Physics - Comp Sci research FINALS

[MrConcrete04]

Hi, I wanted to ask the following: In school for my finals I need to do a project in my physics class which would consist of a 1200 word essay and some sort of research or data collection about something. I want to link it to computers cos I have a bunch of hardware I can test. I was thinking about checking how different TDPs of coolers affect the temps of a cpu and somehow link it to specific heat capacity, I have tower coolers, dual tower coolers, AIOs.

Thank you,

e.

[mariushm]

Sounds like an interesting project, though maybe a bit too complicated and time consuming for 1200 word essay.

 

To be correct, you would have to log the rpm of the fan or power the fan from a separate power supply so that it's at a fixed rpm (or several fixed rpms), you'd have to log ambient temperature, and other things. 

A lower TDP cooler could cool a cpu just fine, but the fan would scream trying to push as much air through the heatsink. A huge cooler could have weak performance because basically the motherboard will default to making the system silent and therefore spin the fan at lower rpm

Ideally you'd also have to monitor the VRM temperature and the power consumption of the cpu  (ideally you'd have a current probe on the 12v input to the CPU) because how well the vrm is cooled will affect how well the cpu is powered, and therefore how well it will reach high frequencies in some tasks ( ex may boost to 4-5 ghz for 10s or so until the cooler gets warm and then cpu throttles down, but if vrm is too badly cooled cpu may not high as high turbo frequencies)

Some coolers will blow air down onto the board through the heatsink, and some of that colder air will also flow down and cool the vrm heatsinks ... and that can help, as opposed to big tower coolers or water coolers. 

 

So.. I guess depends on how thorough you want to be. 

 

[jaslion]

21 minutes ago, mariushm said:

Sounds like an interesting project, though maybe a bit too complicated and time consuming for 1200 word essay.

 

To be correct, you would have to log the rpm of the fan or power the fan from a separate power supply so that it's at a fixed rpm (or several fixed rpms), you'd have to log ambient temperature, and other things. 

A lower TDP cooler could cool a cpu just fine, but the fan would scream trying to push as much air through the heatsink. A huge cooler could have weak performance because basically the motherboard will default to making the system silent and therefore spin the fan at lower rpm

Ideally you'd also have to monitor the VRM temperature and the power consumption of the cpu  (ideally you'd have a current probe on the 12v input to the CPU) because how well the vrm is cooled will affect how well the cpu is powered, and therefore how well it will reach high frequencies in some tasks ( ex may boost to 4-5 ghz for 10s or so until the cooler gets warm and then cpu throttles down, but if vrm is too badly cooled cpu may not high as high turbo frequencies)

Some coolers will blow air down onto the board through the heatsink, and some of that colder air will also flow down and cool the vrm heatsinks ... and that can help, as opposed to big tower coolers or water coolers. 

 

So.. I guess depends on how thorough you want to be. 

 

You'd also best use the same fan(s) for every test to eliminate as much variables as possible and then as you said rpm lock it (there are plenty of those 3 pin fan dial things to control fan speed).

[shadow_ray]

You could use a simplified model based on thermal resistance and explain the results through it.

- Resistance of the silicon chip and package (i think it's constant for each processor)

- Resistance of the thermal paste  (depends on thickness, area and material)

- Resistance of the cooler (depends on it's area, fan speed, material etc)

If you add these together you get the full thermal resistance.

 

After figuring out all of these parameters you should be able to calculate (approx.) the chip's internal temperature given the the power draw and ambient temp.

 

A more precise model would involve diff. equations and what not which is over my head..

[shadow_ray]

21 minutes ago, shadow_ray said:

A more precise model would involve diff. equations and what not which is over my head..

But it would allow you to put fancy weird looking equations (with lot's of Greek symbols) into your paper, making you look smart. And also to predict temp. changes over time.