The Quad Chronicles - Restoring a HOT G5 Quad

[mac57mac57]

Agreed @iModFrenzy, if you don't have to, don't go ANYWHERE near a G5 Quad LCS. Tough, tough stuff, and most of us are electrical/computer type people, not plumbers... no experience with liquid cooling and none of the parts, tools and procedures needed to do this successfully.

I read through your air cooling conversion post, and I have to say that it sounds even MORE complicated than the LCS stuff... you are machining metal on top of everything else. Waaay outside of my experience set, not to mention available tools. Definitely, at least for now, "a bridge too far" for me.

I am blundering my way through it and learning lots as I go, but it is a massively frustrating task, completely outside of anything my 30+ years of computer experience has equipped me for. The good folks here at MacRumors are perhaps the most valuable resource available, and have been, and continue to be, a HUGE aid as I walk down this road.

I am adopting a "failure is not an option" mindset, but I have to admit that after nearly two weeks with no appreciable progress, I am getting mighty sick of tearing down Quads, changing something in the LCS and reassembling, only to find that this too did not work. Rinse and repeat!

Eventually.... I hope ... I will get at least one of these two machines back on its feet.

 

[Doq]

you are machining metal on top of everything else. Waaay outside of my experience set, not to mention available tools. Definitely, at least for now, "a bridge too far" for me.

I think this is my cue to say "you could just use pc coolers" and link the relevant thread.

I bid you good luck in reviving your Quads though.

 

[iModFrenzy]

Agreed @iModFrenzy, if you don't have to, don't go ANYWHERE near a G5 Quad LCS. Tough, tough stuff, and most of us are electrical/computer type people, not plumbers... no experience with liquid cooling and none of the parts, tools and procedures needed to do this successfully.

I read through your air cooling conversion post, and I have to say that it sounds even MORE complicated than the LCS stuff... you are machining metal on top of everything else. Waaay outside of my experience set, not to mention available tools. Definitely, at least for now, "a bridge too far" for me.

I am blundering my way through it and learning lots as I go, but it is a massively frustrating task, completely outside of anything my 30+ years of computer experience has equipped me for. The good folks here at MacRumors are perhaps the most valuable resource available, and have been, and continue to be, a HUGE aid as I walk down this road.

I am adopting a "failure is not an option" mindset, but I have to admit that after nearly two weeks with no appreciable progress, I am getting mighty sick of tearing down Quads, changing something in the LCS and reassembling, only to find that this too did not work. Rinse and repeat!

Eventually.... I hope ... I will get at least one of these two machines back on its feet.

Probably better ways to go about it(like @Doq said) than machining.

Also I’d pressure test the LCS with an air compressor(at a low PSI, see if it holds for X amount of time). Then proceed with filling it after you’ve got everything built. Less messy and will save you from finding out there’s a leak after you’ve powered the machine/LCS pumps on.

 

[mac57mac57]

@iModFrenzy, yes indeed, before reassembly this time, I am powering up and testing the pump and by extension, the entire cooling loop, by following the LCS pump control pinout linked in a post by @NikolaPPC. This involves using alligator clips to connect power from a spare power supply's MOLEX connector to the pump control connector. This will operate the pump and quickly show up any leaks without the need to reassemble the whole Quad.

I did not have any alligator clips on hand, so I ordered a set from Amazon. They arrive on Wed.

In the end though, everyone is pointing in the same direction... air in the cooling loop. No matter what you do, unless you have a reservoir and a valve to isolate it, you will always have at least some trapped air in the system. Does anyone have any tricks for bleeding that out?

 

[MisterKeeks]

If you have a decent fountain pump you can pump a great deal of solution through the system which will help force air out. Then you can close it while introducing as little air as possible. My system has some air in it, but it’s small enough that I don’t think it really matters.

One other way to purge air is to navigate the bubbles to the CPU plates. You can then remove the plates, add more coolant, and seal it back up. You’ll still have some air, but it might be less than before.

If you don’t have clear tubing, it’s going to be a nightmare.

 

[mac57mac57]

@MisterKeeks, I don't, and it is!

HOWEVER, today I went out to the hardware store, purchased clear hosing, a three way (plastic) T-insert with an integrated cutoff valve and some metal screw down clamps.

