Do you have a picture of the loop? My two hypothesis are a blockage in the radiator and kinked hoses.
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.
The Quad Chronicles - Restoring a HOT G5 Quad
- Thread starter mac57mac57
- Start date
- Sort by reaction score
Funny, @MisterKeeks, I was thinking along similar lines. I have decided to extract the LCS one more time and check it over thoroughly for any obstructions, etc. I just don't know how to quit when I am ahead!
No new photos needed - I haven't changed the loop itself since my "FrankenCooler Rev 4" posts earlier in this thread, so it is the same loop as earlier. Here is that earlier photo:
I flushed the radiator thoroughly when I started this effort, so I am confident in it. Redoing the flush would require dismantling and draining the loop... THAT is not "on the cards" at this point.
No new photos needed - I haven't changed the loop itself since my "FrankenCooler Rev 4" posts earlier in this thread, so it is the same loop as earlier. Here is that earlier photo:
I flushed the radiator thoroughly when I started this effort, so I am confident in it. Redoing the flush would require dismantling and draining the loop... THAT is not "on the cards" at this point.
The CPU B outflow line into the T fitting looks very kinked. I am not surprised you see higher temps on that CPU.
The pump outflow line into the T fitting also looks marginal.
Do you have the original T fitting that was in the loop? I might compare the inner diameter of that with what you installed. These fittings reduce flow. I also wonder if placing the additional fitting right on the pump outflow line is problematic, relative to other potential locations. I don’t know.
What’s going on with the service port?
Re: radiator, we saw earlier that the coolant you are using left a chalky residue in the pump. I could imagine that kind of behavior causing problems in the narrow passageways of the radiator.
Many of your radiator fins appear bent. That will require higher fan speeds because less air will flow over them. I’d consider using a toothpick or other soft tool to try to straighten them, though I don’t really expect this to make much of an impact.
Status update: I extracted the LCS, examined it carefully, tightened up the last known leak location and added a little more cooling fluid to the loop. I then reassembled and tested.
Murphy just couldn't resist - what a mess! The silicon sealant that I used to seal up the holes from the injection of the extra fluid did not seal properly. When I turned the machine on, pressurized by the pump, the loop sprayed coolant all over the place before I could kill the power. I lost all the coolant I had added and then some. I just can't win for losing on this particular effort.
I have replaced the lost coolant and resealed the leak locations. Once the sealant is cured, I will try again. I will keep at it until the machine is finally leak free once more... or something else goes wrong! I am mighty sick of G5 Quads these days!
THEN it will be time to do the fan replacement with uber-quiet ones, which arrive from Amazon today.
Murphy just couldn't resist - what a mess! The silicon sealant that I used to seal up the holes from the injection of the extra fluid did not seal properly. When I turned the machine on, pressurized by the pump, the loop sprayed coolant all over the place before I could kill the power. I lost all the coolant I had added and then some. I just can't win for losing on this particular effort.
I have replaced the lost coolant and resealed the leak locations. Once the sealant is cured, I will try again. I will keep at it until the machine is finally leak free once more... or something else goes wrong! I am mighty sick of G5 Quads these days!
THEN it will be time to do the fan replacement with uber-quiet ones, which arrive from Amazon today.
Today's update: After reassembly and testing, the Quad seems to be leak-free and CPU temperatures are well controlled, if a bit higher than I would like on CPU B. Despite the addition of more coolant to the loop, overall CPU temperatures are about 2 C higher than before. However, like before the fans remain maxed out at 3200 RPM to maintain those temperatures, and the pump is running at its full speed of 3600 RPM. An XRG screen shot tells the tale:
I have to conclude that I am probably never going to convince Mac OS X's thermal/fan control software to turn the RPMs down on those fans, so instead it is time to make them audibly tolerable at these speeds.
My new "uber quiet" fans were delivered yesterday, so I will start working on substituting them into the CPU intake and exit fan blocks, hopefully creating uber quiet versions of them. I have a Power Mac G5 Dual (the one I posted about some back, and then purchased when the seller made me a compelling offer) that I will use as the test bed for this work. No more jiggling of my now operational, leak free Quad!
