Unexpected results for hard drive heatsink/suspension

echoes · Post by **echoes** » Mon Jan 31, 2005 9:08 pm

My computer had 3 fans:

PSU (Super Silencer 350)
CPU heatsink. The heatsink is by Alpha, so this fan sucks air from the heatsink and then exhausts it from the case through a duct. This fan is controlled by Speedfan.
Hard drive. This fan was controlled by Speedfan. The hard drive is decoupled by grommets.

I wanted to eliminate the third fan. I could get away with just removing the fan; it still ran cool enough, but I wasn’t satisfied, so I decided to make a heatsink/suspension mount.

I didn’t like the U channel my hardware store had, so instead I bought a 6” x 24” sheet of aluminum. After a few hours, I obtained the results seen below. In the side view, note the tight tolerances on that middle screw hole

(which I later realized wasn’t even necessary- duh).

The elastic material was taken from a sewing kit at home, and the washers are used to clamp down on it, which was very sturdy and easy to assemble.

After installing it, I got a huge surprise- seek noises were MUCH louder. There was some rattling noise. I don’t know what was rattling- maybe the hard drive was bumping against the plates, or maybe the whole structure was bumping into the hard drive cage. Temperatures were higher too.

I decided to remove this contraption and go back to the old mounting system. After removing the heatsinks, I noticed that the sides of the hard drive were very warm, several minutes after shutting down the computer. So a lot of heat is getting trapped in the air between the hard drive and heatsink. Despite that, the heatsinks got quite warm while in operation, so I think this design has a lot of potential. If you look at the heatsink on the left side, you can actually see some curvature at the mating surface… no wonder it didn’t work well.

This was a very rushed job, and it shows in the results. Eventually I may retry this, but I’ll need a better way to mount the heatsinks, so that they don’t get bent. Or maybe throwing a bunch of thermal compound in there would help.

nuttybing · Post by **nuttybing** » Mon Jan 31, 2005 9:19 pm

Is the aluminium sheet touching the hdd cage? From the first picture it seems like it does which i suspect causes the rattling noise. The entire thing should be suspended...
Anyway good idea on the heatsink part, you might want to bend the aluminium against the table to get a better bends

echoes · Post by **echoes** » Mon Jan 31, 2005 9:25 pm

When I installed it, it wasn't touching the cage. Maybe it was very close though, and oscillations from normal use would cause it to touch. I think it might have been the heatsink vibrating against the harddrive.

StarfishChris · Post by **StarfishChris** » Tue Feb 01, 2005 1:47 am

If the heatsink isn't fixed down everywhere (such as at the top, it seems) then you will get vibrations. Aluminium is still somewhat flexible

If you could get the bends a bit sharper so it becomes a much more snug fit it looks as if it will work well. Personally I don't think thermal compound is necessary since you're covering the whole drive - it's a lot more conduction area than you'd normally get in most cases.

colin · Post by **colin** » Thu Feb 03, 2005 4:43 pm

SpeedFan's temperature readings should let you compare the effects of different cooling strategies, rather than relying on touch. Record temperature figures for a whole usage cycle -- from a cold start through several hours of use.

I bet half the HD's heat loss is through the metal top, so as you note, anything that restricts airflow is going to hurt. The idea of putting a machined sink directly on the top, which someone else recently tried, (http://forums.silentpcreview.com/viewtopic.php?t=18943) is interesting.

The advantages of aluminum extrusions bolted to the sides are

(a) it's easier to get more direct contact with the sides of the HD because they're quite flat -- the fact that your bent metal sheet was bowed outward reduced direct metal-metal contact. (And you *do* want to use all three screws per side, because it's those tightened screws that will give you close contact between the HD sides and the heatsinking metal.)

(b) they move the heat out and away from the HD, rather than back around it. And depending on the space available, the pieces of extrusion can be quite a bit longer than your HD, extending out beyond the bay-space. Look e.g. for the extrusions meant as a guide for screen doors, which are flat on one side with fins on the other.

Also think about removing that upper bit of cage to give yourself room. Clear out as much space as you can, which will also let you use longer elastic and get better isolation. It may also let you maneuver the HD into a place where it gets better airflow.

Mats · Post by **Mats** » Thu Feb 03, 2005 7:58 pm

colin wrote:I bet half the HD's heat loss is through the metal top....

I'm not sure about that. Remember that all heat that reaches the sheet metal lid must first go from the moving parts through the casted frame, or go through air.
Cooling the lid is an easy but rather indirect method.

colin · Post by **colin** » Fri Feb 04, 2005 1:16 am

Hi Mats: This is an empirical question whose answer someone must have worked out, but the fact that almost all heat released through the top plate goes through the cast frame first does not by itself invalidate my guess about the plate's importance in shedding heat -- the plate has a large surface area, and it's far more extensively coupled to the frame than anything else -- even if you screw sinks tight to the side of the frame you usually get only a modest area of direct contact. I'm still advocating cooling via adding sinks to the side, but my caution was against reducing the effectiveness of what's probably the HD's most effective sink, the top plate. Cheers.

Mats · Post by **Mats** » Fri Feb 04, 2005 3:41 am

colin wrote:Hi Mats: This is an empirical question whose answer someone must have worked out, but the fact that almost all heat released through the top plate goes through the cast frame first does not by itself invalidate my guess about the plate's importance in shedding heat -- the plate has a large surface area, and it's far more extensively coupled to the frame than anything else -- even if you screw sinks tight to the side of the frame you usually get only a modest area of direct contact. I'm still advocating cooling via adding sinks to the side, but my caution was against reducing the effectiveness of what's probably the HD's most effective sink, the top plate. Cheers.

Well it's not that hard to figure out, just run Ansys or something similar!

I didn't say that using sinks to the side of the frame was the better alternative. The frame acts like a heatsink and while it did it's job quite good earlier with slower running HDs it's a different situation now. The heat output has increased but the frame has not, primarily for two reasons:
-Cost (needs no further explanation)
-Standards (the HD is expected to fit current systems)

The best place for a heatsink is of course where the motor is attached, under the PCB. I could make a heatsink made to fit there but the problem is the PCB, it needs to be there somewhere... and I get new problems when doing that.
This idea is quite theoretical right now but I'd like to try it out someday, possibly with an older HD.