The game plan is simple enough now. Remove all the original Apple hosing, use the manual syringe pump I purchased last week to flush the radiator and then replace all the hosing with the new transparent stuff. Then, arrange the output line from the pump so that it has the T-insert inline with the rest of the (now transparent) hosing. Fill the system with a brightly colored coolant (so that it, and any air bubbles are clearly visible) via the top port of the T-insert, and when it is full to overflowing, plus a suitable time interval to allow any trapped air to bubble up and out, close the integrated T-insert cutoff valve. This should guarantee almost no air at all in the lines, and thus hopefully works first time.

I have not seen this approach anywhere else in these forums. Does anyone see why it might not work? It SOUNDS complete enough to me...

 

[mac57mac57]

BTW, a question to those of you who have replaced the original Apple hosing on an LCS... is there a commercially available tool to easily remove those darn clamps on each line? I did one "by hand" and it took forever, and a huge amount of mechanical force to accomplish. There HAS to be a better way! I would look this up online, but I don't even know what these things are called! Thanks!

 

[MisterKeeks]

@MisterKeeks, I don't, and it is!

HOWEVER, today I went out to the hardware store, purchased clear hosing, a three way (plastic) T-insert with an integrated cutoff valve and some metal screw down clamps.

The game plan is simple enough now. Remove all the original Apple hosing, use the manual syringe pump I purchased last week to flush the radiator and then replace all the hosing with the new transparent stuff. Then, arrange the output line from the pump so that it has the T-insert inline with the rest of the (now transparent) hosing. Fill the system with a brightly colored coolant (so that it, and any air bubbles are clearly visible) via the top port of the T-insert, and when it is full to overflowing, plus a suitable time interval to allow any trapped air to bubble up and out, close the integrated T-insert cutoff valve. This should guarantee almost no air at all in the lines, and thus hopefully works first time.

I have not seen this approach anywhere else in these forums. Does anyone see why it might not work? It SOUNDS complete enough to me...

I doubt this will work. It sounds like you only have opening to the system at the top of the Tee connector. You are putting coolant through this opening. How will air get out? You need another opening.

If the cutoff valve incorporates metal I’d still be concerned about galvanic corrosion.

BTW, a question to those of you who have replaced the original Apple hosing on an LCS... is there a commercially available tool to easily remove those darn clamps on each line? I did one "by hand" and it took forever, and a huge amount of mechanical force to accomplish. There HAS to be a better way! I would look this up online, but I don't even know what these things are called! Thanks!

Hose clamp pliers. I used standard pliers.

 

[mac57mac57]

Good point @MisterKeeks, thanks! I think I can accommodate that with another T-insert elsewhere in the hosing to provide an air exit while filling. My thought had been that the air would come out via the pump line, as it has when I have been filling via that (cut) line, but of course by adding the T-insert, I closed off that avenue. That likely would have occurred to me as I put this together, but you have brought it to the fore sooner - thank you!

The information on the packaging for the T-insert says that it is all plastic, so I think we are OK on that front.

I will give this a whirl as soon as my alligator clips arrive (Wednesday) and I can power up the pump independently of the Quad itself. I will post results once I have them.

Thanks again!

 

[mac57mac57]

Big day today, and big delay starting tomorrow!

Big day today - today my alligator clips were delivered and so I can finally test my latest LCS recharge and see if it leaks. To do this, I will power the pump independently of the Quad itself by using the alligator clips to connect the 5v and 12v a free standing power supply's MOLEX to the pump connector. This will crank the pump at full speed and any leaks should be IMMEDIATELY apparent!

That will be it for several days however, as we quite unexpectedly have family arriving for a multi-day visit tomorrow AM. That will be lots of fun, but no more Quad work will occur during that time. So, I haven't given up, I just won't be able to focus on this for a few days now.

 

[Certificate of Excellence]

I know this is premature but I cant help but think about the coolant. If you use a car coolant, manufacturers add a UV reactive fluorescent dye (for mechanics it makes it easy to spot a leak). With you adding the clear tubing, a UV LED would look awesome in that case with just the deflector in. I know my kids love the LED case lights I've put in my PCs & PMG5s.

 

[mac57mac57]

Similar thinking here, @Certificate of Excellence. I have purchased both transparent hosing and deep red colored coolant. I think it will look fabulous when it is in service.

Our guests leave tomorrow and I will be able to get back to working on this then.