I will start with the CPU intake fan block, replacing its fans with the new quiet ones, and running the control through this bad boy (which I highlighted some time ago, and which was delivered a while back):
This leaves Mac OS X out of the loop entirely on fan speed control, which will now be under my direct control via the dials on the above. The new fans are interesting... they are infinitely speed-adjustable, from 0 RPMs all the way up to 4300 RPMs, faster than the Apple fans can go. I suspect that at full tilt (4300 RPMs) these fans are probably as noisy, if not noisier, than the Apple ones are at 3200 RPM, so I will not be running them that fast except perhaps when I am doing something really taxing with the CPUs - video editing for example.
Once done, I should have both quiet fans and speed-adjustable fans. Lets see if that is the outcome, or whether Murphy strikes again in an unexpected way! Perhaps once I deal with the fans, I will find that the pump itself makes an intolerable amount of noise? Something else? This entire adventure has been plagued by unexpected results, so anything is possible yet!
I have to conclude that I am probably never going to convince Mac OS X's thermal/fan control software to turn the RPMs down on those fans, so instead it is time to make them audibly tolerable at these speeds.
My new "uber quiet" fans were delivered yesterday, so I will start working on substituting them into the CPU intake and exit fan blocks, hopefully creating uber quiet versions of them. I have a Power Mac G5 Dual (the one I posted about some back, and then purchased when the seller made me a compelling offer) that I will use as the test bed for this work. No more jiggling of my now operational, leak free Quad!
I will start with the CPU intake fan block, replacing its fans with the new quiet ones, and running the control through this bad boy (which I highlighted some time ago, and which was delivered a while back):
This leaves Mac OS X out of the loop entirely on fan speed control, which will now be under my direct control via the dials on the above. The new fans are interesting... they are infinitely speed-adjustable, from 0 RPMs all the way up to 4300 RPMs, faster than the Apple fans can go. I suspect that at full tilt (4300 RPMs) these fans are probably as noisy, if not noisier, than the Apple ones are at 3200 RPM, so I will not be running them that fast except perhaps when I am doing something really taxing with the CPUs - video editing for example.
Once done, I should have both quiet fans and speed-adjustable fans. Lets see if that is the outcome, or whether Murphy strikes again in an unexpected way! Perhaps once I deal with the fans, I will find that the pump itself makes an intolerable amount of noise? Something else? This entire adventure has been plagued by unexpected results, so anything is possible yet!
Finally, positive results to report!
Both the intake and exhaust Apple CPU fans have been removed and replaced with Arctic P9 Max Black fans. The photo below shows the front fan assembly with the stock Apple fans replaced with the P9s.
These fans were of great interest to me because they go all the way up to 4300 RPM, well in excess of the top speed of the Apple fans, and they push a whopping 58.25 Cubic Feet per Minute (CFM) of air.
These fans were connected to the below variable speed fan controller, which I installed into the final free slot in the PCI expansion area of the Quad. Despite its enormous utility, this is not a robust unit: build quality is poor and I get the sense that if I plugged and unplugged the MOLEX power connector more than a few times, it would just break off! Nonetheless, despite its obvious weaknesses, it does a GREAT job!
Speaking of MOLEX, there was one enabling step that had to be taken. There are no visible MOLEX connectors inside the Quad, much less any free ones, and yet this fan controller needs one for power. To accommodate this need for MOLEX, I used a MOLEX splitter on the all but invisible MOLEX that powers the DVD reader/writer, and then pulled the other end of the splitter through a gap at the bottom of the PCI Divider plate, yielding a free MOLEX inside the Quad. From this, I was able to power the fan controller.
With all of this done, the net result was excellent. Here are the CPU temps achieved with the fans between 1/4 and 1/3rd speed:
The P9 fans are not perfectly quiet, but they are a HUGE improvement over the "back side of a jet engine" noise level the stock Apple fans make. I have ordered a quieter fan for testing purposes, the Noctua NF-A9 fan (I have one already, so that makes two, enough to replace either the front or back fan set) and I will try these. They are only 2000 RPM max, but that is clearly more than enough, and they push a similar volume of air relative to the P9 units despite the lower RPMs. They are rated at a lower acoustic noise level as well (22.8 dBA), so they look promising.
Operationally, both the intake and exhaust P9 CPU fans are wired only to the above variable speed fan controller, with no connection whatsoever to the motherboard fan control headers. The speeds of the intake and exhaust fans can be increased/decreased just by turning the associated control nob, which of course is external to the case.
