|Mar 4, 2011, 06:24 PM||#1|
LCS repair info for Powermac G5 dual 2.5
I want to share my experiences with my G5 beginning with the symptoms that led me to discover that my LCS (liquid coolant system) had been leaking. Then share what I have done to the machine over the entire repair.
My machine specs:
Dual 2.5 Ghz running 10.5.8 used mostly for processing and manipulation of digital image files. It has 8 GB of RAM and 2 internal hard drives. The video card is what came with the machine. Other than adding the extra hard drive and the extra RAM there are no other hardware additions. I have a 17" and 23" Apple Displays running off the single video card. Externally I have 2 LaCie Pocket Drives (20 GB & 40 GB) and a 1.5 TB Western Digital MyBook hard drive all daisy chained via firewire.
This machine is usually running 24/7. On occasion when I had to restart, usually for a software update, I would get the chime, solid status light (the light above the power button), and the folder with a question mark. In the beginning I could do a hard shut down (hold down the power button) and try to boot again with success. I would then run Disk Utility and repair the permissions and go on about my business.
Over time the machine on start up would chime but the monitor would not come on and the fans would slowly speed up until they sounded like a jet fighter taking off. If I didn't do a hard shut down the machine would eventually shut off.
Sometimes I could restart from the OS install disk, run the disk utility, and the machine would start up fine. When that solution stop working I would dig around on the web for similar experiences and try other solutions. With the machine being 6-7 years old I thought that it might be the internal 3.6 volt battery and replaced it. Didn't seem to make a difference.
---I DID CALL APPLE AND SEE IF I COULD GET A FREE REPAIR OR A NEW MACHINE LIKE OTHER FOLKS I READ ABOUT WHEN THEIR LCS FAILED, EVEN IF THEIR MACHINE WAS OUT OF WARRANTY AND OUT OF APPLE CARE. IT SEEMS THAT AFTER 5 OR 6 YEARS FROM THE MANUFACTURER DATE APPLE WOULD TAKE CARE OF IT. MY MACHINE IS 7 YEARS OLD AND THEY NOW CONSIDER IT 'VINTAGE' AND PARTS ARE NO LONGER MADE FOR IT. I MIGHT TAKE IT TO MY LOCAL APPLE STORE IF MY ATTEMPT AT A REPAIR DOESN'T WORK AS A PLAN B------
Funny enough...if the computer was kept turned off over night it would boot the next day just fine. Now at this point it was obvious that it didn't like being turned off so I would wait to do any updates on Friday and the just power down the machine and hope it would start up on Monday.
I would also try to reset the PRAM, reset the PMU (pressing the coda switch on the mother board), and pull the power cord out of the machine for a couple of minutes before reconnecting and attempting to boot up.
In the final days leading up to the eventual inability to get the machine to boot I would be batch processing jpeg images in Lightroom and the machine would spontaneously go to sleep. At first I thought I had a Hot Spot active in a corner of the monitor that would activate the sleep mode. That was not the case.
After it would go to sleep I would wake it up by clicking the mouse or hitting a key. It would wake up, continue to process files, and after a minute or so it would go back to sleep.
I did have other apps running like Safari and PhotoShop (usually with a couple hundred megabytes of files open).
I had a Temperature Monitor app running that would show me the CPU temps and they would usually hover around 180 F and 200 F. The fans were usually running around (my estimate) 50%-75% power.
The final state of the computer before I started to check for a leak was upon start up I would get the chime, solid status light, no diagnostic LEDs internally, no video, and the machine would not respond to keyboard commands. So, I couldn't boot in another mode like safe, verbose, open firmware, or target. Also couldn't reset the PRAM or force open the CD drawer because the keyboard was unresponsive.
Thinking that it was a temp issue I throughly vacuumed out the inside of the case, radiator, and the outside of the case. I also cleaned all the fans. Still the machine wouldn't start.
I went exploring inside the machine looking for any gross fluid. I didn't see any but discovered that the absorbent pad under the LCS was saturated. I also tilted the machine and after a short time a small amount of liquid would appear. And there it was....my machine had sprung a leak.
If you are reading this then you might have read other posts on this site and others and learned about the horror stories on how the coolant killed their machine by either coming into contact with the logic board, processors, or power supply. I think I might have gotten away with no additional damage but don't know for sure until I get the LCS repaired and attempt to start up the machine.
As of this post I have removed the LCS and drained it by removing one of the hoses.
-----BE AWARE THAT THE SYSTEM MIGHT BE UNDER PRESSURE. THE COOLANT IS NOT VERY FRIENDLY TO SKIN, EYES, AND REALLY ANYTHING. HAVE EYE PROTECTION ON IF YOU ATTEMPT TO REMOVE A TUBE!--------
I have also removed the power supply and checked for any corrosion---there was a couple of trails of dried coolant on the outside of the power supply but no corrosion or visible damage. I put the power supply back.
I followed the post at http://www.xlr8yourmac.com/systems/G...ant_leaks.html to guide me on the LCS removal and disassemble. The plates that sit on the processors showed some corrosion around the edge and dried coolant. I cleaned the inside of the LCS behind the plates where there was a 'jelly' that I can only assume was old coolant.
It seems that the o-rings were the culprit in allowing the coolant to leach out of the LCS.
