First of all, you have to be totally honest and be aware of what you have at hand. A liquid cooling system, be it AIO or custom, is a “living” set, active by definition and with wear and tear. It is sensitive to temperatures, their changes, dirt, micro evaporation, capillarity and of course to failures in various components, so everything is complicated.
Why are there leaks in liquid cooling?
A very interesting topic to deal with and that in all cases are produced by the same causes, although not the same agents and forms. Here we must logically differentiate between AIO and custom, because the components and assembly, as well as the liquids used are not even remotely similar.
In an AIO system, the manufacturer has designed each part to be accompanied by another in particular and therefore it is a much more reliable system from a technical point of view, but in practice this is not the case. Leaks in AIO systems continue to occur and the factors are always the same:
- Defective component.
- Capillarity and entry of micro dust.
- Bad assembly.
That we get a defective nylon braided tube, an O-ring in poor condition or pinched, that the inner coating of blocks and tubes is not adequate or that they have a mistake on the assembly line is within what is plausible and is usually the most common failures. Obviously there are others that are a problem, such as polarized 4-pole pumps that end up not working and raise the internal pressure and literally make a component burst, or for example tubes that are pinched due to bad positions and that generate the same effect described.
In custom liquid cooling the factors are much higher because logically we are talking about loose parts and a custom assembly that results in a huge number of possible problems:
- Bad fit of the fittings.
- Oversized pump for the system.
- Excessive pressure on components with Plexi or acrylic.
- Off-center pump shafts.
- Pinched O-rings.
- Bad assembly or little force when tightening fittings, elbows or various extenders.
- Badly designed and executed system.
There we exclude for supposed poorly thought-out component designs (there are many examples in the industry and unfortunately very recent), which in themselves make the leak irreparable and have to process RMA.
How to avoid leaks in liquid cooling?
First of all we are going to give some keys that we have to be clear before we get down to work. In an AIO we cannot do anything because the system is precisely that All In One, that is, it comes pre-assembled as standard ready to work. The only thing we can do foresight is to test the AIO out of the box and before mounting it with a tapped source and an external controller type Aquaero or similar, a rehobus with PWM and 3 pins would also be optimal.
What we will have to do is make it work in its entire RPM range and if possible heat the cold plate of the block progressively to a temperature of 50 degrees (a heat laser gun or thermal probe is needed for thermal control). Thus we will simulate the operation of the GPU or CPU and we will give the water time to heat up and thereby gain pressure in the system to see if it has a leak.
As it is an AIO if it had it then it would be necessary to speak with the store or the manufacturer in question. What about custom liquid cooling? Well, things get quite complicated here, so we will try to go through steps, but first we have to buy one more component: the leak tester for liquid cooling. This will be vital to everything we have to do and it will be the only useful tool that will determine the viability of the entire system.
The first step and check is to do it component by component except in the tubes, whether they are soft or rigid for obvious reasons. It is tremendously complicated for a tube to be pinched, cracked or simply with micro cracks, a visual inspection with a good eye will be enough, but if we are suspicious then we will have to use a connector at each end where we will place the tester in one of them and in the other a plug.
The method for everything is the same and it is really slow, very slow if we want to do it right. Saving the tubes as we say, what we will have to do is make a check of between 1 hour and 8 hours for each component of the system, be it blocks, radiators, tanks or tanks with pumps, as well as rotary extension fittings, duplicators, T or valves.
The pressure must be between 0.5 bar and 0.75 bar, where each of them must maintain the pressure without moving an iota of the mark where we stop introducing air. As we say, time is essential and we recommend 0.75 bar of pressure because in cold we can be at approximately 0.4.
Once we know that all the components are 100% watertight, the first step arrives, which is none other than mounting the system in sections. For each section of pipe made, the system must be closed at one end and pressure and leaks checked at the other. This is extensible as the circuit grows and is crucial to detect bad assemblies or errors in them without having a nice liquid festival for the PC.
It is the one that is done with the system mounted and logically without liquid. We suppose that we have left a drain valve in the system, so that is where we will introduce the leak detector and after reaching 0.75 bar of pressure it is necessary that we leave the system on standby for 24 hours.
If after this interval of time the needle is still in place, perfect, then it is when we can refill without fear. After this and with the consequent cloths, napkins or liquid collection system that we prefer, the PC has to be another 24 hours circulating for its correct purging and of course it is necessary to be aware of each connection. You do not need to be under a stress test, at least the first 8 hours and as long as everything goes well.
After that time we will be able to demand it to see how much heat the system is capable of evacuating and how it behaves under pressure and thus we will avoid a leak in our liquid cooling.