Apple is blissfully unaware of all of this. iStatMenus reports that Mac OS X is spinning its phantom intake and exhaust CPU fans at 3200 RPM, but since they are not physically there, the usual roar they make is blissfully absent. The one downside of this is that ultimately, it spins the pump up to its full 3600 RPM, which is unfortunate; that is a source of noise that I will not be able to eliminate.
In the end then, I have my FrankenCooler Rev 4 providing the cooling loop function, and the P9 fans moving the heated air picked up from the cooling loop via the radiator out through the back of the Quad. The Quad now runs cooler and quieter than it has in a very long time!
So, FINALLY, this story comes to a happy conclusion... but believe it or not, this is not the end of the road. I will try the Noctua fans, to lower the noise level even more, and then, heaven help me, but I *may* have just have one more run at rebuilding the cooling loop - there are minor improvements I can make to the current one, particularly in the area of removing minor narrowing in some of the tubes as they turn corners.
I will post more about these results as I get to them.
BTW, this post is coming to you from the very Quad that this thread is all about! It is working well, and I LOVE the low noise level, soon to be even lower! I will never achieve a whisper-quiet Quad - even right off the factory floor they were not perfectly quiet - but I can get darn close!
Both the intake and exhaust Apple CPU fans have been removed and replaced with Arctic P9 Max Black fans. The photo below shows the front fan assembly with the stock Apple fans replaced with the P9s.
These fans were of great interest to me because they go all the way up to 4300 RPM, well in excess of the top speed of the Apple fans, and they push a whopping 58.25 Cubic Feet per Minute (CFM) of air.
These fans were connected to the below variable speed fan controller, which I installed into the final free slot in the PCI expansion area of the Quad. Despite its enormous utility, this is not a robust unit: build quality is poor and I get the sense that if I plugged and unplugged the MOLEX power connector more than a few times, it would just break off! Nonetheless, despite its obvious weaknesses, it does a GREAT job!
Speaking of MOLEX, there was one enabling step that had to be taken. There are no visible MOLEX connectors inside the Quad, much less any free ones, and yet this fan controller needs one for power. To accommodate this need for MOLEX, I used a MOLEX splitter on the all but invisible MOLEX that powers the DVD reader/writer, and then pulled the other end of the splitter through a gap at the bottom of the PCI Divider plate, yielding a free MOLEX inside the Quad. From this, I was able to power the fan controller.
With all of this done, the net result was excellent. Here are the CPU temps achieved with the fans between 1/4 and 1/3rd speed:
The P9 fans are not perfectly quiet, but they are a HUGE improvement over the "back side of a jet engine" noise level the stock Apple fans make. I have ordered a quieter fan for testing purposes, the Noctua NF-A9 fan (I have one already, so that makes two, enough to replace either the front or back fan set) and I will try these. They are only 2000 RPM max, but that is clearly more than enough, and they push a similar volume of air relative to the P9 units despite the lower RPMs. They are rated at a lower acoustic noise level as well (22.8 dBA), so they look promising.
Operationally, both the intake and exhaust P9 CPU fans are wired only to the above variable speed fan controller, with no connection whatsoever to the motherboard fan control headers. The speeds of the intake and exhaust fans can be increased/decreased just by turning the associated control nob, which of course is external to the case.
Apple is blissfully unaware of all of this. iStatMenus reports that Mac OS X is spinning its phantom intake and exhaust CPU fans at 3200 RPM, but since they are not physically there, the usual roar they make is blissfully absent. The one downside of this is that ultimately, it spins the pump up to its full 3600 RPM, which is unfortunate; that is a source of noise that I will not be able to eliminate.
In the end then, I have my FrankenCooler Rev 4 providing the cooling loop function, and the P9 fans moving the heated air picked up from the cooling loop via the radiator out through the back of the Quad. The Quad now runs cooler and quieter than it has in a very long time!
So, FINALLY, this story comes to a happy conclusion... but believe it or not, this is not the end of the road. I will try the Noctua fans, to lower the noise level even more, and then, heaven help me, but I *may* have just have one more run at rebuilding the cooling loop - there are minor improvements I can make to the current one, particularly in the area of removing minor narrowing in some of the tubes as they turn corners.
I will post more about these results as I get to them.
BTW, this post is coming to you from the very Quad that this thread is all about! It is working well, and I LOVE the low noise level, soon to be even lower! I will never achieve a whisper-quiet Quad - even right off the factory floor they were not perfectly quiet - but I can get darn close!
Attachments
Last edited: