Quad G5 Liquid Cooling System overhaul help

On the tubing, any of these help?

http://www.frozencpu.com/cat/l2/g30/c99/list/p1/Liquid_Cooling-Tubing.html?id=wT if2bJz

http://www.frozencpu.com/cat/l2/g30/c457/list/p1/Liquid_Cooling-Tubing_-_Tygon.h tml?id=wTif2bJz

http://www.frozencpu.com/cat/l2/g30/c289/list/p1/Liquid_Cooling-Tubing_Accessori es.html?id=wTif2bJz

I'll try to take some pictures of the pump connection. So far, I've only used the pump while cleaning the radiator, and now in retrospect, I think a better way to do that would have been to connect a faucet, using a garden hose connector, straight to the rad. I didn't have the proper connectors for a faucet connection, though.

Originally, I was hoping that I wouldn't have to replace the tubing and that cleaning would be accomplished simply by running hot water (or vinegar+water) through one open connection (with inlet/outlet in a bucket) in the loop with the pump. Didn't happen.

Pump powered using alligator clips and a molex from an old cpu fan:

The green thing is there to protect the tachometer pin from accidents.

It's a temporary connection, so, it's not supposed to look nice. Let's see if I could explain it.

First, the pin numbering comes from the pump connector. It has numbers written on it. Not all the pins are physically there, though. I guessed the pins based on the Dual G5 pinouts posted here and elsewhere.

In the first photo, there's the pump connector, with red alligator clips attached to pins 1 and 5, and black alligator clip attached to pin 7.

In the third photo you see the other ends of the two red clips and the black clip. There, the red clips are both attached to a yellow wire (+12V) from a male molex (the lower molex of the photo). The black clip is attached to the black wire (GND) from the same male molex. The male molex in question is the one from the old cpu fan, which just means that it was ok to cut the fan out to produce a male molex with open leads. You could source a suitable male molex from anywhere, or try to stick the alligator clips straight to the psu female molex.

The male molex in the picture is connected to a standard psu female molex, which has its wires coming from the top of picture (from the psu, of course).

The other leads from the male molex, extending to the bottom of the picture go to a pc motherboard fan connector, which was still attached to the molex. The fan had dual power connectors. That's all irrelevant, of course.

I didn't take a picture of the hack to allow using the psu outside of the computer because this was actually an old Nokia server psu, which has its very own set of weird connectors. However, the ATX equivalent is documented in many places (short green+black wires iirc).

Thanks for all that Heikki. I wonder if you (or anyhone else reading this) could throw any light on the problem I have of replacing a dead (Delphi/Laing) pump with a new Laing DDC-1T.

Earlier Laing pumps had a PCB on which (supposedly) you could remove a wire bridge between two terminals and then use all four terminals to connect to the original four wire Mac connector, but at some stage Laing changed to a new PCB (marked DDC3.1-PWM / 075-162) which has no bridge and switched-around connections, including one blank.

The connections for the existing three wires are straightforward: yellow = 12v, black = ground, blue = tacho. It is tempting to assume that the fourth Mac wire, for variable speed control, could simply be connected to the blank connector on the pump PCB, but there are various things I need to understand before trying this.

1) In the original G5 single Delphi pump setup (2.5 Dual, June 2004), what did the fourth (“V mot”/variable speed, pin4) wire actually do? Did it involve PWM (pulse-width modulation) or did it supply a variable voltage?

2) Is PWM speed control something that the Delphi/Laing pumps have always supported, or is there something new about the latest generation of PCB with “PWM” included as part of the designation?

3) If the blank terminal on the latest PCBs is intended for a PWM control wire (and its use for this in a Swiftech adaptation, the MCP35X seems to confirm this), is there any reason to doubt that the G5 could control it via the fourth wire?

Nick, I'm basically having the same problem, at least, soon. I managed to damage the pump's tach output somehow and now it runs full rpm all the time. I tried contacting various Laing support addresses but couldn't find an e-mail address that actually works anymore (after their ITT merger).

First, are you sure you can't transfer the PCB over from the old pump? There are also E-bay sellers with a stock of pumps pulled from G5s.

