Dew point anti condensation tips

Spudley

Spudley

Spudley

Associate
Joined
3 Oct 2014
Posts
184
Location
Taunton, Somerset
Ok tinkerers............my question is, what nuggets of wisdom are out there for combating this issue.

I've cobbled together the basic guts of a water cooler + chiller system that will drop a cpu cold plate down to sub zero temps (gotta love Generator glycol based coolant).
the problem is, I dont really want to slap this onto £1500 of new mobo/cpu/ram guts for a complete rebuild when it can suck this kind of moisture out the ambient air to dribble & freeze all over it.
Temp reported @ time of the photo at the copper plate was -0.1 deg C
Any ideas?
;)

kzjJ1VK.jpg

bWdSvkA.jpg
 
Last edited:
What are you using to get it so cold?
Ie is it adjustable?
Bringing it up a few degrees might help

Covering everything in a conformant coating
Is probably what people use against condensation

And you have nothing attached to the cold plate
Is it actually going to get that cold
Once a cpu is providing heat to it?
 
  • Like
Reactions: cho
Tailoring your thermal removal to your system thermal output will go a long way to managing it, but there's nothing you can do to straight up eliminate it.
Insulating your tubing and block itself will help, drowning your board and socket etc in something to seal it, all the standard stuff.

I assume you're not planning on trying to run it as a 24/7 setup? If so, an upside-down horizontal mount would probably be your best bet but wouldn't guarantee no condensation on the upward facing part of the mobo if the socket is getting too cold.
 
Definitely threads in the forums somewhere
With people using phase change
Or water chillers
Just don't remember who it was

Could cover the block and pipes in foam
And cover everything near the cpu socket in whatever
People doing liquid nitrogen use

I would be doing any testing on an old cheap pc
Before trying it on £1500 worth of stuff

More information would help
What are you trying to achieve?

Wouldn't aiming for 5c to 10c coolant temperature
Work and eliminate the condensation risk?

Plus as I said earlier there's nothing putting heat into it
With a cpu and presumably a gpu attached to the loop
Will it still be as cold?

If its immediately that cold some sort of delay switch?
To allow the cpu and gpu to dump some heat in there first?

Some boards/cpus may even not post at 0c?
Never tried this stuff but seem to recall some cold bug?
Think my board might even have a ln2 setting for this

Pretty sure one of the guys in here has his radiators outside
So gets pretty cold but not enough for condensation
Even in winter
Which is why am wondering if you actually need it at 0c
 
With a load (an idle CPU), the whole thing won't get so cold. FYI I did something like this with a peltier cooler back in the day.
 
Shame peltiers are so inefficient
And the other side gets so hot
Guess that's why only been a few attempts at
Peltier heatsink from big companies

If the op ever answered it would help lol
But yeah I did mention there's no load/heat input in his experiment

Wonder if you attached that to a couple of external radiators
How well that would work
Would probably remove the condensation risk
Or move it to outside the pc at least
 
Conformal Coat as much as you can, dielectric thermal compound - throw it about like you're icing a cake including in and around the socket, sticky back foam with spray adhesive and more foam on the back and round the socket area. Heat pads where you don't want things too cold can be useful, just want to keep it above the dew point, no need to make it hot.

Probably don't need all that with only a 0 deg C unloaded. As soon as you apply a decent load that should be running above dew point I'd imagine.

Never really found huge benefit in water chillers, always presented more issues vs performance improvement. Saying that my VapoChill loved to munch on AMD cores, chipped edges and clean cracked a couple with too much clamping force applied. Made all my gear unsaleable afterwards anyway so you get carried away and start probing boards to add more voltage here and there, end up with soldering jumpers left right and centre and things just go poof and no more. Becomes an expensive hobby buying chips by the tray to test out their clocking abilities.
 
Agree with what's been posted so far.

However, this test you have setup doesn't have any heat applied to it yet. It'll run significantly warmer once setup with a CPU at load.

Do you have a way to control the temperature of the chiller? I'd run it a lot warmer, as close as you can get to due point, then crank it when benching.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom