DetroitAC's Load Tester Block

[DetroitAC]

Jinu117 wrote an excellent guide to an easy load tester, I don't want to hijack his thread, but thought I should share mine.

The idea here is to make a loadtester that has the same distribution of heat flux as an operating CPU. So the block that conducts the heat is the size of a CPU chip, an IHS is used to properly spread the heat out before the evap. This will be used for an evap calorimeter that I'm building for use in developing my evap designs, and testing other evaps (I have a Kayl that I'll be using as a benchmark, It "looks" like one of the best designs to me).

This was made in a CNC milling machine, although it was so simple, I don't think CNC functions were ever used, so it was essentially a manual milling machine. I had the same logic as Jinu117, all of the heat from the cartridge heater will flow to the evaporator if that's the only path. That being the case, I decided to use aluminum (cheaper and easier) and a long path to allow the heat flux to spread out evenly over the cpu chip sized column. I haven't finished the apparatus yet, this is only the heater, block, and an AMD IHS to simulate that effect.

I bought an Athlon CPU off ebay that had broken pins for cheap, and I machined off everything but the IHS to get it to use for this loadtester.

I'll be using a variac, but with a clamp current meter for AC amps, same multimeter for AC Volts. The cartridge heater is a 300W @120V size, allows plenty of room to hit most heatloads. I've worked with some test apparatus that are sized for exactly the conditions, and because of minor variations they cannot always achieve them. For instance if I used 200W @ 120V cartridge, and my house only has 115V I can achieve 192W

Next steps are to build a baseplate that will allow me to just clamp evap, IHS, block together. Then build an insulating box to surround everything.

[Jack]

Nice Can't wait to see some pics and results

[bazx]

DetroitAC very nice work

when will we see your cascade to test it?

[jinu117]

Awesome!!! Been thinking about it but the only thing I never sure was the effecitive heat transfer from cartridge up through the height. With perfect insulation, that might not be an issue... I have few IHS ting here to try it out but didn't get time to ehehe Show us result. Oh... if you can get me some delta read out as well as evap temp that would be awesome as I will know how effective it is almost right away Going to sticky this one

[DetroitAC]

I'll definitely be doing some fundamental studies like you're thinking when I get the evap calorimeter finished...performace with IHS vs performance without, check performance of an evap, mill off 0.05 inches and recheck, mill off another 0.05...

[bazx]

I'll definitely be doing some fundamental studies like you're thinking when I get the evap calorimeter finished...performace with IHS vs performance without, check performance of an evap, mill off 0.05 inches and recheck, mill off another 0.05...

sorry for this OT but your evap is looking good

[LittleDevil]

DetroitAC any updates?

regards

[johann]

Looks nice but one thing. Why is the cartridge so far away from the surface?

I think also the big mass of that block is not the best. Overall looks really good.

[DetroitAC]

LittleDevil, sorry no updates really, I'm waiting on some material.

Why is the cartridge so far away from the surface?

I think also the big mass of that block is not the best.

That's a good question Johann, basically I want the most distance I can get within the block so that the heat flux has a good chance to even out over the surface of the contact face. I think a CPU probably has a very uniform distribution of heat generation over it's surface, so I want to create a uniform distribution of heat flux (flow).

The problem is very similar to a fluid flow problem. If you want uniform fluid velocities over the cross section of a pipe, as long as it's turbulent flow, you need a long length of straight pipe to let the eddys entance effects dissipate.

For this block I will not be able to measure the heat flux distribution, but I'm using what I would call engineering judgement. A longer length, and more mass will not affect at all the amount of energy delivered to the contact face if all other surfaces are well insulated. If the energy has no other path by which to escape...it will flow down the only path.

I did a rough analysis in which I calculated the time required just to cool this block down, something like 61 seconds assuming the evap has a capacity of 200W, but during a pull down, the evap capacity is way higher, so I'd estimate 30 seconds to cool the block from ambient to -40C.

I figure a single evap test may take 1 hour of tweaking and stabilizing to achieve the conditions, so the mass of the block has no effect on it's function as a load tester

[jinu117]

That's a good question Johann, basically I want the most distance I can get within the block so that the heat flux has a good chance to even out over the surface of the contact face. I think a CPU probably has a very uniform distribution of heat generation over it's surface, so I want to create a uniform distribution of heat flux (flow).

Well.... guess what, heatflux is huge on cpu as well as not so uniform distribution of heat. Some parts gets used for certain operation some parts not (and they make it that way on purpose to reduce total heatoutput from CPU core itself). Yup... not the easiest thing to model for, hence gross approximation we are doing most of time. I am just worried that the distance might add another variable on your setup to balance out the system in reasonable time frame. Even my load block at times takes QUITE some time to stablize on temperature with load on typical single stage I make.

[johann]

Well reason I said that is that I have used exact resistors before on load blocks but the bigger the size of the block the bigger the heatload seems to be even after the block is cooled down to temp.

Perhaps they were not insulated very well, who knows.

I would make a design as small amount of metal as possible.

[DetroitAC]

Yeah, I see the point you are trying to make Johann. In my case, I will be able to perform a heat balance, as this will be used in my evap calorimeter.

Electrical power to the heater = (insulation losses) + (evaporator power)

If I find that I have large insulation losses, I will have to reexamine the setup.

Jinu, you've been holding out on me! Please, please, please! send me a link or a file of the data you have on CPU heat flux distribution! Wow, I had no idea you knew such a thing.

[pythagoras]

If you want to seriously explore this side, heres a few links:

http://www.irisa.fr/caps/projects/AT...x%20picture%22

http://www.enerdynesolutions.com/dow...x%20picture%22

http://www.lib.ncsu.edu/theses/avail...ricted/etd.pdf

http://www.intel.com/technology/itj/.../p02_intro.htm

Regards

John.

[wdrzal]

this is from NASA(18 pages pdf) they relese some basic software frome time to time but think you may have to pay for this one . NIST is also a good place to find thermodynamic software. again they release some for free but most you will pay.

In any case good reading ,for engineers.

http://www.pdas.com/tmx3572.pdf#sear...ling%20NASA%22