---THE INSTRUCTIONS LISTED AT http://www.xlr8yourmac.com/systems/G...ant_leaks.html ARE FOR A POWERMAC G5 DUAL 2.7. I FAILED TO ACTUALLY MEASURE THE O-RINGS THAT I TOOK OUT OF MY DUAL 2.5 AND JUST ORDERED THE RINGS LISTED IN THE POST. THEY ARE TOO LARGE FOR A 2.5 MACHINE--------
I purchased the o-rings at MSC Industrial Supply. 800-645-7270 and www.mscdirect.com. I was able to find the coolant listed in the xlr8yourmac.com post at www.performance-pcs.com for $19.95. The brand of the coolant, Cooler Master, doesn't carry that particular coolant any longer. The O-rings were around $13 for 50 rings (the smallest amount I could order).
After learning that the o-rings I ordered were to large I ordered the next smaller size. The specs on the smaller o-ring have a thickness of 1/16", Outside Dimension (O.D.) 1.375", and Inside Dimension (I.D.) 1.250. It is made of Viton, which according to a tech person at MCS is the best material for coming into the coolant (Glycol) and be under some pressure and heat (rated to 400 degrees F). The order number for this o-ring at MCS is 02049971. The AS568A Dash No. is 026. The Dash number is a standard across multiple industries based on OD, ID, and thickness.
I will conclude this post by saying that I have the coolant and I am waiting on the replacement o-rings which will be here in 2 days. If they fit then I will begin to reassemble and begin the process of recharging the LCS with coolant. I have read on other sites a bunch of methods to get the coolant into the radiator so there are no air bubbles. I will share my experience with that in a later post. I will also replace the rubber tubing on the LCS with something newer and hopefully clear to I will be able to visually see if the coolant, which is blue, is still in the LCS.
I am going to see if I can power up the pump on the LCS with an external power source and make sure it is running and might take it apart and clean it out or at the very least run water through it. IF I need a new pump I have read other posts where a pump located at http://www.newegg.com/Product/Produc...82E16835108063 is $60. Better to learn that the pump on the LCS is broken now that the LCS is out of the computer.
I hope this post is helpful. I will answer any questions as promptly as I can.
|Mar 7, 2011, 05:13 PM||#2|
The O-rings arrived
The rings arrived today just as expected by MSC industries today.
They are the correct size for my PowerMac G5 dual 2.5. Let the reassembly begin. I going to replace the tubing with clear tubing on the LCS and try to use an external water pump to pull the coolant into the system.
The O-ring info that works can be found in the original post in this thread.
|Mar 7, 2011, 05:42 PM||#3|
Since you are replacing the O-Rings, leave the bottom CPU blocks off and fill it that way. I actually found that to be the easiest way to refill the system without using a vacuum.
Great write up BTW. 5 stars to you and I'm sure there will be many googling and ending up on this thread.
Good luck to you sir. Let us know how it all goes.
|Mar 8, 2011, 01:29 PM||#4|
slight set back
In the ongoing saga that is the repairing of my LCS I have over estimated my strength and broke one of the plastic stems that the tubing connects so the coolant can have access to the plate that touches the processor. I am glad my kids were not around to hear my wonderful and Oscar worthy performance of an agitated sailor as I went on a profanity filled monologue for a good while.
That was late last night. I did some research on the best adhesive for connecting plastic to plastic on different manufacturer sites and I think Loctite makes such a product that I hope will do the job. I am off to find it, make the repair, let it sit overnight and continue with the rebuild tomorrow.
The current state of the repair is I flushed out the system with water and removed all of the old tubing. I simply put a piece of tubing on one port on the radiator and held the other under the faucet. While I have a water pump for an outside fish pond I didn't want to risk damaging any internal components since it moves a lot of water quickly. I drained all the coolant when I removed the LCS from the computer so there was only water that came out. The water flow out of the LCS didn't seem to be as fast as the water going in I would attribute that to the many turns withing the heat transfer block (the large 'wall' with all the soft metal folds).
The rubber tubing had to be cut off with an exacto knife because there was some corrosion on the metal where the rubber tubing connected to the aluminum tubing. I disassembled the pump and cleaned it out (not as difficult as it sounds). Its 2 torx screws to remove it from the LCS and 4 torx screws to release the top plate. I also used a dremil tool with a wire brush attachment to remove the corrosion on the aluminium tubing tips.
While I was wanting to use clear tubing, so I could see if the coolant was flowing, I was not too trusting of some others doing rebuilds using tubing found at their local hardware store. My concerns were with the temperatures melting the tubing, chemical resistance and possible pressure failures. While MSC industrial supply (www.mscdirect.com) carries some possible clear tubing that can be purchased by the foot for less than $1.00/per foot, I have been without my machine for over 2 weeks and didn't want to wait for the order and possibly purchasing the wrong size as I did with the O-ring.
----MY CUSTOMER SUPPORT EXPERIENCES WITH MSC HAVE BEEN NOTHING SHORT OF TOP NOTCH. I CALL AND TELL THE SALES REP WHAT I AM NEEDING. SINCE EACH TIME I HAVE CALLED MY QUESTIONS HAVE BEEN OF A VERY TECHNICAL NATURE. THEY ARE MORE THAN HAPPY TO CONNECT ME WITH A TECH PERSON. MY QUESTIONS WERE RELATED TO THE BEST MATERIAL FOR MY APPLICATION WHEN PURCHASING AN O-RING, WHICH CLEAR TUBING IS BEST, AND EVEN HELPED ME FIND SOME POSSIBLE SOLUTIONS TO BE ABLE TO VISUALLY CHECK COOLANT FLOW IN LIEU OF USING CLEAR TUBING. I NEVER FELT LIKE THEY WERE RUSHING TO GIVE ME AN ANSWER AND TOOK THE TIME TO DO THE RESEARCH AND DOUBLE CHECK THAT THEIR SUGGESTED SOLUTION MET MY SPECIFIC NEEDS OF CHEMICAL RESISTANCE, TEMPERATURE, AND PRESSURE. AND IF YOU ORDER BY 8 PM EST YOU CAN HAVE IT DELIVERED NEXT DAY AT NO ADDITIONAL CHARGE. THE ICING ON THE GEEK CAKE IS THAT THEIR CATALOGUE (CALLED 'THE BIG BOOK') IS A 6-7 POUND HARDCOVER BOOK THAT IS OVER 5" THICK. YOU CAN REQUEST ONE AT NO COST AND THEIR WEBSITE SHOWS, IN THE ITEM DESCRIPTIONS, WHAT PAGE IN THE BIG BOOK YOU CAN FIND THE ITEM. YOU CAN READ MORE ABOUT THE ITEM, TYPICAL APPLICATIONS, AND I HAVE FOUND IT AN IMMENSELY USEFUL AND HELPFUL RESOURCE FOR EDUCATING ME ON SOLUTIONS AND PRODUCTS I AM NOT TOO FAMILIAR WITH. COMBINE THAT WITH THE CONVERSATIONS WITH THE MSC TECHNICAL PERSONS, I FEEL CONFIDENT WITH MY CHOICE OF ITEMS FOR MY REPAIR. AND ON A DIFFERENT NOTE THEY ALSO SELL LOTS OF OTHER ITEMS RANGING FROM THE MORE PEDESTRIAN TO THE TECHNICALLY EXOTIC. BEING A STRINGENT DIYer I AM LOOKING FORWARD TO TACKLING SOME HOME PROJECTS THAT WERE HAD TO BE LEFT TO A PROFESSIONAL BECAUSE I COULDN'T GET ACCESS TO THE SAME MATERIALS LIKE CLEANING MY HVAC EVAPORATOR COIL. ITS AN EASY ENOUGH PROCESS TO DUPLICATE AFTER WATCHING THE HVAC REPAIRMAN; SIMPLY OPEN UP THE UNIT AND SPRAY ON A CHEMICAL AND WIPE AWAY ALL THE GRIME. MSC SELLS THE COIL CLEANER SO NEXT TIME I CAN SAVE SOME $$ ON A SERVICE CALL!!! IF YOU LIKE READING MANUALS AND TECH SHEETS THEN YOU WILL FIND THE MSC 'BIG BOOK' A WONDERFUL READ. I WON'T SPOIL THE ENDING BUT WILL SAY THAT IT IS A BOOK THAT YOU HAVE TO PUT DOWN FROM TIME TO TIME SINCE IT WEIGHS SO MUCH -----
I choose to go to my local auto parts store and purchased radiator tubing and will be using that. MSC also has lots of options that you could incorporate into your rebuild like flow indicators and flow sights. Both will allow you to visually detect if coolant is flowing. If I find that refilling the radiator is not too difficult I might add some kind of piece that will allow me to visually check coolant flow at a future date.
I have before and after photos that I will post later of the steps discussed in this post.
The lesson here is before you attach the tubing to the tips, be sure to have ALL the tube clamps on the tube so after you attach the tube on both ends you have the clamps on so you can put them into place. The connection I was working needed one piece of tubing going from the radiator to aluminum tubing to the plastic part in the base of the LCS that has 2 tips. I had 3 clips on instead of 4 and needed to remove the tubing from the radiator (which was the easiest to grasp) to add the additional clamp. It took a tremendous amount of exertion to get the tubing off. I was still pulling with lots of strength when the tubing released, unexpectedly, I ended up pulling back away from the LCS too far and pulled on the aluminum tubing, which was already connected to the plastic port and broke it clean off.
I have every confidence that the adhesive will handle the pressure and heat (i read the technical sheet that is available on the Loctite site). But will run the system out of the computer to check for leaks and make sure the adhesive will do the job.
The plastic plate that has one of the 2 ports that I broke is just a simple item that allows coolant to flow into one port, make contact with the metal plate that touches the processor, then flow out the other port and continue onto the next processor. If the adhesive ends up not holding I will have to either have a machine shop make me something that does the same job or possibly modify the plastic piece by drilling out the broken port and use some kind of solution that uses items available at my local hardware store using off the shelf plumbing parts. Funny how the least expensive components can cripple a very expensive machine. Murphy's laws seems to repeatedly come to mind.....
Thanks for reading and I will update soon. Should you have any comments related to my approach please feel free to post as they might help me and other DIY'ers willing and crazy enough to attempt this fix that might read this thread in the future. The first time always take longer. Maybe I will get another G5 and do a repair on it so I can have another great workhorse of a machine. Ram is cheap enough to load up a new machine. Ah...to dream
|Mar 8, 2011, 01:35 PM||#5|
Sorry to hear you broke the block. Sorry I stopped reading after that but I do have an extra block if you want it, just send me your address and I'll send it to you.
