• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Peltier controlled by fan controller idea.

M

menlatin

Gawd
Joined
Sep 4, 2004
Messages
612
So i hate condensation. My 1st peltier project went south because of it and i haven't ventured back for 4 years. Well i'm upgrading my box, and before i've gotten all my parts, the bug has already bitten.

As the title reads, i want to use a fan controller to control my peltiers. Now, i know peliters draw WAAYY!!!!! to much current to just hook them up to the controller, so what about the a controller that is just On/Off and not speed control. You can wire that to a relay that is wired between your 12v and the peltier. Have a temp probe between the cpu and heat plate (still called that?) to the controller, set your range from like 18c-20c, and you should be condensation free right? What you guys think?

relay like this

http://www.radioshack.com/product.a..._name=CTLG_011_002_013_000&product_id=275-226
 
well, there are two key problems with what i am reading here. one is easy to fix, and the other is fundamental.

the easy to fix one is that a usefully powerfull pelt draws too much current for a computer PSU, which you seem to be indicating that you plat to use to run the pelt. the solution for that is to use a dedicated 12 V supply to run the pelt, keeping the other aspects of your current plan the same.

the one that is unfixable is that to keep your CPU colder than a good, modern, straight watercooling loop can keep it will mean that at least some parts of the pelt assembly will be going sub-zero, and will tend to form condensation.

that even leaves out the fact that you're simply not going to get a usefull performance improvement with a system that is within a few degrees C of being at ambient temperature, as compared to a watercooled system that is a few degrees above ambient.

seriously, if you have the budget to pelt cool, waterproof your PCBs, use insulation, and go at it full-force. otherwise, you're just pissing time and money away.
 
well, i already have a water cooling setup, just need the pelt, insulation, relay and fan controller. ~$50.

Now as far as the gains go, i'd say that most water cooled systems will do what 28c idle, and maybe 36-40c under full load? Thats what my p3's run at atleast. Well, with this setup we're talking about 18-20c under LOAD. Thats a solid 20c difference.

The issue that i'm more aiming towards solving is the no load situtation where a computer is left on 24/7, and the cpu can get really cold, for a really long time.
 
well, you don't have specifics in this thread or your sig, so i don't know for sure, but odds are pretty good that you need a new waterblock, unless you are already running with a DD maze 4 or a swiftech 5002, and a cold plate. odds are also good that you are going to need some added radiating capacity to keep good coolant temps, at reasonable noise levels. i run two BIX2 rads to keep coolant temps in the low to mid thirties with my loop, with what i consider to be reasonable noise levels in the low thirty dB range. i was able to keep coolant temps below 40 with my two panaflo U1A fans running at full, on one BIX2, however that's a bit loud for everyday use.

the CPU itself doesn't usually get too cold, when you are running overclocked and over volted. that's not where you're looking at the big danger of condensation. the biggest worry is always the cold plate. even in your planned configuration, to keep the processor down around 20 C, you're probably going to have to draw the cold plate down to about zero, for the low limit, and let it rise to around 10 for the high limit, you are certainly going to be looking at condensation, unless you are in an unusually dry room. the reason that i'm picking those numbers is that unless you are willing to cycle the pelt on and off so frequently that the realy will click over several times a minute, you are not going to be able to target temps much closer to your target load temp.
 
hhmm...... i see your point.... what about... a thin cold plate? say 1/16" or 3/32"? Less heat stored in the plate and more to the cpu sooner? I guess there's still going to be some amount of condensation though.... hhm.... We'll then. We'll see. Thanks for your quick responses.
 
menlatin said:
hhmm...... i see your point.... what about... a thin cold plate? say 1/16" or 3/32"? Less heat stored in the plate and more to the cpu sooner? I guess there's still going to be some amount of condensation though.... hhm.... We'll then. We'll see. Thanks for your quick responses.
Click to expand...


I don't understand what you aim to achieve here, but a coldplate is necessary to spead out heat from the CPU core. Making it thinner than it should be defeats that purpose, and I don't see it helping fight condensation at all, it would probably actually make it worse since the outer edges of the cold plate would be even colder.
 
You could use a PWM based fan controller and instead of going to a fan have it go to some heavy duty power transistors to switch the current for your pelt. Additionally, you can put a hefty capacitor between the +/- for your pelt and then the afftect will be more like a voltage scaling than a solid on 12v / off 0v. Something to think about.
 
Hm... well i see thay my mobo uses pwm to control the fans, so i cna use that. What about putting a cap on the input signal that way the transistor doesnt have to turn on an off a bunch, and then a cap on the output for the ramping effect. Now i just need to find the right trasistors and caps. Any suggestions?
 
menlatin said:
What about putting a cap on the input signal that way the transistor doesnt have to turn on an off a bunch,
Click to expand...


But that's how PWM works... :D seems to me like that would only shift the pulse by some amount of time, anyway.
 
You must log in or register to reply here.
Back
Top