 

[mac57mac57]

The new "tools of the trade": transparent hosing, Blood Red EC6 coolant, alligator clips and metal screw down hose clamps. On Monday, I should be able to put all this into action and see how it goes.

 

[mac57mac57]

Finally, some progress to report today...

Just before taking the plunge and removing all the existing LCS hosing and completely rebuilding the LCS cooling loop, I decided, with nothing to lose, to try one last "quick and dirty" LCS recharge option: cut the hose coming out of the pump and use one end of it as the drain/fill line, and the other as the air exit line.

I did this, drained the existing coolant (which is the coolant I added on my last attempt), added new coolant (got nearly 200 ml in using this approach, vs. the ~140 ml I got in last time) and then sealed it in a novel way. The inspiration for the sealing approach was this:

I literally just happened to have a small length of 1 cm diameter metal tubing about 1.5" long. I inserted one end of it into the pump exit hose cut and one end into the other side of the hose cut and then with some effort, pushed the two ends together until you could no longer see the insert. I then sealed both sides with construction grade building tape and then "clamped" each side, in this case with cable ties. A day or two after I did this, I found the metal screw-down clamps that I really needed, and set them to the side, on standby. The net result looked like this:

Then I took a free-standing PC power supply with an independent on/off switch, added a MOLEX extension to it, cut the other end of the MOLEX extension off and stripped each of the four wires to expose the copper underneath. I then used my newly purchased alligator clips to connect +5v, +12v and GND from the MOLEX to pins 1, 5 and 7 of the LCS pump connector respectively.

Ready to go! I turned on the power supply and ... NOTHING! The power supply did not come on. I checked all the connections and then retried. Still nothing. Thinking that perhaps the pump simply doesn't draw enough amperage to activate the power supply, I hooked up an old Macintosh hard drive to another of the power supply's MOLEX connectors. That did the trick... the hard drive spun up AND the LCS pump took off.

The test setup then looked like this:

It is running now (has been for 10 minutes or so) and there is no leakage. According to what I have read, the pump should be running at full speed, and it is definitely audible, but it makes less noise than I expected.

Now for the rub... I have NO idea what sort of metal the insert was made out of, and so no idea if this is a safe configuration or one that will kick off galvanic corrosion. Really, this was just a watertight connection technology trial - I expected it to fail. I am delighted that it seems to be water tight, but I will let it run like this for a few hours to be sure.

If there is still no leakage, I will reassemble the whole thing into the Quad, cross my fingers and toes, pray to the gods and fire it up. Maybe, just maybe, I will have dodged the "rebuild the LCS" bullet...

 

[mac57mac57]

Well... no such luck. I have not dodged the bullet after all. I reassembled the Quad with the latest LCS recharge, and the machine runs just fine, but the fans still quickly ramp up to 3200 RPM and stay there.

This makes the machine intolerably loud and so it is time to move on to the last step, the one I have been dreading... a complete rebuild of the cooling loop with transparent hosing, brightly colored coolant and of course, a preceeding repeated flush of the radiator.

Despite the noise, CPU temperature are nicely in range, in the mid 40s, up to the mid 50s under load. These are good safe values. If the fans didn't make so much noise, I would be happy to declare victory... but they do.

So... on to the next step; this may take weeks yet.

If YOU are thinking of servicing a Quad LCS, be prepared for a long and difficult journey. As William Shakespeare wrote in MacBeth, you will need to "screw your courage to the sticking place" and dive in with the full expectation of a monumental struggle. This is nasty stuff!

 

[mac57mac57]

Well...between iStatPo and XRG, I was able to see that CPU B was the problem child...CPU A was idling in the mid-40s C while CPU B was in the mid 60s C. I decided to try repasting just CPU B before embarking on the full LCS rebuild. If I could lower CPU B into the 40s, maybe the fans wouldn't need to run so hard...

This was successful, just not successful enough. CPU B now idles in the low 50s C, so repasting lowered the temperature by 10 C, a very good result, but the fans are running at full tilt to maintain that.

I find myself wondering if I could just swap out the Apple fans and replace them with something much quieter. Both CPUs are now operating in a "safe" temperature range. With quiet fans, I could probably leave well enough alone. I will do this simultaneously with the full LCS rebuild; LCS rebuild on one Quad, and quieter fans on the other. We will see how that turns out.