I'll read the rest of your post later when I get more time.
|Mar 16, 2011, 02:13 AM||#6|
LCS refilled and testing for leaks
First, I haven't forgotten about the photos of this repair that I said I will post. I just wanted to get down on virtual paper of the trials and tribulations of what has been quite a journey and it isn't over yet.
When we last left our hero he was waiting for the very technically specific piece for his super computer commonly known as....wait for it....THE O-RING!
So the new O-Ring fits like a charm. Let the reassembly begin!
I used jewelry cleaner to get the junk out from in between the copper blades of the heat block after I used an exacto knife to go between the blades and scrape out gunk. I was careful to not bend the blades to much but just enough to get the knife in.
I put in the new O-Rings and used automotive gasket sealant before attaching the heat blocks to the LCS. Be sure to tighten the screws uniformly so the gasket sealant makes a good tight seal. The repair I did on the broken plastic lead that connects tubing so the coolant can come into contact with the heat block worked like a charm. Another note here is if you use plumbing clamp on the plastic parts be sure that the large part of the clamp, where you use your screwdriver to tighten the clamp, isn't moving the plastic part off center. If you don't have the plastic part screw holes lined up with the metal screw holes in the LCS plate you won't be able to screw the heat block plates back. If you remember that little tid bit then you just saved yourself 30 or so minutes trying to fit a screwdriver or needle nose pliers into the middle of a bunch of metal and rubber tubing to move the clamp.
I disassembled the pump to clean it out before putting back onto the LCS. Remember that this thread is about a G5 dual 2.5, which has a single pump. ------REVISION: I would apply the gasket sealant also to the pump the next time I do a replair. Couldn't hurt and as long as the pump was open it takes no time add the sealant.----The pump has 4 screws holding on the 'cap'. Inside is a large metal half sphere that is solid and it has some weight. It is magnetically held in place, which is how the motor spins it to move the fluid. You can simply pull it out but the magnet is strong so there will be some resistance.
Upon disassembly there was only some mild 'crust' in the bowl where the metal sphere goes. I just scraped it clean and put the motor back together.
I tested the motor by attaching 2 leads to metal probes in the housing that connects the motor to the computer power supply. In my research regarding how to power up the motor I found info that didn't work for me so I had to figure it out on my own.
I first tried to plug the motor back into the computer and that didn't work. I had an old 12v power supply block that I put small clips on so I could attach the power supply to the motor leads. It was a bit of trial and error but I was able to get the motor to operate.
The motor has a 6-lead connector. Where the wires go into the back of the connector you can see numbers above the wires. Not all the wires are attached to metal leads. I put the clips on 2 and 4. Since I was working with liquid I filled the LCS in my bathroom. I plugged the power supply into a GFCI outlet (it has a test and reset button) that I was able to use as a switch instead of plugging and unplugging the power supply.
I replaced all the tubing since its cheap and the inside was very cruddy. I went to my local auto parts store with the old tubing in hand for sizing and got 3 feet of radiator tubing. It is the 3/8" flavor. I also got a foot of some clear tubing that is rated at 175 degrees F and 75 lbs of pressure, which I expect (and hope) is enough. While my favorite industrial supply company, MSC, has clear tubing that has higher temp and pressure rating I didn't want to wait any longer to get my computer back up and running. The clear tubing will allow me to visually check that the coolant is moving though the system.
This is where my rebuild takes a turn so I hope you are wearing your seat belt.
As I hope you are able to use this thread as one of many threads to educate you on what others have done and learn from their successes and mistakes. My original concept for my repair was to put a valve inline. I did this using brass plumbing fittings. I went to Home Depot and stood in front of the brass fittings looking for the parts I needed. There are compression fittings and threaded fittings. I found a 3/8" compression-style valve and a standard threaded Tee fitting. The Home Depot guy said the two can't be connected. I was thinking that I was going to have to order the parts but thought I would go to another hardware store and guess what....Ace Hardware had the part that was able to connect the compression valve with the threaded Tee fitting. I attached 1 foot of clear tubing on the valve. More on that later...
I put 3/8" brass barb fittings on each end of the Tee to connect the Tee to the tubing. I reused the clamps that were originally on the LCS and swopped out some plumbing clamps. Don't forget to use teflon tape on the treads before you attach the fittings to each other for a water tight fit. The threaded fittings are tapered so it will appear that the fitting isn't screwed in all the way since you will see threads after you have tightened the fitting. That is normal and expected. The taper helps to make a tight fit.
Now the tubes and clamps are on, the Tee and valve are in place, with the exception of connecting the tube that comes off the output port on the motor (there is an arrow showing the direction of the flow next to the port). The goal starting was to open the valve on the Tee, attach the coolant bottle onto clear tubing coming off the valve, and fill until there is coolant coming out. I didn't have an aquarium pump to pull the coolant, as seen on one thread, and running the pump didn't pull the coolant out of the bottle that I thought it would. I got a hand pump at an auto parts store but that didn't work as it induced air into the LCS.
