Radiator failure

Associate
Joined
2 Oct 2009
Posts
829
Location
Belfast, UK
A while ago my custom loop started very slowly dropping water levels.

The loop consisted of 2x Thermochill PA 120.3 radiators (pretty ancient, 9+ years old but always looked after) and a Laing DDC Ultra 18W. The fluid was distilled water and a kill-coil was used to inhibit any growth.

I checked it many times to try and isolate any potential leaks and never found anything, so I kept it running but kept an close eye of the fluid levels, which indeed continued to drop (the system would lose about 150ml of fluid over 2-3 months, so it seemed it indeed had a small drip-leak.

Eventually I said enough was enough and did a pretty extensive leak-test with the components off, and probably because the loop was now cold I found signs of evaporated water drops on a hard-drive bay, so luckily the system leak had not occurred near any components.

I then did a test and sure enough the top mounted Thermochill radiator had a very small drip leak that had obviously worsened over time. So I redid the loop, removing LQ for the cpu and just using the 1 good radiator for my GPU, and I decided to use Distilled + EK-Cyrofuel for the new loop, removing the kill-coil.

The water itself of the loop was clear, with a bit of black tarnishing on the copper blocks, but nothing particularly noticeable.

The strange part is I isolated the leak point in the radiator to inside the fin-pipes. At first I thought I might have punctured a pipe with installation of radiator fans/screws, but it turned out the leak coming from deep in the middle of the radiator away from any screw holes. The leak only occurred with the high-pressure generated with the pump-loop running, so I was basically unable to locate the exact location of the leak once I removed it from the loop.

Has anyone had anything like this happen? The only theory I really have is some sort of galvanic corrosion that slowly ate away at the internal radiator fin-pipes, but I am not sure if such a thing is even really possible.
 
Soldato
Joined
28 Dec 2017
Posts
8,385
Location
Beds
9 years is a long time to keep copper wet. The pipes are thinner than the end tanks and have lots of solder points so I wouldn't be surprised it that's the first failure point. It is thinner, has been heated up and has been joined to.

I think you got a good life out of that radiator but I'd keep an eye on the other if it's the same age.

Does loop pressure really go up with pump action? I guess near the pump output but the loop as a whole shouldn't develop all that much pressure.
 
Soldato
Joined
18 Oct 2002
Posts
7,813
Location
Scun'orp
I had a pretty much identical experience with a Thermochill PA120.3 of about the same age, only with mine there was actual water to see. With the rad in the case top water was collecting along the join between fan and rad. Just enough to appear moist but not drip or anything, so was probably just a weep more than an outright leak, with the water running to collect where it did. Anyway, I dismantled and tried to find where the leak was. First I submerged it in the bath and blew in one end, expecting to see a stream of bubbles coming from where the leak was...nothing. Set up up in its own loop with a spare pump...again nothing. Not drop in water level and no signs of any moisture. Still haven't been able to explain it, but can't trust it in my main rig again. Am planning to use it in a second PC but have it as an external rad, so if it does leak again in use it won't damage anything.
 
Associate
OP
Joined
2 Oct 2009
Posts
829
Location
Belfast, UK
Interesting. Was your radiator that failed vertical or horizontal? My top-horizontal radiator was the one that failed, and it was also closest to the pump. My remaining radiator is vertically mounted outside the back of the case and seems fine so far, but I will leak-test it for sure before I even think about rebuilding my loop.
 
Soldato
Joined
18 Oct 2002
Posts
7,813
Location
Scun'orp
It was horizontal in the case top. My circumstance was I noticed the drop in water in my tube res was such that the negative pressure actually caused one section of flexible tube to suck in on itself and collapse. I had slackened off the blank plug on the res to let some air in, refilled and reinforced the vulnerable section of tubing with a coil and thought no more of it. Prior to this fix I had seen no signs of water collecting anywhere, or overtly coming out the rad. I put the drop in water down to evaporation as it normally does over time. But after I did this fix it the water level dropped again fairly rapidly over a period of days and at the point I noticed the moisture I described before, as I say, prior to this there was no signs of a leak here. You can play the Twilight Zone/X Files music now if you want, could never explain it afterwards as no leak could be found in the rad once I removed it, and there was no way it could have come from a barb. A blow test should have definitely revealed it if there was a hole/split somewhere in it, since that would have produced way more static pressure than you would ever get from a pump. I also have a PA120.2 mounted vertically in the case, and that is fine up to now, same age.
 
Associate
Joined
13 Jan 2018
Posts
458
Location
London
I had something similar happen to 2 XSPC radiators over the 2 decades plus I've been water cooling. Either slow corrosion or a failed solder joint over time. Fortunately it was slow drips both times and on the second one took months before I bothered locating it even though I noticed the water level dropping. Think I was in denial coupled with laziness. Both times the leaks were in the fin tubes.
 
Soldato
Joined
11 Jun 2003
Posts
10,795
Location
Hampshire
Anything better than the Beyerdynamic DT1990 for the same (or less!) money?

I considered the 1770 but I think they might be a bit fatiguing (from what I’ve read its impressive bass/sub-bass and pretty sparkly highs).
 
Last edited:
Soldato
Joined
11 Jun 2003
Posts
10,795
Location
Hampshire
Slightly wrong forum :D

Ummmm, my bad.

On topic. Did you use inhibitors in the system? I’d never use liquid that didn’t inhibit growth, rust and galvanic corrosion. It’s a recipe for real problems imo.

9 years is quite a while in service, I suppose if you didn’t have inhibitors then galvanic corrosion is possible.

I’ve never heard of a radiator spontaneously going like this with proper components, liquid and maintenance. It’s a weird one.
 
Associate
OP
Joined
2 Oct 2009
Posts
829
Location
Belfast, UK
The first loop I only used distilled water and a silver-kill coil. No growth at all but a a fair bit of black tarnishing, especially on the CPU block. My interim GPU-only loop I added EK Cryofuel inhibitor just to play it safe.

It's pretty interesting I am not the only one who has had this issue, must be a fatigue or corrosion issue with the fin-tubes, maybe exacerbated by powerful pumps? If I do watercool again I don't think I will risk putting a rad inside the case again.
 
Soldato
Joined
6 Jun 2008
Posts
11,616
Location
Finland
A blow test should have definitely revealed it if there was a hole/split somewhere in it, since that would have produced way more static pressure than you would ever get from a pump.
Thermal expansion might explain crack not showing itself always.
If leak was slow enough for water to evaporate instead of forming clearly visible leak signs, leaking crack can't be big.
Different temperatures could cause it opening and closing when we're playing with low pressures.
Also such tiny crack might get clogged by tiniest amount of anykind gunk/drying residue...


The first loop I only used distilled water and a silver-kill coil. No growth at all but a a fair bit of black tarnishing, especially on the CPU block.
That black stuff is likely cupric oxide.
https://en.wikipedia.org/wiki/Copper(II)_oxide
 
Back
Top Bottom