Both the front and back Quad CPU fans are 90 mm fans, which complicates things a bit, since 92 mm seems to be the "new 90 mm" - most 90 mm fans are in fact 92 mm. 92 mm will NOT fit into the tight confines of the existing fan enclosures and space. I have one 90 mm fan in stock, but its maxes out at 1500 RPM, almost certainly not fast enough to maintain cooling. So, the hunt is on for a true 90 mm fan that can rev up to at least 2500 RPM with a full speed noise level of less than 34 dbA or so.

This is proving elusive to find, but I have only searched a few places so far. I am hoping that OWC or some such may carry fans of this dimension, speed and noise level.

 

[MisterKeeks]

If the machine is running at 50 Celsius with fans on max then the liquid cooling system is not performing correctly. I think it’s unlikely you’ll be able to find fans that can be quiet while also proving the airflow to compensate for a faulty LCS. It’s also not a given that the PWM protocol that the G5 uses is standardized - I don’t know - so you might also run into issues with the fan control not working correctly and/or the system being unhappy with n the replacement fans. Again, I don’t know. But I think trying to change the fans isn’t going to solve the fundamental problem you have, and it’s going to introduce new thorny problems that you’d probably rather not deal with.


Before you mess around with fans, I’d try to do a thorough rebuild of your other LCS. It’s annoying, yes, but if you do it successfully once I suspect you’ll find repeating that process is preferable to trying to use new fans to cover up this problem.

My $0.02.

 

[mac57mac57]

I absolutely agree @MisterKeeks, changing out the fans does not solve the underlying problem; it merely buys some time. The LCS *is* failing; it *will* fail to the point where even full blast fans won't keep the machine running. I understand that. That is why I am embarking on a full LCS rebuild for the "other" Quad.

However, noisy fans have been a bug-a-boo of mine since 1999, when I had the incredible misfortune to replace my aging main PC (a really excellent Dell R450) with a new HP Pavilion 9800. That machine was SO loud from "day 1" that I could not stand it and ultimately simply gave the machine away to a local GoodWill store. It was just a case of cheap, noisy parts: power supply fans, CPU cooler fan and a noisy hard drive.

At the time, I was not yet into building my own PCs, so I lacked the skills, tools and parts to change all those noise makers out. For a year or so I toughed it out and then I had simply had enough and replaced the machine with a new custom built "quiet" PC that was SO quiet I dubbed it "Whisper". That was my last PC. After several years of fine and very quiet service, I replaced that machine with my first personal Mac (had used them at work for years), the Power Mac G5 DP 2.3 GHz machine that I still have. I have been driving Macs ever since.

So, I am interested in replacing these fans anyway out of both historical imperative and of course, curiosity. As Albert Einstein said always have "holy curiosity"!

 

[mac57mac57]

OK! The process of rebuilding the LCS has begun! I extracted the LCS from my "other" Quad, cut two of the hoses and drained it, stupidly not capturing the drained old coolant (so I don't know how much there was), cut away most of the excess old hosing and then began the arduous task of removing the metal clips Apple so thoughtfully placed on each hose where it attaches to its mating barb.

Those clips must have been designed and manufactured in hell - they are diabolically difficult to remove! The amount of mechanical force needed to remove each one astonishes me. I had to use two sets of heavy duty pliers and a LOT of force to accomplish the job. They are now all removed except for two inside ones... I have to completely remove some hose ends from adjacent barbs to create enough physical clearance to work on these last two. I suspect I will have to cut off the hose ends with an Exacto knife / box cutter. I can neither pull them off nor twist them off.

Once this is done, I will hook the radiator up to the Syphon Pump I bought earlier and repeatedly flush it until it is running more or less clear. Then I can start rebuilding the cooling loop with the new transparent hosing and screw down metal clamps I have purchased.

For now, two clamps to go and then onwards!

 

[bkmoore773]

OK! The process of rebuilding the LCS has begun! I extracted the LCS from my "other" Quad, cut two of the hoses and drained it,....

I rebuilt an LCS on my G5 Quad with mixed results. (Once CPU passed thermal calibration, the other one didn't.) I think the main problem is you have to clean out all the mineral deposits in the radiator and the cooling blocks, otherwise it won't properly cool the CPUs.

