Watercooling Question?

[shaft103]

I am currently assembling my first watercooling system and there is one concern that I have. I leave my computer on 24/7 and I need to protect against pump failure. I was curious what the people who are watercooling do to safeguard themselves against this. My idea was to use motherboard monitor to monitor temps. and set it to shutdown the computer when it gets to a certain temp. I have tested the method and it does work but will it be able to save my tbird if I did have pump failure? Thanks for the help everyone!

Shaft

[ondaedg]

If Motherboard Monitor provides an autimatic shutdown of your system once a certain temperature is reached, then you shouldn't have a problem. The key is to make sure you have motherboard monitor running at all times. If the pump did fail, the cpu wouldn't burn up immediately. The waterblock would soak up the excess heat first. It would probably take ten to twenty seconds after the pump failed for it to damage the cpu.

[UnaClocker]

The great part is most motherboards have a thermistor in the socket, and it takes a helluvalot longer than 20 seconds for it to notice a CPU temperature change. In my tests, my system has locked hard long before the socket probe notices a difference, and if the system is locked, there's no way it can shut itself down. What you need is a circuit designed to monitor the block temp. Which nobody has made or designed yet. I'd say just get a Danner or Eheim and you won't have that worry.

[Memphis]

Try a twin pump system.

Cheers

Memphis

[shaft103]

Una,
Well to try and counteract the thermistor being in the socket, I have bent it up so that it it touching the back of the chip right underneath the core and I also applied a dab of thermal greese on it so that it detects the heat easier. My temperature readings change very quickly when I am idle and I turn on Seti or folding.

Shaft

[ck42]

Una, I've been thinking about this issue pretty seriously lately.

Possible solutions:

1. Electronic water flow meter/sensor. Only problem is that these things are EXPENSIVE. This would also ONLY take care of a pump failure or partial failure.

2. Use a probe on the cpu itself and somehow substitute it for the one built onto the m/b. This way you're not relying on the stupid probe in the socket. A flat probe on the top side of the cpu, like some of us are doing it, would seem as tho it would detect rising temps pretty fast. This probe being monitored by MBM would prolly work.

3. My prefered method is a standalone piece of hardware which monitors the cpu temp via a topside cpu temp probe. This way you don't have to worry about Windows locking up from RAPID cpu heating and therefore rendering MBM useless. This setup would have to do something like control a relay which would simply shutoff power to the computer...screw proper shutdown! When the cpu is about to nuke itself...

It would also seem to me, correct if I'm wrong here, that #3 would also take care of a pump failure AND a TEC failure. If the pump were to fail, the TEC would not be cooled...which would mean the cpu temp would rise...voila.

So, it all seems to come back to monitoring the actual cpu temp.

I've decided I'm going to make an attempt at finding a hardware solution, ie. #3.

I REALLY don't like the idea of having to shutoff my puter everytime I step away from it for 10 min.

I'm also thinking seriously about a standalone piece of hardware for detecting water in the event of a water leak. Only problem is that you need to place a sensor EVERYWHERE you think a leak might occur or the water would be. This solution would also have to do something like control a realy to shutoff computer power.

[warp_zero]

if you could get something that would monitor a temp and short a path if temp rises, just hook it in parrallel to the ATX power button leads, set windows to shut down immediatly when pressed