Quad G5 Liquid Cooling System overhaul help

I was in contact with Laing or whatever their new company name is. They have the nowadays-standard confidentiality note in their e-mails, so I'm not reproducing the exchange here, but the upshot is that sourcing a G5 pump from ebay looks like one's best option now.

I too had a reply from Laing/Lowara today, which brought me to much the same conclusion, though it did not contain direct answers to my questions. I likewise received a lot of interesting info from bmaverick, who has a stock of old DDC pumps which he says have the correct PWM and “SINE control” to talk to the G5, but they do not have volume compensators. He provided another useful little nugget of info, that the VC pumps have vents under the ceramic ball which non-VC pumps lack, so there is no point in fitting a VC to a non-VC pump.

So yes, all in all it is looking like ebay. Frustrating, because I received the impression that people had successfully wired up earlier (but not OEM) Laing pumps to G5s. It is still not quite certain that the current generation will not play, but it seems doubtful, and a potentially expensive experiment.

Melted?!? Oh, my...I hope the pump is still alright.

If you did the connections correctly (polarity? +12 V line and not the +5 V), then a few things to consider: are you sure the ATX PSU was ok to begin with? Were the clip wires thick enough for this application? Most ATX PSUs would require some load on the 5 V lines as well. I connected an old useless hard drive to one of the other molex outlets from the PSU. The pump apparently does need a much larger startup current than what it needs under operation; maybe too weak PSUs (would probably have to be in the < 100 W range) wouldn't cut it. The PSU I had was 900 W, and good quality.

You could just drain and refill, but you must ensure that you get all of the Delphi stuff out before refilling with any other coolant. Air is very evident from the pump sound. Depending on how much air there is, the pump will wear out more quickly and the cooling performance of the whole system will be worse. When I tried bleeding air out, I let the pump run for more than 6 hours straight in hopes that the air would slowly gather to the top of the radiator and out of the main circulation. That didn't happen.

Btw. You might be interested to read the work log I posted of my rebuild/refill, linked in the thread I started on the subject. I'm still in the process of adding a couple of photos there.

Nick, could you post a link to a definite "case" of a, say MCP350, being connected to a G5 and actually working as a drop-in replacement (and not just powering up)? I was hoping to get something like a compatible PCB revision list from Laing, but didn't, and maybe they don't want to share such info.

If only. Yes, Laing (and Swiftech) seem very coy about giving out the information that they must surely have. Most of my conviction about the compatibility of MCP350 and DDC-1T (a generic Laing model number that covers a wide range of modifications), comes from this Apple thread:

https://discussions.apple.com/thread/1940728?start=30&tstart=0

Since the thread is archived, it is not easy to contact the posters for details, though I did manage to contact “Remoteneeded”. His LCS (with Laing DDC3v20 PCB, sticker 3.3) is still working 2 years on, with a speed control wire soldered on, but when I asked if he was sure that the speed was actually being controlled rather than the pump working permanently on 12v, he was not. The most that can be said for certain is that the pump does work. I am beginning to wonder if this applies to others, including this guy for instance:

https://discussions.apple.com/thread/2802360?start=0&tstart=0

There are various modding fora that are full of arcane information about PWM and Laing pump revisions, but mostly they are silent on Mac compatibility.

I would myself love to find a case of any non-OEM pump where the speed control can be confirmed to work. How in fact could one confirm that it was working without specialist equipment?

Well, if the motor speed control works, I'm sure the pump rpm will show change (in Temperature Monitor or such). The 18 W pumps are also noticeably louder at max vs. min speed. I think, you could connect just the pump and take out the other fans from the G5, start up, and wait through the OF delay to see whether it gets the pump going (should start out slow and go to full speed after the delay).

Testing outside the G5 whether some pump has PWM control should be possible by applying a constant voltage (at least 3.3 V should safe with new models) to the PWM pin, which should have it running max speed, and then

grounding the PWM pin, which should make it run at the lowest possible speed.

If the older Laings or MCP's (DDC-1, 2, 3.1, 3.2) indeed were drop-in replacements then it's a shame that nowadays everything on the market seems to be based on the new 1T models, and buying a Swiftech, you just don't know what's inside afaict.

A couple of reactions to your last Heikki: my understanding from Dr Bresink, dev of Hardware Monitor, is that the sensor technology used (only) in the G5 does not actually give reliable state readings for pump speed, just reports what speed the pump rpm *should* be, based on the temp.

Secondly, I was advised by bmaverick not to use an 18W pump in the G5 due to the fragility of the thin copper wafer water-blocks.

I am really tempted just to hook up my Laing pump with the v-mot wire on the blank pad and see what happens. I am having difficulty however in assessing the potential risk of damage to the pump or (much worse) the G5 if it goes wrong.

Nick, Dr. Bresink is probably correct about the rpm readings. I don't have the tach working at all and yet the pump rpm says 1250, which is the minimum on the Quad, but the pump is running full rpm. However, I can't get the pump rpm to react upwards when stressing the system - actually, it's hard to get any of the fans to go up from the minimum after rebuilding the LCS. ASD's pump test might be an alternative. At least, it fails when the tach is not working.

The Quads use 18 W pumps, Duals apparently the 10 W versions. I don't have one, but I'd assume it's more difficult to hear differences in the pump sound based on speed on the 10 W version.

A total guess, but I'd say the worst scenario is breaking the pump by overvoltage from the G5. Differences in signal types are probably not destructive and I don't think amperage could have relevant differences in a control signal. Be sure to report back when the smoke clears 🙂

Quote from previous page: "However, the Intel specs say that the max voltage from PWM headers should be around 5 V whereas at least the Quad gives out 12 V, which might be a problem."

How do you know the voltage from the G5 on the speed control line Heikki? Forgive my ignorance, but if there is 12V permanently available from the 12V wire, and if the Intel spec is limited to 5V (presumably for a 12V pump), does it not imply a completely different method of speed control if the Mac is delivering up to full voltage down it's speed control wire?

Dan, 12 V is always a yellow and 5 V always a red line in PC connectors. I reviewed your original post: are you sure the pump was ok to begin with?

Nick, I think we've hijacked Dan's thread here, but anyway, the 12 V wire from the G5 seems steady 12 V, always. The V-mot wire also measured around 12 V. I used a voltmeter to measure it with the G5 running without CPUs, or the pump. Yes, the intel spec would imply a different, I wouldn't say method, but at least voltage on the PWM, or in the Mac's case, potentially PWM (V-mot), wire.

Did bmaverick or someone else say anything about what the SINE on the pump meant? Would it mean that the PWM is of that type, or that the pump actually expects sinusoidal alternating current on the V-mot wire? Do other DDC pumps have such marking? There was a mention of sine control or something regarding a sanyo pcb replacement for the Laing pumps but I've now lost it somewhere..

Yes, you are right Heikki. (Sorry Dan.)

Heikki, could you email me: my first name [at] curloadfarm [dot] co [dot] uk?

I don't want to clog up this thread with a lot of extraneous matter, nor do I want to quote bmaverick extensively without permission.

QUOTING: BDAqua "I think you can use regulat Anti-Freeze/water mixture."

Nope. The regular EG coolant will have issues with the residue of the old harsh saline coolant that was originally used.

Your only hope is Sierra Propylene Glycol that has the required corrosion inhibitors and using Distilled Water.

Being that the LCS is a mixed metal loop, the ratio of the coolant is 66%PG and 33%DW. Or 2/3rd to 1/3rd mixture.