I would guess the Laing PCBs haven't changed that much over time. The Quads apparently used various revisions of the Laing DDC pumps, or at least, I've seen pictures of Quad pumps that have different stickers than the imprint on mine. They are probably all compatible or easily modified. I have a faint recollection that the connectors on the PCB were in different order in different revisions however.

In the quad, the pump runs at full rpm without the tach, so, I think if slightly increased noise isn't an issue, it doesn't really matter if the G5 can control the pump or not. Not having the tach appears to cause the firmware to supply the pump with what I'd guess is voltage enough for 3600 rpm operation. The pump's 50000 hour mtbf should be for full rpm operation and should be more than enough time to outlast the G5. It probably doesn't matter if the V-mot wire from the G5 isn't connected at all.

Like you, I would like an official word on the pump connection, however.

At anandtech forums and others, there's a guy "bmaverick" who might have an answer.

Thanks again Heikki.

Good tip about ebay; I wish I had seen that before investing in a new Laing.

I might be able to swap over PCBs, but it would involve more soldering than I am comfortable with, and really raises as many questions as connecting to the exiasting pcb, since the motor configuration looks quite different. And I don't even know if my old PCB is still good.

I have asked questions elsewhere, including attempts to contact Laing, but I was really asking here to try to find out more on the Mac control end of things. Do you know the answer to my question 1), about PWM?

Nick, I'm sorry I have no facts about the PWM control. It would seem likely that the "V-mot" control from the Mac is PWM because (1) that's what the Laing pumps seem to have if they advertise speed control and (2) if the G5 pump was voltage controlled, why have two voltage lines when one would be sufficient? Then again, the Quad pump didn't even start without applying 12V to the V-mot...

A data point (see the comments): http://martinsliquidlab.org/2011/03/09/laing-ddc-1-ddc-1t/

Some further thoughts: As it shouldn't matter if fixed voltage is applied to PWM input, it should be safe to test. The empty tab in the Laing -PWM PCB apparently is a PWM input:

http://www.pcmasters.de/forum/wasserkuehlung/87558-howto-laing-ddc-1t-auf-pwm-um bauen.html

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. If no conclusive answers to this question appear, I'll just borrow a scope. Don''t want to spend 80 € on a useless pump.

Yes Heikki, there is an increasing amount of evidence that the new PCB Laing pumps (LAING DDC3.1PWM) do support PWM, and that that is what the empty terminal is for. The crucial question for us now is just what comes down that G5 V-mot wire, and is it the right flavour of PWM control?

One of the things that worries me is that the PCB on the original Mac pump seems quite unlike the standard Laing DDC-1 PCBs, and that on the side of the pump casing it says “Type: DDC-1VC”. DDC-1 makes sense, but what about VC: voltage control? It also says “SINE”; why?

Someone must know the answer to these questions, but until they are answered I am tempted just to hook up the pump with three wires, or go for a secondhand one on ebay.

The VC part I know. It comes from "Volume Compensator" and means the extension chamber at the bottom of the pump that is not present in normal DDC models.

Of course; thanks Heikki.

You might be interested in the comments on this page

http://martinsliquidlab.org/2011/02/25/swiftech-mcp-35x-reservoir/

It seems as though there are uncertainties from both ends again. Do you, or does anyone, know how much of a problem it might be just to run the pump with three wires (no V-mot)? Does this cause the system to run all the fans at full speed as well as the pump, or otherwise produce difficulties?

Interesting link, Nick. Too bad they couldn't confirm anything either.

A thought about the DDC PWM control: Maybe all the Laing pumps have PWM control (assuming that's what the G5s and old Laing models use), but they normally don't work with the 3.3 V or 5 V voltage supplied by PC motherboards. The G5 gives out 12 V from the motor control line (as measured by a cheapish voltage meter). Maybe the MCP35X is just modified to handle the lower voltage and that's the reason it works with "common" PWM controllers while the others don't.

If the Dual is like the Quad, it doesn't matter whether the motor control (or the tach) are connected or not, the 12 V line works anyway. So, you should be able to run the pump. Although, I'm not sure what happens when the tach is connected but the motor control isn't. Other fans are not affected by the pump. ASD will fail, but that seems to be about it for adverse effects.