Cooling Harddrive Thermal Dynamics

So I'm probably going to be milling some custom heatsinks for hard drives, and I've come up with a road block. Which part of a hard drive gets the hottest? The bottom, the top, or the sides and front? Even better, if someone has a FLIR camera and can stress test a hard drive, I'de be in heaven. I wish I had one of those $15000 cameras.

I've also noticed that most companies cool only the sides, like Zalman's heatpipe cooler from the fanless case. Is that because it's the best zone for thermal contact, or because it's easier to encompass it side to side than to create a whole 360deg heatsink?

 

AFAIK, HDDs are designed that way that they move most of the heat to the sides.

 

Thanks, I had figured as much, but with the motor and chips being centrally located and on the bottom, I wonder how hot those components get in comparison. In other words, is it worth it to engineer some contact points to those locations I wonder... *ponder ponder*

 

good airflow is enough. it's not like it's going to overheat.

 

I'm designing them for use in an entirely custom 1U server mod with liquid cooling/ custom hot swap bays and air cooling. Unlike in desktop use, hard drives DO get cranked out in a server, 140deg is a temperature rarely hit in desktop use, but most SATA drives will hit that temp. under constant use. Keeping it down under 110 is my goal.

In case you're wondering there will either be 2 3.5in in RAID1 or 4 2.5in in RAID6 in a little itty bity 1U chassis. I am planning on building my own distributed computing system.

 

all hard drives require is a slight breeze over the top of the case for adequate cooling.

 

I take it noone has a FLIR Camera or anything.

 

Chances are slim tbh. HD's are designed to channel heat through the casing, to act like a heatsink. Light airflow over the chips, and a decently cooled mounting should help keep 'em nice and chilly.

Can we see pics when you've finished?

 

Once I get going, I intend to post the whole worklog on here. At least what's corporately allowable. We'll see, right now this is all just planning and speculation. And me being entirely dissatisfied with Icy Box and similar companies.

 

The use of rubber mounts in "silent" systems suggests the edges don't normally transfer much heat to the cage, it's convection over the top & bottom sides. I've known a drive in a system with a broken fan get hot enough to crash the system in under an hour.

There is a thick thermal pad available for transferring heat from circuit boards to a metal heatsink.

 

That's some pretty cool stuff... but what do you think about the coolers like the Zalman fanless cooler that only connect to the sides?

 

In my line of work, I see a lot of servers, though stress testing isn't one of my main job functions. Still, all the 1U systems I have worked with do just fine with air cooling. The problem is that the fans are often cooling everything in the system and are a bit on the loud side. Still, it doesn't take much to keep them happy.

Seems there are some thermal images of mainboards and video cards, but the only images with hard drives in them were from angles that made it hard to get any information

There were some HDDs with integrated heat sinks, they were always on the sides and back of the drive.

 

Any body hotter than its surroundings will lose heat and set up a thermal balance at some equilibrium temperature, and being a good conductor the drive case forms a useful path (if not blocked by rubber, etc). The Zalman cooler has nice chubby cheeks to move the heat into the pipes and away from the drive, but where does it go then? Some air movement over the heatpipes is needed to convection-cool them, shiny tubes are poor radiators. Is it better than just the same airflow over the conventionally-mounted drive body? Probably.

Now this is crude but may give comparative results if distances are standardised. I've an IR "ear" thermometer, cost me £8. Pointed from 1cm at my morning coffee, it says it's 42.9°C.