Sure enough, this latest leak was from the pump housing. I found it and sealed it with silicone sealant.
I then retested, and there is
still a very small leak from the pump housing, but it only occurs during Thermal Calibration, which really flogs the pump hard. During routine operation, there is no leakage. Remember, I use bright red coolant and I put a clean piece of white paper towel under the LCS when the system is running, so that if anything leaks I will be able to spot it right away. I can therefore say with confidence that it runs leak-free normally, but leaks a small amount during Thermal Calibration.
I have come up with a novel idea for getting more coolant into the loop without opening the loop back up. It is an odd enough idea that I will hold it for now and report it and it's results together, in the next few days - stay tuned!
I tried out said "novel idea" yesterday and it works well. What is it? It is a mechanism for getting rid of those last few air bubble/pockets in the loop, or viewed alternately, adding that last little bit of coolant to the loop.
.... and the idea is ... [drum roll please!] ... using a hypodermic needle/syringe to poke two pin prick holes in the hosing. I then add more coolant via the syringe inserted into the second hole while the air the coolant is displacing is is expelled through the first hole.
These syringes are available on Amazon as a four pack for less than $US 5.00.
In the photo above, I am using the needle to poke the first hole (the air exit). I then repeated with the second hole (coolant injection) and then used the loaded syringe to pump in the coolant it holds. I then sealed up both holes with silicone sealant, waited one hour and I was ready to go (it sets it in about 30 minutes, so I give it twice as long, just to be sure).
It worked well, but it made
zero difference to CPU temperatures and fan speeds, which surprises me. The machine is thermally stable, holding steady in the high 30s on CPU A and the mid 40s on CPU B, but it requires CPU intake and exit fan speeds in the 2900 - 3200 RPM range to hold it at those temperatures... still way too loud for routine use.
The machine is fully stable and runs well in this configuration; it is simply too loud, which has been the problem all along.
I cannot think of anything else I can do to make the loop achieve any more cooling, so I guess that this is my personal "state of the art" for the moment on LCS cooling. To complete this effort, next up is finding
really quiet 92mm case fans that can run at full speed in a fairly low noise way. I will then replace the existing Apple CPU intake/exit fans with these new ones, and I *believe* the machine will then be thermally stable
and reasonably quiet, which has been the end goal all along.
After two months of solid work on this topic, I have successfully lowered CPU temperatures into a good range, but I have failed to tame the fan revs; they are still far too high and I am fresh out of ideas for bringing them down further. Unless someone reading this has a brilliant idea for achieving further cooling from my new loop, this is the end of the road for liquid cooling for me - it is just too finicky and too difficult to optimize.
This has been a long thread, and if you are thinking of servicing your LCS, don't bother reading the whole thing. Instead, read my A-Z guide to LCS servicing, which you will find at my other related thread here at the MacRumors "PowerPC Mac" section:
Comprehensive Guide to Servicing G5 Quad LCS
This guide is rich in photos and tips for making the work easier. While long itself (55 pages!) it will provide an amply illustrated guide to LCS servicing, from "my Mac is too loud" to "ahh, peace and quiet".
I am now moving on to
air cooling for my other Quad, a whole new journey, and undoubtedly, a whole new set of large engineering challenges. Let the fun begin!
