Wind Tunnel for Audheid K7 NAS Case

TLDR Version: If you want an easy upgrade, get better 92mm fans that move at least 45 cfm, print tunnel-by-cpu and tunnel-behind-ps, and install them. It'll be good enough, but not the best you can get. If you want the best, print rearbox-under-cpu and rearbox-by-ps, get 2 x 92mm fans with at least 45 cfm, build the shroud behind the drive cage, and tape it with conduit silver tape, optionally printing complete-seal-ps and complete-seal-cpu (or use paper).

SLDR Version (Semi Long Didn't Read): Air flow around the drives sucks and the only way to ensure your drives don't get too warm is to restrict the air through the drive chamber by either pushing air in through a sealed front or pulling air through a sealed rear of the drive chamber. The front is easy, but ugly and inconvenient. A rear-sealed setup with 2 good 92mm fans and rear-under-cpu and rear-by-ps is harder to set up, but pays off in the form of aesthetics and convenience. Leaving the original fans in place helps lower the temperature even further.

Long version and why:
The Audheid K7 NAS case is a great case for NAS with a backplane which will accommodate not only SATA drives, but also SAS drives which can be obtained for cheap. It seemed, however, that the case designers had either no idea how to vent air past the hard drives, or were told to make it look good and not worry about air flow. As a result, airflow is roughly nil between the drives. During my testing, operating the machine without the case garnered a couple of drives at 71 degrees C. Don't do that! Operating the machine with the cover on resulted in better temperatures, but as others have said, the drives still ran too hot. So I decided to set up an experiment I could redo with different parameters.

For the experiment, I noted the temperature of all the drives, busied all 8 drives for 1 hours and took measurements at 25 minutes and 60 minutes. I then measured how long it'd take for the total drive temperatures to fall down while there was no activity.

Baseline:
The first test was to set up a baseline without any mods. Initial drive temps were 238. 25 minute temp was 320 and hour mark was 348. The temperature continued to climb after activity on the drives was stopped. 30 minutes after the test was over the temp was 362. It was clear the fans and the venting mechanism weren't doing any good.

Looking at the case, it was clear that the two fans in the back weren't pulling much air through the front vents because it was easier to pull air from elsewhere. I considered getting 92mmx92mmx38mm fans to replace the fans in the back, but the fans were $16 each and I wasn't sure if they'd actually work. Then I wondered if I by channeling more air through the front vents and restricting airflow elsewhere I'd get better temperatures, so I tested the case with cardboard stuffed on the CPU side of the drives and the slit above the power supply separating the upper and lower chambers. It made a difference, so I decided to create some STLs for those pieces.

CPU and Above PS Brackets:
I printed tunnel-by-cpu and tunnel-above-ps and tested again, starting with drive temperatures of 208 C across all drives. 25 minute temps were 268 and hour mark was 360. After the process was done, I expected the drives to start to cool down, but after an hour of inactivity, they were still sitting at 360 C. So I removed the tunnels and put the case back on with the unit running. The temperature climbed to 402 C. Thus I believe the tunnels do help on the order of 10-12 percent, but not enough. I didn't want HDs to sit at 45C all the time after some busy activity.

CPU Bracket and Tunnel Behind PS:
Next I printed tunnel-behind-ps model to sit behind the power supply. For this test, I removed the tunnel-above-ps strip since I wanted to make sure the PS was going to get enough air from up above. Testing started with temp of 248. After 25 minutes, the drive temperatures had climbed to 311 C. Interestingly, the CPU temperature had also climbed a few degrees, but it still averaged around only 38 C and I wasn't worried about that. At the 60 minute mark, the temperatures totaled 330 C. I was hoping the temperature would drop now that all the air the two fans in the back were pulling would be coming from the front between the drives only, but to my dismay, the temperature did not budge. After 25 minutes, the temperature had not changed. So the new tunnels helped, but just barely. My only conclusion was that there was basically close to no air flow coming from the front vents by the drives. This was confirmed by seeing if any air was coming out the fans in the back and there was none. Once I removed the drives, plenty of air started moving out in the back. I believe this to be a design flaw in the case. I took one of the caddies completely apart to see if there was any room for improvement on the air flow from the front. Short of cutting out most of the metal bracket under the drives by the circuit board, there's not much room for improvement and even that wouldn't be that helpful.

