Working on the G5 quad liquid cooling system

I have a G5 quad with the one pump Delphi LCS ("version 1"). I'm guessing the LCS is clogged because when the machine sits idle (even in reduced CPU setting, which cuts the GHz in half), the fans and the pump slowly, over several hours, go to full speed. Temperatures of CPU B, particularly the second core, shoot to near 100C when loading the CPUs. ASD 2.6.3 usually passes, but sometimes ends in checkstop and overtemp. A local mac service center, which I had the machine checked at (in hopes of getting free repairs because of a leak), said it's fine.


As a simple first measure, I tried replacing the thermal paste on the CPUs, with little effect. The LCS didn't seem to be leaking or corroded anywhere, but when handling it, it sometimes made a bubbling noise, so, there's probably air in there. I thought I'd take the LCS apart next, but before that I have some questions which I hope someone can answer.


Which one, the upper or lower, is CPU A and which is B? I've read the machine runs on one CPU just fine, but which one?


If I switch the CPUs the other way around and later switch back, does something in the nvram reset so that thermal calibration needs to be run? I'd rather not run it as it seems to have mixed success.


What material/size are the CPU block O-rings on this system?


If I use a vacuum pump on the service valve (I gather it's of the "r134a high" type(?)), what would be a safe level of vacuum? Has anyone tried using the manual brake kits, or car cooling system refill kits that use compressed air and venturi valves, on the LCS?


I suppose I'm just going to try attaching a hose as a reservoir to the service valve and fill through that first, but if that doesn't bring satisfactory results, I'll level-up to some sort of vacuum method.

PowerMac

Posted on Sep 17, 2011 11:59 PM

Reply
252 replies

Feb 11, 2012 2:40 AM in response to Ramón G Castañeda

Ramón, about having someone do these sort of overhauls: I was actually playing with the thought that I could start doing these on the side, but there's very little demand here where I live. A few years ago, I remember seeing an Australian Apple service tech advertising that they can repair the LCS. Generally, anyone handy enough to do car maintenance can probably do the overhaul, but of course, experience of having done it before, and the proper tools, will help. One problem I see with DIY repairs is that the "pool" of spare parts gets smaller and smaller as parts get destroyed by the inevitable oopses of the DIYers.


As to my approach, I'm too nitpicky and, apparently, had too much time on my hands. Unlike current electronics, water cooling isn't really black magic; if it was, cars would be blowing up all around, central heating wouldn't work, and so on. Careful material selection and as-clean-as-possible process should give a larger maintenance interval, but that's about it. Off-the-bat duct tape methods can give results that work just fine. For example, tap water is always said to be a big no-no in water cooling because of its impurities, but one guy ran a WC loop with tap water for almost a year with little ill effects.

Feb 11, 2012 4:33 PM in response to Heikki Lindholm

Heikki Lindholm wrote:


…Generally, anyone handy enough to do car maintenance…


That rules me out. 😝

Heikki Lindholm wrote:


…A few years ago, I remember seeing an Australian Apple service tech advertising that they can repair the LCS…


Autralia is a wee bit too far away from my location in Northern California. Hope someone has a lead closer to me. 😀


Thanks anyway. Your posts have been tremendously helpful. 🙂

Feb 13, 2012 3:37 PM in response to Chris323i

The install has worked very well and I am actually posting this from my G5. Installation of the LCS was not an issue at all.

I have been running the computer since last night and no leaks or stange issues/noises. Fans have been running at a slight to moderate pace during use. What a machine this is! Very fast!

I give this project about an 7 out of 10 in difficulty but mainly because of the precision work that must be done when reconstructing items. Overall, I spent about $100 in parts. I don't believe anyone will be able to pry this machine away from me in the future.


Here is a shot of iStat I have just taken after running all day and night with a multitude of programs running.

User uploaded file

Feb 13, 2012 4:10 PM in response to Chris323i

Great work, Chris, congratulations! …and thank you for sharing your images and your experience.


I do note that both of your pumps run roughly at 3 x the speed of the single pump in my G5 Quad, made in late June of 2006, which runs pretty steadily at 1250 rpm most of the time.


I'm extremely happy that my G5 Quad is running so well—knock on wood!—but at least I know there's hope for when the day comes that the LCS has to be refurbished. I "just" have to find a skilled tech willing to undertake the job. I agree, this is one heck of a machine!

