) and use task manager to check it isn't underload. BIOS' can be weird, and if it was only 79 degrees on all cores, that looks excellent! You could reset the bios just to be sure and start from a known starting point, but get in to windows, that's a better environment to debug.
 
1. All the pumps I've had must be powered from a MOLEX / SATA Power and the connection to the Mobo is only for PWM speed control, so that might explain this, at some point you might want to have a profile that lowers pump speed at lower coolant temps, but most pumps spin at 75-100% with just the MOLEX/SATA power connected.

2. I'd boot in to windows, use any of the diagnostic tools (e.g. HWINFO) to get more detailed stats) and use task manager to check it isn't underload. BIOS' can be weird, and if it was only 79 degrees on all cores, that looks excellent! You could reset the bios just to be sure and start from a known starting point, but get in to windows, that's a better environment to debug.

That’s very interesting ! I’m going to give it a try because right now it runs at full speed and is therefore quite noisy. Couldn’t find anything about this in the product description but my mobo manual says which pins are for speed control so I may be able to check easily. Thanks !

Regarding 2., all the temps are fine now that I ahem... peeled the plastic cover off the block...
they were fine on idle cores but climbing to 100° on active cores, then switched cores and so on, which is relevant for this problem... :)
 
@Insorior Looking at the pictures from your other thread, the one that you posted about the "problem", I think you have the cpu block backplate the wrong way around (180 degrees out) as you should see the three screws for the socket clamping mechanism through the three holes in the backplate. Potentially this could put uneven pressure on the socket and may damage it. At least you wouldn't have to drain the system to swap it through 180 degrees.
 
Back
Top Bottom