Die Simulator...

[Rami Saikali]

Humm...

Well i did look but i came up with a lot of confusing things....

I need to build a reliable Die Simulator that i can calculate c/w with...how would i go about building and calculating this...thats all i really need to know please.. thanks

[pHaestus]

http://www.hardforums.com/showthread...hreadid=266016

You will need a lot more gear than just a die simulator though. Better start with some questions:

What is your goal?
What precision is needed to achieve that goal?

Design and prioritize based upon the answers to these quesitons. A list of Bill's test equipment is here (along with some advice):

http://www.thermal-management-testin...em_testing.htm

[Rami Saikali]

Okay...

YEah i knew this would happen now im more confused then before on what to do

Starting a new site.. in near future we want to review waterblocks...

I have part of the procedure down.

The first part is we very methodically apply each block with the same clamping pressure.

We have a thermal probe affixed on the CPU touching the side of the die.

We turn on the computer and for 3 days have the ultility record the idle temperature.

Then we run F@h, SEti, Cpu Burn, And other programs to tax the cpu to full load for 7 days. the ulility records the temperature at 15 minute intervals again.

we take the recordings at the intervals (leaving out the first hour of each tests since the die might still just be warming up in that time) and then average to find a total Idle and Load temperature.

REpeat with every block.


(All this is done with a controlled room air temperature)


now we also want to find C/w so we show there arent as many variables.

I need something that i can build fairly cheaply but that willl serve my pourpose... I cant spend hundreds upon hundreds of dollars however i want to have something that will calculate c/w and do it wel enough to be reliable.

Thanks

[pHaestus]

I cant spend hundreds upon hundreds of dollars however i want to have something that will calculate c/w and do it wel enough to be reliable.

I want the same exact thing! No such critter though. Reliability implies accuracy, and that costs money. You need to measure both voltage and current to a pretty high precision and supply a nice smooth power. This alone will eat up a couple hundred dollars with secondhand equipment (all I use but gotta be careful).

Some observations about waterblock testing from another amateur at this task.

You MUST measure: Flow rate, water inlet temperature, die temperature.

Water inlet temperature must be held constant somehow.

You will probably want to measure: Water outlet temp, baseplate temperature.

You MUST get a reproducible mounting of block (10 reps seems to be pretty representative from Bill's testing). Mounting pressure is part of this problem, paste application and torque from hoses are others.

Look at the errors involved in EACH of your measurements and propagate them forward. With typical cheapo overclocker stuff, the errors in the temps alone is likely 3-4C. What about the error in W? 10%? Now calculate the error in the final C/W you get from those.

If the uncertainty in ONE of the measurements alone is more than the difference in performance of many waterblocks (and it is) then the results you produce are worthless. Well perhaps I am too harsh. They are not at all my definition of reliable. How's that for tact?

[Rami Saikali]

Well looks like i cant get to do any c/w calculations :/

Which also means probably no-one will want to send thier block in for review. OH well... The clamping pressure i have down to the number of threads through the hole it is and such but if i cant calculate c/w im betting i wont be able to do much..