If geeks love it, we’re on it

Awesome HSF test rig

Awesome HSF test rig

Not all HSF tests are created equal, but some strive to be more equal than others. This is possibly the best test platform I’ve seen.

Comments

  1. lemonlime
    lemonlime Very well thought out testing methodology. I agree that a TEC will provide very predictable heat output and can be varied easily with voltage. The only thing really missing is an actual PC :) I know that when I read an HSF review, I want to see what sort of mounting and clearance issues will be found on a typical mainboard. Its also nice to be able to see that 'heatsink A' managed to keep a Q6600 below 50'C for example and provided cooling to the mosfets and northbridge area on 'mainboard x'. I think that accuracy is not always as important as consistency in this type of review. Its true that mainboard sensors are laughably inaccurate but consistency is certainly achievable with hardware test rigs. Variables like ambient temperature, interface material, mounting and configuration can easily be controlled. I also don't agree that AS5 is the best choice for HSF testing due to a small but somewhat significant performance improvement during the curing time. Each HSF would have to be tested for the same duration to keep that variable in check. There are other interface materials that do not require much break-in, or simply don't improve much over time. Arctic Ceramique is a good choice as performance improves only very slightly after break-in.

    Very cool setup though! I'm struggling through all of the same issues as I develop Icrontic's HSF testing methodology for some upcoming reviews.
  2. BuddyJ
    BuddyJ Yeah, I thought you'd find it interesting. Their choice of using AS5 also struck me as odd, too. My two issues with their setup are 1) it doesn't give any indicator as to clearance issue the HSF and mobo may have (which is such a big issue these days with the advent of MEGA size HSFs), and 2) it's a closed setup, yet no cases are that airtight. They'll get results that should ease in cross-comparisons, but they won't be real world numbers.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!