Feb 13, 2012 10:22 PM in response to Chris323i

Chris, the fan speeds do seem high, all of them. The min is 970 for intake, 1000 for exhaust, and 1250 for the pumps. In "normal" ambient (less than 25 C), the fans should be running at min speed when the machine is idle. Fully stressed (Bresink's SystemLoad does that well enough, but careful with it; i.e. enable one CPU at a time or so), CPU temps should definitely stay below 70 C and probably even below 60 C if the LCS works well and ambient isn't hot. However, I'd wait for the situation to "settle" for a couple of days.

Feb 15, 2012 3:12 PM in response to Heikki Lindholm

Yes, you're right Heikki. The fan and pump readings were a bit high because I tried to run quite a few programs to flex its muscle. It didn't help that the room was a bit warm, around 80*F.

At idle with iTunes and a few other programs running in a 77*F room, I am having my fan speeds around 1400RPM with the pumps just under or around 1800RPM. The G5 certainly keeps my small room warm. Once summer arrives, I may need to sleep it when it's not in use.

Feb 29, 2012 5:44 PM in response to Heikki Lindholm

Glad, i found this thread as the most recent one for the topic and hope it is still alive.


I have a Quad eqipped with the one pump Delphi/Laing LCS and also had to check it for temperature issues. First suspect was the pump. But this has proven wrong. After some cleaning it runs powerful and fine...


...but it has some hard work to do! My question is: how easily should the water/coolant pass the LCS. Or - in other words - is some really noticeable resistance in the flow normal behavior or caused by too much sediments or corrosion in the sytem?


Running on it's own and sucking water directly from a dish, the pum spills a powerful, tube diameter stream around. Doing the same through the complete LCS, there's just a very small flow dripping of the outlet.


There wasn't too much material in the coolant i drained from the system. And also the pump didn't look that bad for nearly 7 years of unserviced use. But, what there was, was very hard and sticky to the parts. Sure, i could check the waterblocks and tubes one by one. But as the whole system seems to be 100% not leaking, i would like to take apart as little as possible to keep it like that.


If this is not nomal, another question could be, if something like this (for cars) could be used to clean the inside without taking the LCS appart.


Thanks in advance for any hint!

Feb 29, 2012 10:28 PM in response to Fipps

"Small flow dripping" doesn't sound right. There are three places that gather buildup in there: the two inlets of the CPU blocks (they have filters that protect the cpu blocks) and the radiator (probably the bottom of it more so). If you don't want to rebuild the pipework, use hose removal pliers to detach (carefully) the three hoses from the radiator and clean the rad thoroughly. Then reverse flush the detached pipework one flow path at a time to clean the cpu block inlets.

Mar 1, 2012 6:01 AM in response to Fipps

I see no harm in using a rad cleaner, but I'd only use it for the rad and rinse it out when done. Maybe I'm paranoid, but these things are difficult enough to get spare parts for.


I detached the hoses from the rad by first gripping them with normal pliers, with a cloth between the pliers and the hose, and rotating the hose around the barb a a few millimeters back and forth in order to make sure it's loose. Then I pried a sewing ping or some implement like that - a blunt steel pin - a few mm under the hose so that I could lift the edge of the hose enough to insert the hose removal pliers between the hose and the rad. That was the most difficult part - with another pair of hands it would have been easy. Then patient "pumping" with the pliers and the hoses detached nicely without damage. The hose removal pliers I used were similar to what google shows from amazon with "bikemaster hose removal pliers."

Mar 1, 2012 5:27 PM in response to Heikki Lindholm

Thanks for the detailed instructions, which helped me alot!


I'm very happy that i finally managed to remove those hoses without destroying anything. It's been a very hard task! Sadly these hose removal pliers are exactly not the kind of thing, the next door hardware store has in stock. So i had to build my own (maderine not included). Doesn't look that nice, but it worked.


The findings after this: The radiator itself seems to be free and relatively clean. Flushing it, even shaking and draining did not show a lot of buildup or other stuff. Water runs through it free and easy...


...but both of the CPU-blocks have very little troughput! In both directions. This could be the point to open them, to clean them or to see, what's inside. But i'm afraid to, cause they are not leaking now and i don't know if they will be like this again afterwards. Also the screws seem to be very tight and additionally secured with some stuff like locktite. I carefully tried to losen one of them with a usual phillips screwdriver but wasn't able to. I didn't want to use extensive force on that.


Now i wonder, how much liquid should normally pass there. The blocks look very flat, much unlike the ones of the early Dual 2.5s. In no way, as far as i see, there could be room for any copper fins in there. Also i wonder why they put a "filter" in there, as the LCS is a closed circuit with no way for anyting to get in, these filters could protect the CPU-blocks against. So the only thing they can do is exactly what they seem to have done: locking up with buildup. 😟


Did one of you open the Quad’s CPU-blocks to clean them inside, getting them closed ok after? As far as i see, there are no o-rings in there but some kind of flat rubber gaskets. So, if you wreck up one of them, you're done!

Mar 1, 2012 9:15 PM in response to Fipps

The blocks of the early Delphi LCS have air-channel fins, much like a straight tube condensor has air-centers. the later blocks used a technology first developed by IBM in the 70s called micro-channels. The channels are about 60u wide by 1.5mm deep. The channels are like micro capilaries and require a pre-filter to prevent getting clogged. However, as time moves forward, the block themselves will react in time as the corrosion inhibitor is used up in the loop. The mixed metals of the copper blocks and the aluminum RAD in time will have galvonic corrosion.


Even if the blocks are not leaking, the flow rate has slowed greatly. It's a strong sign that the block's micro-channels have corroded and are clogging up more as time goes on.


Since the Delphi LCS production for the G5 has ended, other retail blocks have improved to perform nearly as well. Swiftech has the micro-channel pins and there are a few europe makers of micro-channel blocks today. The micro-channel zone needs to be a 32mmx32mm area inside the block to cover over the center of the G5 CPU.


All this talk about fixing the LCS is fine. However, there have been others who yanked out the old LCS and put in their own WCing unit. The best one I'm aware of is an external box running two hose into the G5 case and 24-inch lead-wires to the external pump. Of course, the external pump is a DDC-1T or the DDC-1T-VC.

Mar 1, 2012 10:05 PM in response to Fipps

Fipps, you can find a link to a web page of my project earlier in this thread. It has photos of the blocks and filters, freshly opened, so you can also see the level of contamination a weakly working unit had when opened. In my case, the blocks were very clean and opening the screws was basically useless, i.e. the filters had worked very well. One of the filters, however, was more clogged up. There is a silicone O-ring in the block, but that doesn't seem to be the problem part in this version of the LCS design. Fwiw, I couldn't find a silicone O-ring of matching size where I live.


I cleaned the filters with very mild dish washing soap + hot water mix and then rinsed with hot clean water. Hot water alone would probably have done the trick. The type of scaly buildup in the filters as well as some crystally buildup in the rad required many rounds of cleaning.


Reverse flushing the filter with hot water several times should unclog it as much as it will unclog anyway. Leaving the CPU blocks and their hose connections alone is a good idea as it seems many people manage to break the plastic parts of the blocks in the process of detaching or reattaching the hoses. Opening the screws and opening the blocks that way, and later closing them, is less of a risk imho - just mark the screws so that you know how far to tighten them.


I'm with BMaverick on replacing the whole or most of the unit. That way, you will get something that's maintainable into the future and possibly performing better as well, and with reduced possibilty of coolant leaking into the case. The G5 LCS units have plenty of parts that when broken or damaged are hard to replace as is - or at all. Many northbridge water blocks from the PC realm look like something that could be quite easily fashioned into the G5 and the rest of the WC unit could be put outside the case. The Quads don't seem to complain much if some of the fans are removed or if the pump is removed.

Mar 3, 2012 3:25 PM in response to Heikki Lindholm

Thanks, Heikki, for pointing me to your website, which is a perfect step-by-step with pictures that leave little to the imagination!


Sad to admit, that im still to chicken to open these CPU-blocks. By now it's less for beeing afraid not to get them closed again, but more for not getting them open. Those phillips screws seem very tightened. I tried with force but i didn't want to spoil them ore worse, get the first two or three of one block losened and failing at the last. So i kept these things closed.


But at least know - thanks to your pictures - i had a peek inside and i'm not so worried anymore. As far as i see it a let's say somewhat limited througput is what can be expected, cause these channels inside the blocks seem to be very narrow and fine. Also extended flushing in both directions improved the situation to a point. To give you an idea, how much the pump pulls trough by now, i made some pic of a flushing session. What comes out there is what the pump is able to pull from an about 25cm lower level over both CPU-blocks. Vacuum on both intakes feels relatively even, "tested" with a finger touch. I hope the amount is fairly ok.


Replacing the hoses in my case, i think is not necessary, as they have little to no buildup inside and are still smooth an flexible. The inside is still black, even when dry. Also there didn't come much out of the radiator even after alot of flushing and shaking.


So maybe the thing can be refilled soon and it comes to the coolant again. As Sierra is not sold in Europe, choice is the same, you had: Thermochill EC-6 vs. Glysantin G48 / Water. If you had to do it again, which would be your choice?

This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

Working on the G5 quad liquid cooling system

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.