Quote:
|
Originally Posted by Ls7corvete
It is a 3000+ winchester running at 280*9 1.56v(DMM tested). I have not tested stock settings with this setup, Idle will get too low and I have to worry about condensation. Those are not my CPUs max, but close enough for testing purposes. I should have put this in the article but I think the comparison to my old setup is more meaningful.
I will try BurnK7 but for now I am happy with a 42c prime temp(overclocked). Yes there is one fan on my HC. I am using a chevette with no shroud, I should be able to put my bonneville(dual 120 ) core in and see an improvement.
LHG, I believe that a fluid with a higher thermal expansion should give better flow rates and possibly better temps than water, water has an unusually low thermal expansion. I am beting over 100$ on this.....
A couple degrees from improved HC and a couple from a better coolant and I can call it a storm killer. heh.
EDIT: Burnk7 shows 48c, makes it look rather shitty like that. Lol.
|
Yeah, that's what I was concerned about. The TEC's will do a decent job, undervolted, so long as you don't exceed its ability to move the heat load. This was the whole reasoning behind a
TEC water-chiller I've been working on.
TEC's are notorious for providing good cooling up until you get close to their heat-pumping capacity, and as soon as you exceed that, things start going awry, which is why I suggested the BurnK7 deal. The Winchester CPU's run fairly cool (in my experience anyway until you start pushing >1.6v through them), and with your moderate overclock/overvolt, running BurnK7 would be pushing out about an extra 5-10W of CPU heat over something like Prime95, and as you noticed, it results in a rather non-linear increase in CPU temperature.
This is where things would start to change dramatically if you instead stuck a San Diego core onto your setup and started dumping >100W of heat, instead of your current 60-70W. Whereas with straight water you would get a rather linear increase in CPU temperatures with respect to heat load, on a TEC you will exceed its heat-pumping capacity and CPU temperatures will sky-rocket in a non-linear fashion.
What I would suggest to you is that you have constructed a very finely balanced passive TEC heat-pipe, where so long as the CPU heat load remains within an effective operational range and the TEC's COP remains high enough (undervolted), the passive cooling nature of your loop is not being overwhelmed, and so long as the CPU heat load is high enough, you will remain above the dew-point where condensation occurs. If the CPU heat load gets too high, thermal runaway will occur and the CPU will toast.
It is a very effective, as as you demonstrated cheap, way to keep a CPU that is operating within a rather narrow heat-load range about as cool as water-cooling could maintain, but it has it limitations in that it's not going to satisfy a broad range of heat loads. It is a very delicate balancing act you've managed to achieve with your setup and I'm glad you're happy with it, but it will not provide the same broad heat-load application range that a basic water-cooling setup, or even a decent HSF, can provide.
I'd be curious to see what your results were in comparison to a decent HSF with a quiet fan as well, especially in terms of peak achievable sustained BurnK7 overclock. I would suspect you will find that there will not be too great of a difference.