Quote:
|
Pyrotechnic: I'm not sure if a leak detector would work 100%
|
You make a good point, and are correct about the problem of detecting a small leak. Presumably protecting against that is where the conformal coating discussion comes in. The other option which might be possible is to put some sort of plastic 'umbrella' under the bulk of the plumbing designed to deflect most leaks away from the board so the water would go to the case bottom.
Your idea of a seperate enclosure for the pump and rad is also good, although probably to awkward for most people (including me!)
However I may approach that method, as the case I'll be using is a dual chamber 'server cube'. I'm already planning to put the pump on the hard drive side of the case, and am debating about where to put the rad - I can either hang it off the back outside of the case on the drive side with the system exhaust fans blowing through it, or stick it on the inside on the mobo side with so the intake fans suck (or blow) through it. Either way, If the rad leaked it wouldn't drip on the mobo, unless the fans blow the water there.
I'm not going to even consider air cooling though it's safer in some respects. I don't like loud computers, and even on my ancient boxes with minimal fans, AC = NOISE... (I've also had multiple fan failures, each box is on at least it's second CPU fan)
Quote:
|
Solarian League: ... anyway i think the best way to prevant failure is to just do a regular mantaiance check say onice a week, you just pop open the case check the horse, check the comps to make sure they look like theyre functing correctly etc. or maybe extend that peroid to onice a month do a check up but then again if you bother the system more chance of failure so.... but if you do regularly check upon the system u could possiable avert an severe system failure
|
A bit more reasonable, (especially since you don't want the horse to escape
) but you do point out the dilema of excessive checking, not sure what the solution is. Of course with a Linux box you don't want to shut down so often so I need to design so as not to need frequent checks.
Quote:
|
bigben2k For an OCP (CPU Overheat Protection) alternative, someone suggested bolting a thermal switch to the CPU block: too hot and it shuts down the power supply. The switch is designed specifically for a set temperature, and I think it's available from www.digikey.com .
|
Simple and neat, but I'm mildly inclined against it. I don't know if those switches would be able to handle the power without getting into relays, etc. (Not bad but then you haven't gained much...) I'd also be concerned about what the trigger temp was, and how that would match what your system needs (dependent on the block properties and location of switch among other things)
If one is already using a digidoc, or other on board monitoring, and has a spare fan header then just hook a kill relay to it and tell the monitor system to turn on that 'fan' if temps get to high... Why add another temp switch?