Hi-Z fan testing follies: the SI-128SE [almost!] fiasco

Felger Carbon · Post by **Felger Carbon** » Tue Jun 17, 2008 9:50 am

Toward the end of this thread, there's some photos showing the detailed construction of my hi-Z equal-airflow test jig based on the SI-128SE. Bright and early this morning, I fired it up. It crashed and burned on the runway (old aviation expression referring to really bad aircraft designs).

edit: see added comment at the end of this post, and changed subject.

Applying 12.7V to the Noctua fan, the airflow sensor prop barely rotated at all with the 9-blade Noctua turning 1401RPM! I looked for the obstruction I had overlooked. Nada. The jig, in this example, is the SI-128SE's cooling-fin block and the fan mounted directly on it. I took it apart anyway and confirmed that there was no unexpected impediment, just the closely spaced fins.

Substituted a Slipstream800. At 12.7V, not enough air came out of the rear of the cooling block to move the sensor prop at all.

So, I used my Mk I airflow sensor - my right hand - to try to make sense of what was happening. There was airflow out the back of the cooling block. Very little airflow. At the front of the fan, my hand detected a lot of swirling air (due to the rotating fan blades), but this masked the little movement of air into the fan (and thru the cooling-fin block).

The SE-128SE has a buncha little holes in the fins that the original SE-128 didn't. My hand/fingers could detect air coming out of these little holes, and in fact from every tiny gap etc. But not much, and not much at all from the back of the fin stack. Boy, I hope nobody is counting on all that super VRM cooling from the downflow air out of an SI-128SE!

I chose the 128SE as the basis for my Hi-Z fixture because I thought it would be, well, Hi-Z! A pretty good brick wall was not what I was expecting.

Two planned Z fixtures to go: a Mid-Z based on the S-1283 HSF's cooling-fin block, and another using fan adaptors with different input areas to simulate exhaust fan air impedances. I'm pretty sure the last one will work as expected; there's a whole lot of fan adaptors and if the airflow was blocked I'm sure we would all know about it by now!
-----------------------------------

You can do a crude version of this test non-destructively on your 128SE, or TRUE, or any HSF with closely spaced fins. Just pick up the HSF, mount a fan, connect the fan to 12V, and see how much air flows out of the other side of the cooling-fin block! No computer (or CPU, or mounting) needed as long as you can get 12V someplace.
-----------------------------------

edit: This design attempt was not a total fiasco. I simply failed to use adequately high-pressure fans. There's lotsa 2000RPM fans around, and they would work great on this HS and test fixture. But those fans are noisy, and are of no interest to me - and I'm running this particular show.

Later the same day, I discovered the identical, unmodified fixture base works fine with ~1200RPM fans (or maybe less) using the HDT-S1283 HS, which has slightly wider fin spacing.

Modo · Post by **Modo** » Tue Jun 17, 2008 10:08 am

Have you tried fans that make more static pressure, like the S-FLEX, which supposedly does?

Bluefront · Post by **Bluefront** » Tue Jun 17, 2008 11:53 am

Yep....back pressure. It's a killer. Now if you set up the machine with the fan sucking through the fins, I'll bet your sensor prop will turn normally. Keep at it.....you'll get it right.

Felger Carbon · Post by **Felger Carbon** » Tue Jun 17, 2008 2:34 pm

Bluefront wrote:Yep....back pressure. It's a killer. Now if you set up the machine with the fan sucking through the fins, I'll bet your sensor prop will turn normally. Keep at it.....you'll get it right.

We do agree most of the time, Bluefront...

The HSF Mid-Z fixture works fine. Took zero time to build it! Same as the Si-128SE fixture.

The low (but steady and repeatable) sensor RPM means I have to take my tach readings at night to get the needed contrast. And since summer is coming, that means windows are open at night, so noise measurements have to be taken in the daytime. Recording the fan voltage to 10mv resolution makes this transition possible.

Cistron · Post by **Cistron** » Tue Jun 17, 2008 11:04 pm

Does this mean that high-Z HS would cool equally well if they were only half as wide?

Modo · Post by **Modo** » Tue Jun 17, 2008 11:45 pm

Only with low-CFM fans. Look at TRUE versus HR-01 Plus comparisonâ€”the first one is better with higher rpms, while the latter excels with low rpms. This means the added airflow is required for the whole fin area to be used.

Tzupy · Post by **Tzupy** » Thu Jun 19, 2008 4:05 am

My setup, 19 months old, says that the SI-128 (the old one) isn't as restrictive as it seems from Felger Carbon's test.
With a BIOS-controlled Nexus (now at ~530 rpm), the temperature of the northbridge is about 47C.
If the Nexus fan wouldn't be turning at all (the SI-128 is dusty now, my bad), the northbridge would be ~6C hotter.
So if even with a 530 rpm Nexus there is some airflow going through, at full speed it should be significantly more.