I drilled a hole in the top of the coolant bottle so I could easily attached the bottle to the tubing going to the valve. However, I needed to also poke a small hole into the side top of the bottle so I could use my finger to cover the hold when I am squeezing coolant into the LCS and when I release the bottle I would move my finger letting air back into the bottle so I can resqueeze the bottle to put in more coolant. If the air hole wasn't there when I released the bottle after squeezing coolant into the tube it would get sucked back into the bottle.
I have more to add but going to stop for now. My next post will cover using the valve in order to bleed out air from the LCS.
Thanks for reading
Last edited by photonorm; Mar 17, 2011 at 11:48 AM. Reason: adding photos
|Mar 16, 2011, 01:11 PM||#8|
Great news!! Its alive
If I was a betting man (and I am) I would have bet against the computer starting up after putting it all back together. I was prepared for that possibility this morning when I begun to put the LCS back into the computer and power it up.
Well....I would have lost that bet because it started up on the FIRST TRY! Nothing unusual or out of the ordinary. I only had one monitor, keyboard, and mouse plugged in. I pressed the power button and heard a chime. The fans spun up, the monitor came on and I said, "It can't be this easy, can it?!"
First got the folder with the '?' and thought ok, here we go but I can get through this problem as I have had to do it before. But after a couple of seconds it went away and I saw a blue screen. I felt like a kid waiting for Santa or for the Tooth Fairy. The anticipation was spectacular, but reserved and before I new it I saw my desktop!!!
I wanted to share the great news now and I will return later to post the images, update on how the temp readings are, and all things of importance (as I view it) on this repair.
As of this moment the fans are running, I think, faster than they should but that may only be that I need to recalibrate them since I removed the processors from the logic board. The computer is idling with a few apps running, but the apps are not doing anything. The temps are reading as follows:
CPU A Die: 99.4ºF
CPU B Die: 101.5ºF
Processor Card A Ambient: 78.3ºF
Processor Card B Ambient: 77.0ºF
These values are from the app Temperature Monitor. The app tells me that the Cooling target temperature is 180ºF. I found that value when I viewed the 'Show System Info' and clicked the 'more info' button.
I can recall operating at temps in the 190-210ºF range!! I guess I am very lucky that my chips didn't turn into charcoal!
That is all for now. I can't enjoy this victory for too long as the machine I am writing about is my main production machine and I have been without it for 3 weeks. I have so much work to catch up on but I will eventually return with the photos and details.
Thanks for reading!
|Mar 16, 2011, 06:56 PM||#9|
Running ASD to do a Thermal Calibration
I was able to do some work and put the computer through some processor intensive tasks. The max. temp. I saw was about 155ºF and that was a peak value. With the processors working at 80%+ for a good stretch while it was creating, exporting, and uploading a website out of Adobe Lightroom the temps. hovered around 135-150ºF. At idle the temps. would drop down to around 100ºF (+/- 10ºF).
I downloaded the Apple Diagnostic Disk (ASD) 2.5.8 from a file sharing site and created a CD so I could boot the G5 to do the Thermal Calibration. The download came with plenty of info on how to perform the test and what to do if any errors come up.
I have a screen shot that I will post later of what the software interface looks like.
It took about 20 minutes to complete the calibration on each processor. The first calibration passed. The second processor gave me an error of Tafs is out of range and to replace the processor. I am running the calibration again hoping with better results. I am going to do some more research on this error and see what it means. So that is where I am at currently.
Updates to follow.
Last edited by photonorm; Mar 16, 2011 at 07:01 PM. Reason: adding text
|Mar 16, 2011, 07:05 PM||#10|
I'm surprised the 2.5.8 ASD even booted, let alone calibrated. The G5 Quad required the 2.6.3 disc
edit: N/M you have the dual 2.5ghz Must be confusing you with someone else...
|Mar 16, 2011, 08:37 PM||#11|
Second try on Thermal Calibration
Well, I guess it is time to clean up all the confetti and send the party band home.
I haven't researched what my 'Tafs is out of range' error means yet but after running 2 passes on the Thermal Calibration my machine booted fine but now the fans are moving very slowly and the idle temps are now around 150ºF for each processors with no apps running except the Temperature Monitor and the Activity Monitor.
The idle temps I posted in a previous entry were lower than the current idle temps because the fans were constantly running much higher than they are now. The farthest forward fan was running on high and the middle and rear fan groups were not as fast or working as fast.
I worked the processors at 90+% for 5 minutes and the temps. were up in the 190ºF range and even spiked over 200ºF for a second or two. At the higher workloads the the fans do spin up and slow down when the machine is idling.
I did notice that one processor was around 15ºF to 20ºF when the temps were up in the higher range. Part of me says that is normal since the coolant has to flow over one processor then the next before cycling through the cooling blades to release the heat. At idle the the processors are within 5ºF of each other.
I am uneasy about the higher temps not completely sure of what the correct range of temps are and when I should begin to worry that something isn't working correctly. If you know of a resource, a link, or even have those values I welcome your input.
---I AM WONDERING IF ANY OF THESE ISSUES MENTIONED IN THIS POST COULD HAVE ANYTHING TO DO WITH MY THERMAL PASTE APPLICATION. THAT MIGHT EXPLAIN THE HIGH TEMP DIFFERENCES BETWEEN PROCESSORS BUT THE COOLANT EXPLANATION FITS BETTER. JUST THINKING OUT LOUD ON THIS ONE-----
Maybe I am just being a nervous nelly and the current temp values are within tolerances or not. But between the error on one of the processors when I ran the Thermal Calibration and now the higher operating temps I am not ready to rest on my laurels just yet.
