Absolutely! Nothing substitutes for "safe computing". It remains a dangerous world out there. The platform you are on helps, but always be aware and be alert!
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Here We Go Again!
- Thread starter mac57mac57
- Start date
- Sort by reaction score
My new coolant arrived yesterday and I loaded the newly built cooling loop with it today, and installed the overhauled LCS back into its Quad and tested.
There is good news and bad news.
The good news is that for the first time ever (since I have been working with LCSs) the new loop was perfectly tight... no leaks at all. I have learned a few things! 😉
The bad news is that CPU A overheats and the machine checkstops before it can boot anything. I can get Open Firmware to run, but that is all. Even the Boot Picker freezes.
I *think* I know what the issue is - I had real trouble installing the CPU A card onto the LCS, and I suspect that I messed up its cooling performance in the process.
I will remove the CPU A card from the LCS and reinstall it, but not before, out of curiosity, I try running the machine with only the CPU B card physically present in the box. I have demonstrated that Quads will run with only the CPU A card physically present; I am curious if the same is true for the CPU B card.
It will cost me several hours to run this test, but it is worth it, if only for intellectual curiosity! I will report the results back here so that others have this answer too.
More results when I have been able to run these tests.
There is good news and bad news.
The good news is that for the first time ever (since I have been working with LCSs) the new loop was perfectly tight... no leaks at all. I have learned a few things! 😉
The bad news is that CPU A overheats and the machine checkstops before it can boot anything. I can get Open Firmware to run, but that is all. Even the Boot Picker freezes.
I *think* I know what the issue is - I had real trouble installing the CPU A card onto the LCS, and I suspect that I messed up its cooling performance in the process.
I will remove the CPU A card from the LCS and reinstall it, but not before, out of curiosity, I try running the machine with only the CPU B card physically present in the box. I have demonstrated that Quads will run with only the CPU A card physically present; I am curious if the same is true for the CPU B card.
It will cost me several hours to run this test, but it is worth it, if only for intellectual curiosity! I will report the results back here so that others have this answer too.
More results when I have been able to run these tests.
Quick update: Quads cannot run with just the CPU B card physically present in the box.
The Quad will give a startup chime in this configuration, but only after an uncomfortably long interval, and it will run both Open Firmware and the Boot Picker, but it will not boot. Oh well, I had to try.
I've disassembled the Quad, extracted the LCS, and fixed the physical issue that was causing CPU A to overheat (not a tight enough seal between the CPU face and its cooling block). Hopefully, once rebuilt, all will be well.
The Quad will give a startup chime in this configuration, but only after an uncomfortably long interval, and it will run both Open Firmware and the Boot Picker, but it will not boot. Oh well, I had to try.
I've disassembled the Quad, extracted the LCS, and fixed the physical issue that was causing CPU A to overheat (not a tight enough seal between the CPU face and its cooling block). Hopefully, once rebuilt, all will be well.
All done with this latest LCS overhaul, and the results are the best I have achieved to date doing this sort of work. I definitely have learned a few things along the way, and applying all of them has delivered this result. I am going issue another update to my LCS Overhaul document, capturing a few additional points that turned out to be key.
After resolving the CPU A problem (poor fit to the cooling pad face) and reassembling, the cooling system, amazingly, seems to be running to factory spec:
Note the 970 RPMs for the Intake fans. That is Apple's "easter egg" type node to the PowerPC 970MP that energizes the Quad, and you will only see that if the cores are cooled as Apple originally intended; else, the fans ramp up. This is why I say "factory spec".
RPMs are low and CPU temperatures are low as well - what more could you want from an LCS overhaul?
...and yes, all four cores are up and running!
After resolving the CPU A problem (poor fit to the cooling pad face) and reassembling, the cooling system, amazingly, seems to be running to factory spec:
Note the 970 RPMs for the Intake fans. That is Apple's "easter egg" type node to the PowerPC 970MP that energizes the Quad, and you will only see that if the cores are cooled as Apple originally intended; else, the fans ramp up. This is why I say "factory spec".
RPMs are low and CPU temperatures are low as well - what more could you want from an LCS overhaul?
...and yes, all four cores are up and running!
Attachments
As mentioned above, I have published an updated copy of the Power Mac G5 Quad LCS overhaul A-Z guide. You can get it at:
https://inverary.net/LCS/Power Mac G5 Quad LCS Restoration, A-Z Guide, v1.3.pdf
https://inverary.net/LCS/Power Mac G5 Quad LCS Restoration, A-Z Guide, v1.3.pdf
A quick follow-up on this. A 2005 Quad is indeed an older box, and things do routinely fail on older boxes.
I no sooner got started configuring my newly cooled Quad than the LCS pump failed! You can build the best cooling loop in the world, and fill it perfectly with the best coolant in the world, but if the pump doesn't circulate the coolant around the loop, it is all for naught.
I managed to overcome this latest obstacle and the machine is once more working fine, but I am posting this to encourage anyone who is thinking about overhauling their Quad's LCS to take the Air Cooling approach instead. Air Cooling is inherently simpler, and once done, it doesn't have to be revisited, unlike an LCS, where more servicing will absolutely be needed at a future date.
Air Cooling is (relatively) simple to implement and is both cool and quiet. It is also more or less permanent. Do it once and you are done. You may need to repaste a CPU A from time to time, but you would need to do that no matter what your cooling solution was.
Air Cooling is the way to go!
I no sooner got started configuring my newly cooled Quad than the LCS pump failed! You can build the best cooling loop in the world, and fill it perfectly with the best coolant in the world, but if the pump doesn't circulate the coolant around the loop, it is all for naught.
I managed to overcome this latest obstacle and the machine is once more working fine, but I am posting this to encourage anyone who is thinking about overhauling their Quad's LCS to take the Air Cooling approach instead. Air Cooling is inherently simpler, and once done, it doesn't have to be revisited, unlike an LCS, where more servicing will absolutely be needed at a future date.
Air Cooling is (relatively) simple to implement and is both cool and quiet. It is also more or less permanent. Do it once and you are done. You may need to repaste a CPU A from time to time, but you would need to do that no matter what your cooling solution was.
Air Cooling is the way to go!
I would if I lived in the magical place where G5s appeared at reasonable prices.
I already have a Quad though, perhaps in time I'll find a Dual 2.7 to do that with.
I already have a Quad though, perhaps in time I'll find a Dual 2.7 to do that with.
A counterpoint: I’ve encountered a good number of G5 Quads, currently own 3, and have fully refurbished the cooling system on one. I’ve never seen a pump fail. I also don’t recall having heard about pump failures on forums. Meanwhile, you’ve experienced two pump failures in the process of attempting to repair your LCS. I am concerned that the issue is not with the aging pumps but with the process you are using to repair the LCS, and that anyone who follows that process may render themselves prone to the same sort of failures.
Are you using the same coolant as in your last build? I noted in that thread that it was leaving a chalky residue in the pump. Whether or not that was the true cause of the failure, I would not want that kind of material inside my pump or radiator.
I also note that you instruct in your guide to run the pump directly off a 12v connection. Again, this seems like a good way to break things (you attribute one pump failure to this already). It’s not necessary if you get an external pump, either.
Is it a pain to repair these liquid cooling systems? Yes, no doubt. It’s not rocket science though. If one is patient and methodical it’s a pretty mechanical process- remove the old hoses, clean out the radiator, rebuild the loop with new hoses, leaving it open to allow an external pump to charge the system with coolant and flush out air. Then seal it up, letting in as little air as possible. I recommend Peak Global Lifetime for the coolant (I believe I recommended this in the last thread too) and stand by that recommendation, it had no issues with dye sedimentation that I observed and also purportedly has a nice concoction of corrosion inhibitors.
Good points @MisterKeeks, and largely not things I have not asked myself.
Agreed, I was using a coolant that turns out to not have been all that well regarded, but I no longer use that. These days I use one that I found recommended on these very forums:
I have found myself wondering about the viscosity of this particular coolant, wondering if it is too "thick" for the pump and thus exerting an undo load on it.
In this case however, in the end it turned out that the pump "failure" was simply a loose connection. Once corrected, all was well and the cooling result is excellent, especially those coveted 970 RPM fan readings!
I agree that overhauling an LCS is a methodical step-by-step thing. However, it seems that even the slightest deviation from the optimum processing of each one of those steps seems to leave one with a problem whose only solution is to tear it all down again and redo.
As for the use of the pump with a direct 12v connection, the guide explicitly directs that you do not do this for any length of time and goes into the reasons for that (duty cycle of the pump). Used as directed, brief bursts of pump action can move things around in the loop, helping get an optimal fill. BTW, I got the idea to do this (and the pump connector pinout!) from these very forums!
In the end, I stand by my original comments above. Converting to air cooling is faster and easier, and the result is essentially permanent. You don't have to worry about your (passive) air coolers failing (!) ... and if one of the cooling fans fails (which I have never seen happen), the repair is easy and inexpensive. You do not have to tear down the whole solution to fix it. Your air cooled solution will never develop a leak and no minerals, crystals etc. will ever build up inside the solution over time.
Overhauling an LCS is a different matter. It is much more involved, but it has the upside of when it is done, it returns the Quad to the cooling mechanism that Apple intended. This makes it of historical interest, and of interest as well simply because it IS the more challenging of the two approaches. It is good to master both approaches.
Agreed, I was using a coolant that turns out to not have been all that well regarded, but I no longer use that. These days I use one that I found recommended on these very forums:
I have found myself wondering about the viscosity of this particular coolant, wondering if it is too "thick" for the pump and thus exerting an undo load on it.
In this case however, in the end it turned out that the pump "failure" was simply a loose connection. Once corrected, all was well and the cooling result is excellent, especially those coveted 970 RPM fan readings!
I agree that overhauling an LCS is a methodical step-by-step thing. However, it seems that even the slightest deviation from the optimum processing of each one of those steps seems to leave one with a problem whose only solution is to tear it all down again and redo.
As for the use of the pump with a direct 12v connection, the guide explicitly directs that you do not do this for any length of time and goes into the reasons for that (duty cycle of the pump). Used as directed, brief bursts of pump action can move things around in the loop, helping get an optimal fill. BTW, I got the idea to do this (and the pump connector pinout!) from these very forums!
In the end, I stand by my original comments above. Converting to air cooling is faster and easier, and the result is essentially permanent. You don't have to worry about your (passive) air coolers failing (!) ... and if one of the cooling fans fails (which I have never seen happen), the repair is easy and inexpensive. You do not have to tear down the whole solution to fix it. Your air cooled solution will never develop a leak and no minerals, crystals etc. will ever build up inside the solution over time.
Overhauling an LCS is a different matter. It is much more involved, but it has the upside of when it is done, it returns the Quad to the cooling mechanism that Apple intended. This makes it of historical interest, and of interest as well simply because it IS the more challenging of the two approaches. It is good to master both approaches.
Last edited: