This pump is used for water cooling of processors and electronic components. Due to its size and output, different ranges of application are available. This pump is an electronically commuted ball-motor pump, with running times of more than 50.000 hours at 12 V. The only flexible part of a ball-motor pump is a sphere-formed rotor-/runner unit, which is propped up against an extremely hard, wear-resistant ceramics bearing ball. A conventional axle with journal bearing and shaft sealing is not available. The spherical bearing of the rotor-/runner unit on the ceramics bearing ball provides a lot of advantages: a development of the bearing play and thus a noise increase - is theoretically not possible. The pump is silent over the whole running time.
The bearing is self-adjusting. It is lubricated directly by the delivery medium (wet-running pump). There is basically no need for maintenance. Since the rotor is always held magnetically in the intended position, smaller dirt particles are no problem. Blocking of the pump is not possible under normal conditions. Even after lengthy downtimes a safe startup is ensured. The permanent magnetic rotor/ runner unit is propelled by a magnetic field, which is produced by the surrounding stator. This is built entirely around the rotor. The entire pump is thus with 38mm only a little higher than the stator itself. It easily fits into all conventional PC, Barebone and Mini PC cases. A separate magnetic shielding is not necessary under normal conditions. Theres no need for a separate magnetic shielding. The ball motor principle enables an economical enterprise with comparatively high performance. The tension variation can easily regulate the DDC direct current pump over a large capacity range. All medium-affecting parts are completely corrosion resistant.
This pump has a controller and is five-step adjustable with 1.800-4.800 revs./min. Mounting plate is included in delivery!
Here we go, how to bleed, clean and refill the system.
As the picture with the bukket on page 4 shows, Eaglefour used a aquarium pump for this. The aspiration tube in the bukket full with water and vinager and the pushing tube connect to the system, the last tube from the system into the bukket.
To refill, connect the aspiration tube to the bottle filled with demineralized water with 1% of the anticorosion product and the pushing tube connected to the Eheim valve (15mm inside). The non return valve is disassembled and the green tube connected to the valve, which brings back the liquid into the bottle. After a while, when there is no air coming out of the system, stop the pump, shut the Eheim valve, disconnect the green tube and reassemble the non return valve. Start the pump again, open the Eheim valve, and more complicated, keep the non return valve high but in an horizontal position above the bottle, use eventually a funnel , push on the ball from the non return valve to make the liquid come out. After that, close the valve and stop the pump again. Move the Waterblock upside down to verify if there is still water bubbles inside. This takes quite a while, but even after several purges and manipulations, some bubbles will remain. That's why Eaglefour used this piece of cristal tube with the Eheim valve filled with liquid. This derivation will be filled with the air bubbles and the liquid will come into the circuit. The derivaiton with the cristal tube will need to be refilled then.
I will later on give the references (pages and numbers) of the concerned pictures. I unfortunally lost fourty min of work when my Cube went down. Have to start again, and make a PDF so I can send it to you Meersbroeck otherwise this litterature without pictures is hard to understand. I had to read several times by watching the picture to realized what Eaglefour meant sometimes.
bb you've covered a lot of ground in your last few posts! I'm sure all that info will be of great benefit to many others, not simply me.
I was impressed with the pictures of a Quad Core 2.5 with its two pumps. Now if I had the money I would have got one of these rather then my old Dual 2700, but would I have had so much fun stripping it down and rebuilding it knowing how much capital I had tied up in it? possibly not
Anyway, thanks to you and Nikkonut I have managed to overcome my fear of working with 1: Complex Macs and 2: Computers & water in the same box ;D
The result is that my G5 is now happily purring away on my desk with only a 4 degree varience between processor 1 & 2. I used gravity and a holding tank (plastic tube into a pet bottle) to top up the radiator via the non return valve ball bearing.
So many many thanks to you both and by extension this excellent forum
I am happy for you, Meersroek. It was an interessting journey for me to discover so much. And you got it right with the gravity. Only a 4 degree varience between processor 1 & 2 is great, Eaglefour had a difference of 10 degree before fixing his LCS ;D. Hope others will benefit from the infos.
Hi All, just a quick update as the Beast has been running flat out for a couple of days now.
So far the G5 has handled everything I've chucked at it without any sign of slowing down or locking up. The temperature difference between both processors seems to be averaging out at 8 points under load, with about 5 at tickover. Unless I fit double pumps as Apple did with the Quad 2.5 I'll not get better then this as other members have poined out that the coolant flows across both processors before rejoining the rad; a simple design flaw that if everthing else is properly assembled should not have been a problem.
I'm quite confident now that the cause of the original problem was badly seated heat exchanger blocks coupled with badly applied thermal compound. With these two issues resolved my G5 is behaving like a totally different machine: the main difference being Reliability; without which the Beast was only a very expensive stack of spare parts, so many thanks Guys, Flea Bays not having her yet ;D ;D ;D
Lastly the G5 fan noise has dropped by around 50% so I suggest that a quick clean up and new compound might be the way to go if all the more usual quick fixes for this problem have been tried and failed ;D ;D ;D
Hi bb, that seems like a good idea; simple, cheap and effective!
I'd really like to go down the copper tubing route in the future though, rather like the French guy did in your previous post; I like the idea of wider piping allowing a stronger flow through the splitter and gaining a greater volume of coolant in the system as an added bonus.
My next project though is probably converting my MDD to a faster spec. and then adding liquid cooling.
Bear in mind here that I'm not an engineer, and my knowledge of hydraulics and flow control is based on refitting household appliances due to being too skint to afford a proper plumber
Last Edit: May 30, 2008 11:03:00 GMT by Meersbrook
I've just reread your earlier post about the water-cooled 2.5 in MacBidouille. My schoolboy French is pretty bad but I'm sure that I made a mistake in my response when I referred to "the new 2.5" wrongly assuming that Mac had released photo's of a prototype model. My only excuse is that I'm a foolish optimist when it comes to Mac stuff ;D
On re reading when a bit more awake, I now believe the article concerns a mod by a company called "Cooligy" am I right? if so the system seems to be a fair bit better designed then the Delphi system on my G5. What look like collector boxes should totally overcome the problem of preheated coolant being forced over processor B, and the long diagonal pipes could easily be tapped into for Memory, Bus or HD cooling.
Re. the valve connectors: I don't think that my small local plumbing merchants will have the parts but I'll try on Monday. On reflection the last time I saw anything similar was for high pressure 1 & 2 litre compressed air "buddy bottles" for divers; luckily there's a dive shop in town so I'll try there also early next week.
You are right about referring to Aquamacs articles on the MDD; before I do anything I intend to print them all off and stick them to my study walls ;D The hands on detail, plus knowing that the guy has learned from his mistakes then come up with working solutions give me the confidence to experiment myself. Maybe next year I'll get around to building a Hack ;D ;D