Fluctuating Idle Temperatures

[BonzaiDuck]

I had posted some questions and "food-for-thought" regarding what I thought were anomalous but promising results with my SI-120 and a Prescott 3.2E (478pin) with motherboard ducting.

Some people think "I know some things," others may have doubts -- but asking questions is my stock and trade -- here or anywhere.

My rig seemed to get better and better as the Arctic Silver 5 cured. I was able to tune down the fans in the case considerably and yet maintain the same cooling profile. Even at 80F -- a Baghdad spring morning -- S&M, Prime95 and any other load test forces the CPU temperature sensor to peg at only 43C. And this Prescott is over-clocked to 3.5 Ghz with the FSB running at 1,000 Mhz.

What I notice is that the lower the room temperature, the wider the swing in idle values shown by SpeedFan. The swings occur at fairly regular intervals of about a second or less. So at one point, the idle CPU value may be 77F, then 81F, then 84F, then 77F -- even dropping to 74F -- back and forth. Today's room temperature is 65F.

When the CPU is loaded up, the swing in CPU temperature only fluctuates +/- 3F degrees around a norm or mean.

I'm wondering if the idle temperature behavior is resulting from some harmonic action of the heatpipe cooler -- that is, the condensation and vaporization cycle is not continuous, but harmonic like a pendulum.

Does anyone have any thoughts about this?

 

[GalvanizedYankee]

Welcome back Mr.Duck Long time, no see.

Being that you put a very fine esoteric point on your discussion it is with some trepidation
that I respond to your query

Cutting right to the chase. At higher fan speeds there will be a slight vibration that may
assist the capillary action within the heat pipes. This slight vib *might* help the near
microscopic globs of liquid cooling medium amass much faster and be drawn back to the base
at a higher rate of speed.

Wicking as well as liquid inside heat pipes varies a great deal and is determined by application.

http://www.heatsink-guide.com/ > heat pipes (on the left) Read it. It's dated but good info.
In the second to last sentance is a link. Read this also, as it will put a real fine point on
heat pipe tech.
You might want to check-out the homesite of that link for the scientific info that is way over
my uneducated head.
Tell me what you think, as I'm sure you will, of the offered opinion and links.


...Galvanized

 

[GalvanizedYankee]

I would bet a that what you are experiancing is the lower temperature working
range of the fluid in the heat pipe.

That thought just hit me this morning.


...Galvanized

 

[BonzaiDuck]

<<One problem stays: A heat pipe designed for a certain temperature range will only work well in this range; designing a universal heat pipe for all temperature ranges is not possible. Apart from the vapour temperature range, factors like thermal stability and thermal conductivity influence the choice of working fluid.>>

I would tentatively bet that this explains at least part of the behavior for the SI-120 working with the Prescott 3.2E.

Given the behavior of the temperature values at load as opposed to idle, I rather doubt that it has anything to do with the reliability of the temperature sensor.

I'm also wondering if the implications here are less CPU stability at lower room temperatures [verging on "uncomfortably cool or cold"] -- than at the normally accepted range of comfortable room temperatures.

it might be interesting for me (or anyone else) to see what happens if the fan speed is lowered sufficiently to raise the CPU idle value toward 38C, so that SpeedFan kicks the fan up at 40 to 43C. If the fluctuation is noticeably smaller, it would be an encouraging confirmation that the variation occurs at the lower limit of coolant effectiveness.

We'll give it a look-see, for shuah!

 

[BonzaiDuck]

I think we're correct about this, and thanks, GY, for drawing my attention to that link on heatpipes.

I kicked the idle fan speed down to around 1,300 rpm. The MEAN idle temperature moves up about 3 or 4F to around 95F, and the fluctuation is reduced by about 2F at the higher value. The maximum fan speed is around 2,800 to 3,000, and kicks in at about 102F. With SpeedFan and Prime95 LargeFFT torture test, the mean value at load is around 102.5F.

So I have actually reduced my idle-to-load spread from 10F to around 8F. Thus, letting the CPU idle at a slightly higher temperature, with the fan set to kick on close to the expected load temperature value, has made for an overall more stable temperature. One would expect that this would also lead to a more stable CPU performance, although this might not be so easily detectable without pushing the CPU, FSB and VCORE to the extreme limits. And I'm comfortable with the existing OC settings.

System is even quieter than before, too.

 

[BonzaiDuck]

Well -- it's a tradeoff between a closer range of CPU temperatures and a wider range of chipset temperatures. I think I like keeping the chipset at a more constant temperature and a 5 or 6F lower.

Interesting, nevertheless.