Its time to mount up and take to the Information Superhighway...UP...UP...AND AWAY!!
I went into Energy Saver Preferences and there is a pull down menu for Processor performance. There are 3 settings: Automatic, Highest, Reduced. I was at Highest. I changed it to Automatic and the temps immediately went down 20ºF and are now, at idle, 130ºF (+/- 3ºF). That is a step in the right direction.
When I put a heavy load on the processors for 5 min the temps hover around 185ºF on one and 200ºF (+/- 5ºF). When the load is reduced back to idle the temps return to the 130ºF range.
Last edited by photonorm; Mar 16, 2011 at 09:04 PM. Reason: new info
|Mar 16, 2011, 11:02 PM||#13|
What thermal paste did you use? like the online documentation? arctic silver 5? I hope you never slaggered the CPU die like they did. It was far too much! Too much is as bad as too little.
Also read up a little on your error...
Bright light. Bright light.
|Mar 16, 2011, 11:21 PM||#14|
I think those temps are pretty high. I folded with my quad for a bit and the temps stayed in the 150-160*F at the very most with all cores @ 100% I believe. 200*F seems very hot to me so be cautious.
You got all the air out of the lines right? Oh and running a thermal calibration is really only needed when switching CPUs/logic boards. Just changing the O-rings shouldn't have mattered but doing the TC is necessary I think if you are having problems like you mentioned.
|Mar 16, 2011, 11:25 PM||#15|
Somebody has similar temps he has with a dual 2.5.
I think these ones were air cooled?
It wouldn't let me calibrate mine as it said it wasn't needed.
Bright light. Bright light.
Last edited by vohdoun; Mar 16, 2011 at 11:33 PM.
|Mar 16, 2011, 11:30 PM||#16|
this is one of the best threads i have read trough all day
good job you are a smart man
... i actuallly had an idea that is simaler to your fixing plan (alibit with a quad g4 that isnt almst dead) and its good to hear it workd well
internet high five to you!
: 13in MBP, MDD FW800, iMac g3 slot loading, iBook g4
PC: Custom build(main), IBM Intellistation p4, IBM Thinkpad 390x, Dell latitude pIII, Headless Acer Netbook
|Mar 16, 2011, 11:39 PM||#17|
photonorm, what was the pump that came out of the leaky system. Was it a Delphi or Panasonic?
Delphi were the ones known for the leaky systems from what I've read over the years. I really should check mine to see what it is.
Bright light. Bright light.
|Mar 17, 2011, 12:46 AM||#18|
I have more...
So I did some research and didn't get my answers about the Tafs error when I did the Thermal Calibration.
I did read about the ambient room temp effecting the temp so I am cooling down the room to see what that does for me.
I also was going to run another Thermal Calibration but the software said I didn't have to. "Thermal Calibration is valid no calibration required." I read that someone did 3 tests and on the third test all errors on the earlier tests were gone.
I didn't want to press my luck since the fans do have variable speed at different processor loads.
So, now I just want a definitive answer to the normal operating temps and what the top end is so I don't fry my machine.
Thanks for reading!
|Mar 17, 2011, 11:18 AM||#20|
Thanks for all the kind words and posts. It is nice to see fellow tech-perts, self proclaimed geeks, DIYers, and just plain ol' problem solvers coming to the aid.
Here are my replies since my last post to comments from y'all:
Vohdoun: I used Arctic Silver 5 and put a about an 1/8" dot in the middle on the top of the processor. I read the online documentation for the thermal paste and followed their directions. Here is a cut and paste from their site found HERE:
Break-In Period Explained:
Due to the unique carrier fluid used and the shapes and sizes of the thermally conductive particles in Arctic Silver's thermal compounds it will take multiple thermal cycles to achieve maximum particle to particle thermal conduction and for the heatsink to metal cap interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink.) On systems measuring actual internal core temperatures via the CPU's internal diode, the measured temperature will often drop slightly to significantly over this "break-in" period. This break-in will occur during the normal use of the computer as long as the computer is turned off from time to time and the interface is allowed to cool.
Break-In Period by Thermal Compound:
Arctic Silver 5:
Break-in period: 200 hours (Break-in period will occur during normal use.) Temps will drop several degrees over the break-in period measured with a thermal diode in the hottest part of the CPU core.
I actually read most of those posts on the links you listed. My machine seems to be inline with what other listed in those posts. Maybe I am running within spec. It does bother me when the machine spikes over 200ºF. When it does it is only for a second (literally). Thanks for sending the links and please keep reading and submitting your idea, links, and comments. All of it will certainly help me and others down the road.
OrangeSVTguy: I did get all the air out, which was a trial and error process for me. Your recommendation to fill the LCS through the opening where the heat block goes wasn't the route I took but it was one I considered. When I replaced the O-rings I also put gasket sealant around the edge for a tighter seal.
Hansolo669: Thanks for the virtual high fives for my work. It is nice to hear and I am glad to know that you think it is one of the best threads regarding this type of repair. Please keep reading and posting
Vohdoun: The pump is a Delphi. It is my understanding that the Panasonic pumps were not used until after my Dual 2.5 was produced. I have read on other blogs that the Panasonic pumps were used in the 2.7 machines both dual and quad flavors.
(I didn't see your earlier post when I was posting late into the evening)
I need to get the photos up sooner rather than later since y'all are asking great questions. I bet the photos will help diagnose problems and certainly help others with their rebuilds.
Thanks to everyone for adding their comments to the collective
|Mar 17, 2011, 12:54 PM||#21|
Adding more photos for reference
I didn't realize until I started to add photos that I can only add 5 per post. So these photos are the photos of the new plumbing fixtures I incorporated into the LCS.
The first photo shows the compression valve fitting. Underneath the valve is a connector that allows the compression fitting to connect to the brass Tee fitting. On both sides of the Tee where the tubing is connected is where the brass barb fittings go. The white that is visible at each connection is Teflon Tape. All parts were found at various local hardware stores. I always like to ask questions of the sales folks to get their ideas on projects. They will sometimes know of a better item than the one I think I need. Case and point was one sales guys told me that compression fittings and brass fittings can't be connected together because the are different 'types' of fittings. I went to a second hardware store, shared my dilemma, and not only did he know that I could get a compression valve to connect to a brass fitting, he had the part! There is a more detailed description in the earlier post dated 3/16/11 titled "LCS refilled and tested for leaks."
Photos #2, #3, #4 show air inside the tubing. In photo #2 you can see very small bubbles. In photos #4 and #5 are shots of the small tubing coming out of the valve. The way I would bleed the air out of the system by having the LCS turned on its side just like it would live inside the computer and opening and closing the valve to release the air. By doing so would allow the bubbles to exit the system.
I had to experiment on how to get the air out. Here are some ideas that didn't work:
--don't try to suction out the bubbles. I would put the small tubing into my mouth and suck on it like a straw with the valve open then close the valve quickly after I couldn't suck any more out either because I would pull the coolant into my mouth or the suction was to great. While I was able to pull the coolant out with air bubbles the negative pressure was strong enough to pull air into the system. I figured this out after a couple of hours of using this technique. My best guess that the negative pressure was pulling air from the pump gasket seal or from a connection on the top of the system.--
--I also tried to use negative pressure from the coolant bottle. Remember that I put a small hole in the tip of the bottle so I could allow air to enter the bottle by moving a finger off the hole after the bottle was squeezed. I would squeeze the bottle, attach it to the tubing, open the valve, and allow the bottle to return back to normal, which would pull coolant and air from the LCS. Same issues as sucking with my mouth...pulls air into the system and we are trying to get air out.---
--I would run the pump with the valve closed. The pump would make a 'sizzling' sound as air moved through. I would cut power to the pump (by tripping the GFCI outlet) and in less than a second I would see a large bubble rise from the pump towards the valve. Before the bubble would reach the valve I would turn on the pump (by resetting the GFCI outlet) and open the valve immediately. The bubble would leave the LCS and come out the tube connected to the valve. See photo #4. While I was able to get the air out this method didn't seem to work. This method led me to the final approach that proved successful.---
I added a step to the description listed above of running the pump with the valve open so the LCS would equalize the pressure inside the system with the pressure outside. Once that was achieved I turned off the pump, closed the valve, and put coolant into tubing. I would then power on the pump, open the valve and the coolant would be pulled into the LCS. I repeated this adding of coolant until no more coolant was pulled into the system and some stayed in the tube that is connected to the valve. I would close the valve and turn off the pump still leaving coolant in the tubing.
When the pump shut off I would see a bubble rise from the pump towards the valve. With the pump still off I would then open the valve first then power on the pump. The air bubble would 'burb' out of the valve and be visible in the tube (See photo #2) I would tap on the tubing to get the bubble to rise through the tubing until it was released. As I would be tapping other air bubbles would come out on their own and I would tap those bubbles out as well.
Now that the system was equalized and coolant was still in the tubing I could see bubble leaving the system. As more air is taken out of the LCS the coolant in the tubing would be pulled into the LCS, replacing the air that was removed. When the system needed more coolant I would power off the system, close the valve, and add more coolant to the tube going to the valve.
I would repeat this method until I felt most if not all the air was removed. If I started with this method of bleeding out the air I think it would take at least an hour not including having the system run with the valve closed for 15-30 min at a time to allow smaller air bubbles to combine to make larger and to check for leaks.
Tilting the LCS in different angles while the pump was running helps to release air. I did the tilting toward the end of the bleeding off of the air.
Air bubbles would be visible in the clear tubing where the flow of coolant was going into the pump. They would just sit there and not move while the pump was running. Cycling the pump combined with tilting with the pump on and off helped to get those bubbles to move.
As the pump becomes much quieter as more and more air is removed from the system. In the final stages of the bleed process I would run the pump for longer stretches of an hour or two (with the valve closed). When I started a long run the pump may be 'sizzling' because of some air inside but when I returned the pump would be whisper quite. I would turn it off and leave it until morning and continue to bleed out more air.
I knew I was finished when I couldn't see any more air in the clear tubing and the pump wasn't making that sizzling sound. Since the LCS as been in the machine and under 'working' conditions of heat and running for hours and hours the pump is still quite and no visible bubbles seen in the clear tubing when inspected with a light shining on the tubing inside the computer.
Last edited by photonorm; Mar 17, 2011 at 01:00 PM.
|Mar 17, 2011, 04:52 PM||#22|
Screen shot of Temperature Monitor
Here is a screen shot of the readings from Temperature Monitor as Photoshop was processing files. You will see in the darkened area that the temp did spike up to around 210ºF/100ºC but was only there for a fraction of a second.
Click on the photo to see a larger version in a new window.
|Mar 21, 2011, 10:21 PM||#23|
Heavy Processor work
Here is a screen shot of my machine exporting files out of Adobe Lightroom. The files are camera RAW files (about 13 MB each) and they are having my file adjustments applied, then resized, and converted to JPEGS. There are 283 files in total.
As you will see from the screen shot with loads hovering around the 90+% range for 5 minutes (the time frame visible in the graph) the temps over around 180-190ºF for CPU A and 195-215ºF for CPU B. This is by far the heaviest load I have placed on my machine since the repair.
The ambient temp in the room at the time of the screen shot was 73ºF.
You will also notice at the top of the photo the Activity Monitor data how much "CPU" each Application is using.
The screen shot below is of iStat which is a widget. You will see the fan speeds, internal temps, and CPU usage. I am not clear as to where iStat gets it temp data from or how frequently it updates because the CPU and fan values change every second.
When you compare the values between Temperature Monitor and iStat the following values are equal:
CPU A Ambient
CPU B Ambient
It's only the temp values that don't read the same between the 2 apps. The manual does say that some PPC Mac are not supported but I suspect that if they were not supported they wouldn't show up in the widget.
This final screen is shot shows how quickly the temps dropped once the export was completed. The spikes represent me working within the app moving around the 'Ruler' that is over the graphical lines before I took the screen shot. The long stretch of lower temps is the machine in true idle and I took the screen shot in this post very quickly before the temp spiked.
The machine returned to the idle temps within a matter of a few seconds. To me that indicates the LCS is doing its job well by removing the heat off the processors efficiently.
So now the question becomes...which are the correct temp reading?
|Mar 22, 2011, 11:57 AM||#24|
New problem since repair
I did a restart for the first time since I rebooted the machine after I attempted to run the ASD Thermal Calibration (at least a week). Upon restart I got the chimes but no video. I checked for any leaks and didn't see any. The fans started slowly then began run speed up until they were at jet fighter levels.
I did notice a red LED light on the lower processor board, my name for the board that has the processor that attaches to the LCS and fits into the slot on the logic board. You can see it in the photo. The upper CPU board has no light on it.
The top of the computer is actually the left side of the photo.
I let the fans run at the high speed for 10 minutes then I did a force shut down by holding down the power button.
I tried a restart and do a PRAM zap (option-command-P-R). Got the second chime but same as before, red LED on lower processor card, fans slow then spun up. I did a force shut down again.
Does anyone know what that light means? I will be doing my own research after I post and will update if I find the answer.
I left the machine off for 2 hours while I was surfing the net looking for what the red LED meant. It seems to be related to temp but can't say for sure. After 2 hours I was able to boot the machine normally and there was no red LED as mentioned in this post. I will note that the lower Processor card is receiving coolant directly from the radiator and not from the other processor. So that leads me to think that maybe the processor card isn't seated well or there isn't enough thermal paste since it is receiving the coolest coolant and it still runs hotter than the other processor. Now I am assuming that the lower processor card is the hotter card only because of the red LED light and it's possible connection to a heat issue.
Still looking for a definitive answer to what the light means.
Last edited by photonorm; Mar 22, 2011 at 01:15 PM.
|Mar 22, 2011, 12:20 PM||#25|
Well, it certainly isn't good.. it can mean multiple things.. The 1st thing that comes to mind is maybe a faulty ram chip.. RAM after a while does go bad, depending on vendor. Memory from OWC has been very reliable and I use that for all my macs.
2nd: Red Light can also indicate pending or failed processor.. I read somewhere that because of a leak in the LCS, one of the processors in a dual 2.5(NOT QUAD) failed to post and thus a red light came on..
Those are the only two areas I can think about.. so here is what I think it might be:
Either... 1. FAILED RAM chip or slot
2. FAILED processor or pending processor due to a possible LCS Leak
3. A failed peripheral, such as the airport/bluetooth combo card, or
some other device.
But the likely cause: Failed memory or slot.. and or Processor.. oh, and I will throw in OVERHEATED processor.
Hope this helps.
|Thread Tools||Search this Thread|
|thread||Thread Starter||Forum||Replies||Last Post|
|My PowerMac G5 Dual 2.5 Liquid Cooled just went up in smoke! *UPDATED*||Crawn2003||PowerPC Macs||24||Nov 14, 2007 09:17 AM|
|New iMac 24" SR 2.4 vs PowerMac G5 Dual 2.5?||Revlimit Punk||PowerPC Macs||6||Aug 12, 2007 10:36 PM|
|FS: Powermac G5 Dual 2.5||50thVert||Marketplace Archive 2||17||Jun 9, 2006 08:23 PM|
|FEELER: Powermac G5 Dual 2.5||50thVert||Marketplace Archive 2||2||May 23, 2006 08:20 PM|
|FS: PowerMac G5 Dual 2.5 (Canada)||Scottyk9||Marketplace Archive 2||4||May 23, 2006 03:39 PM|
All times are GMT -5. The time now is 10:11 PM.