 

[mac57mac57]

I suspect that you are right, which is why my next step is a full flush of the main radiator with a vinegar/water solution, pumped through under pressure via the manually operated siphon pump I purchased just for this purpose. Rinse and repeat until the fluid coming out is as clean as the fluid going in.

Did you encounter as much trouble as I did getting those ...darn ... metal clamps off the original hosing?

What I don't think anyone can do is flush the per CPU "radiators"/cooling blocks... this looks like a completely closed system, at least to me. Are there any tricks by which these two can also be flushed?

 

[mac57mac57]

A little more progress today. I got the LCS fully torn down, finally removing ALL of those insidious metal clamps Apple put on the hosing.

I am all cut to shreds from running into metal edges while placing enormous mechanical force onto the clips to remove them. I have three really nasty abrasions and a few lesser ones. An LCS rebuild is DEFINITELY not work for the faint of heart or body!

I don't believe I have ever seen a photo of an LCS all torn down, so here is one:

Regrettably, the amount of physical force needed to remove the clamps snapped off the barb on the very last one that I was working on, which was one that directly connected to the CPU 2 pad. Now I have to very carefully epoxy that back into place and give it LOTS of time to cure before proceeding further.

I can also see that I have squashed quite a few of the radiator fins, so those will have to be painstakingly straightened so as to get maximum cooling efficiency when this is all put back together.

One final note. While removing the existing hosing, I discovered that there are small inline filters in several of the hoses; as I recall, there were at least two of them, one in each of the two hoses that go into each of the CPU pads. Since all the old hosing is now gone, those filters are gone too, and I will not be placing them into the new hosing. No matter what, they obstruct the free flow of coolant. They served a valuable purpose, but with all the flushing and cleaning that is going on as part of this process, I think that this purpose has been met via other means, at least for now.

 

[mac57mac57]

After a brief hiatus for several days of yard work, I am back at it. Today I flushed the radiator of the LCS using a hand operated siphon pump. The input hose was placed into the fluid source, while the other hose was connected to the intake of the radiator. I flushed several loads of a 50/50 mix of distilled water and vinegar through it, followed by several loads of simple distilled water. Then I drained it completely, or as far as I was able by pumping just air through the system. Now I will let it all dry thoroughly before proceeding to the next step, which is epoxying the broken CPU pad barb back into position.

Here is the setup I used to flush the radiator:

As you can see, the source fluid was held in a gallon distilled water bottle, which I filled with whatever I was flushing with at that time. You can see the other hose connected to the intake of the radiator, and finally, partially hidden by the gallon bottle is the hand operated siphon pump.

The fluid leaving the radiator was pretty clean right from the start, but I pumped the full intended volume of cleaning mixture through it anyway, at the highest pressure I could create with the hand operated siphon pump. I captured the output from the outlet of the radiator into a measuring cup from time to time to check its clarity. It seemed very clean pretty much from the start.

Flushing the radiator is a wet, messy job. Be prepared for that if/when you decide to do something similar.

Next up: rebuild the cooling loop itself and connect it all up to the LCS, prior to refilling it with the actual coolant.

My guess: I should be able to rebuild the loop, fill it, and reassemble the Quad, all within the next three to four days. Then, I will cross my fingers, pray to the gods and power it up. With any luck...

 

[mac57mac57]

Well... I may have hit the end of the road. The CPU pad intake barb, which sheared off while I was removing those ^%$#@& metal clips Apple so considerately put on each hose end, is proving to be impossible to reattach with any form of epoxy, Super Glue, etc. It breaks off again every time I try to attach a hose to it, or if I pre-attach the hose, when I attempt to attach the other end of that hose to the radiator. I am trying one last bit of serious MacGyver'ing to overcome this, but if that fails, I have no place else to turn.

This is the culprit (the sheared off piece, and the location it sheared off of, are both circled in red):

Does anyone out there have that particular black plastic piece, with the CPU intake and outflow barbs on it, or even the whole assembly it is part of? I would be happy to pay for the part(s)... thanks!

 

[mac57mac57]

BTW, live and learn. After extensive hunting, I found the below tool, which does a MARVELOUS job of removing those clamps without the sort of heavy mechanical force needed to do it without such a tool.

I purchased this on Amazon.com for the princely sum of $7.17. In case you are interested in getting one too, here is a screen shot of the listing, so that you can find it easily: