First steps at testing.
 

I had plans of making a testing system slowly going, but last week a local retailer offered me to test a couple new coolers just arrived. Had to hurry up things and test with what was available.

The goal is being able to test complete "cooling solutions". That means heatsink/fan combos or water systems as a whole.

-Tests are made in a somewhat old AMD system with a 1900+ Palomino.
-The main equipment for acquiring temps is a CHY 503 K/J thermometer.
-One K thermocouple is bonded undercore on the cpu, another is held a couple cm. away from the air intake.
-BurnK7 is used for at least 30 minutes before taking temp readings.

After some test it was clear that the undercore sensor was excellent, it reacted quickly to load, and when stability is reached readings are absolutely stable on the 0.1°C resolution display.

The difficulties start with ambient readings: Temps show great variations as air currents inside the room bring waves of hot air from the heatsink outlet.

The CHY thermometer has an average function, so this problem was minimized by reading the average of a 5 minutes lapse. This worked very well on radiator type coolers, where air flows in a straight path from one side to the other. In that case the intake air is almost the same as overall room temp.

The real problem happens on traditional heatsinks. As air flows to the mainboard and has to turn to the sides, a lot of it is recirculated by the fan. On average intake readings were about 4°C higher than room! Even worst is that some areas of the fan draw more recirculating air, while others fresh room air, so small variations on Tc positioning have big impact on readings.

That why I’m seeking advice from the more experienced ones here of what can be an acceptable way of acquiring ambient temps.

Here are some pics of my setup:

Tough one.

You're on the right track, by running the HSF on an open motherboard, I think. I'd keep the area surrounding it clear (i.e. move the reader away).

Other than than, you could drape the work area (table?) to limit the hot air "waves".

It's not going to reproduce a mobo in a case, but you'll have steadier air input temps.

I favor using 3-4 temperature probes arranged around the intake area of the fan (use the same height above the fan every time and then take an average. Air temperatures SUCK to measure though and every heatsink is somewhat different. It's a lot easier to regulate flow rate in a wb than it is to regulate airflow in a test. The only GOOD way to do it would be to make an evironmental chamber and hold the air temperature (and mobo eh) constant

those are the choices
.

Dont worry Ben, the tc reader was there just for the picture.

The problem is that no matter how steady is room temp, outlet is going to be always over it. So by just walking arround in the room air circulation is altered and readings change.

In a tipical heatsink inteke and exhaust are so close together that theres alot of recirculation. I have the idea of ducting intake from a more steady area, but my concern is "ethical". When doing that I'm creating more ideal conditions for the heatsink than the tipical use. At how point isolating distracting variables become improving product flaws?

One of my initial ideas was testing things as delivered; that means using the included paste and sopport system, but this proved to be a big problem for tests:

1- Thermal pastes are the source of nightmares! each one has different set up time, and is not very practical waiting 48 hours after each mount to start testing.

2- Included supports can be dangerous! It is a good idea being able to report possible mounting issues, but discovering them in the test hardware can kill it! While mounting Thermalright SP7, one of the supports cracked an SMD resistor on my Epox 8RDA mobo that was too close to the hole. :eek:

Ever notice on OEM cases where the power supply intake is is right above the HS fan?

That's why it is there, to minimize recirculation. Good case flow is a common sense given.

Testing in open air will generally give you higher temps and without the exhaust air being drawn off you will tend to heat up local to cpu sockett components also.

Lacking a simulated die, I would place a 120/92 mm fan intake side 3-5 in. above the intake of the heat sink it will prevent recirculation to a great extent.

I would be careful not to get the added fan too close or run it at high rpms so the heat sink fan is not competing significantly with the added fan.

the effect is not as noticable with the delta Monster fan HS as they tend to dominate case flow all by themselves. smaller fans tend to show recirculation effects more

Could also do a fan cut out with a piece of card board extended 4-5 in. in a square or round flat plane. This would direct the air away from the HSF far enough not to be drawn back in and would not significantly effect HSF performance.

in hsf testing a flat plane of 3 fan diameters' radius is supported over the fan intake a specified distance (normally that which the case would have)
- there is always some recirculation

I normally use a forced flow environmental chamber with the hsf in free air

so do I occaisonally, When the contract req. it. But test block free air works just as well. @ some point you just end up splitting 'airs

"But test block free air works just as well."

correct in an environmental chamber, not so in an A/Ced room

Why not just close the thing in a sealed container and have the temperature reader and power external, running through a hole in the side that has been sealed? The purpose of testing is not to have good temperatures, just constant ones, is it not?

how is going to be constant in a sealed chamber? with a heat producing element inside, temps are going to be in constant grow!

correct but @ some point you just end up splitting 'airs = I get the pretty much same in my ansi calibrated flow chamber as in free air which is verified at oem test labs with less than 3% difference between facilities.

There will eventually be a point of equilibrium, just like everything else.