CPU Bracket and Tunnel Behind PS with Front Fans for Push-Pull:
Not wanting to give up, I decided to finally take the plunge for some external fans. I bought a pair of 92mm x 25mm fans, but I wanted them in front of the drives, pushing air through, while the rear fans pulled the air out while the tunnels on the two sides restricted airflow from elsewhere. I printed 2 x tunnel-front for the fans. For this test, the starting temperature was 240. After 25 minutes, the temperature was 243 C and after an hour it was 247 C. Woot! We made a difference. Not only that, but after 30 minutes we were back down to 240 C. So there IS a solution to keeping the drives cool.

CPU Bracket and Tunnel Behind PS with Better Fans in Back:
It occurred to me that perhaps the fans which came with the case were just really crappy, so I decided to exchange them with my new 92mm fans which were supposed to move 48 cfm each. I replaced the fans and the first thing I noticed was that these fans were noticeably louder, a sign that they're actually moving more air. I redid the test with tunnel-by-cpu and tunnel-behind-ps installed. Starting temp was 192. 35 minute temperature (sorry!) was 275. Not bad! 60 minute temperature was only 282. It continued to climb to 285, but again, to my dismay, it stayed at 285 even after 20 minutes and 288 after 40 minutes. Could I live with that? That's an average of 36 for each drive (they were all 36 except two which were 35 and 37). The ambient temperature was about 27. I considered printing the fan shrouds https://www.thingiverse.com/thing:469865 and https://www.thingiverse.com/thing:2732939 to put some 56 cfm 120mm fans I had in the back on the outside, but before I did that I wanted to make sure there weren't any more openings I could close to tighten the vacuum the rear fans were creating in the box. So I decided to put some tape at the edges of the tunnels I had created to close some of that gap. It's pretty hard to do that, though.

CPU Bracket and Tunnel Behind PS with Exterenal 120mm Fans:
I could have let things go, but I wanted to be sure there was nothing I could do to improve the temperature coming down after the test. I printed the 120mm to 92mm shrouds for the back, but this time I didn't want to cut any corners, so I covered every nook and cranny I could find around the drives and sealed the seams with conduit silver tape. To get the fan shrouds installed on the case you'll need a small ratcheting tool with a phillips #2 head. I also wanted to be able to easily change the 120mm fans in the back, so I used two fan cable extensions and fed them through the grate in the back after cutting a few of the grates off. Once everything was installed, I started the test again with a temperature of 208. The temperature climbed quickly at first and after 25 minutes, to my chagrin, it was at 296. Bummer, and not a good sign, since the temperature with the 2 x 92mm fans and mediocre sealing was only 275 at the 35 minute mark, but I figured I'd finish the test. At the end of the test, the temperature was 321 and after an hour it was 323. My guess was that the fans weren't really pulling 56 cfm. My main objective all along had been to bring the drive temperatures back down after heavy activity and the only time it really worked was in a push/pull setup. The good thing, however, about having the fan connectors and the shroud on the outside was that I could experiment. So I disconnected one of the fans and connected a brand name fan on top of the fan. Nothing changed. Next I put the shroud in the front with the two 92mm fans and turned off the fans in the back, thinking that there's zero air escaping in the front chamber. The temperature started to fall and after 10 minutes I was at 290 and after 20 minutes it was 274. I was still missing some holes which I hadn't taped (Note: I noticed them later. They're the opening underneath the motherboard which connect the above chamber to the lower chamber).

Rear Shroud Behind Backplane:
Either the 120mm fans I had weren't as good as the 92mm fans, or I had missed some other "hole" which was affecting the vacuum, which I suspected to be the plate underneath the motherboard. I had one more option left before I gave up and went back to 92mm fans inside: create a shroud for the inside of the case and behind the back plane. So that's the two rear-under-cpu and rear-by-ps prints (fan grill is from https://www.thingiverse.com/thing:2802474 ). You have to do some hand modifications to them after they're printed, namely cut the upper corner on one side so it slips nicely over the other one, and then cut away whatever holes you need for your cables.

I printed the shrouds, installed them, and put conduit silver tape everywhere I could see and think of. For the fans I powered them directly from the PS and not the back plane. The test started with drive temperatures of 198. 25 minute temp was 297. A bit high compared to other tests, but I was really shooting for the temp coming back down after the test was done. It didn't look promising, however. I was already thinking ahead to what I was going to do next: put the original 92s back in and run with 4 fan. The advantage to this setup would be that not only would I be pulling the air with the fans in the shroud behind the back plane, but I would also be cooling the case as a whole with the additional 2 original fans. But we're getting ahead of ourselves here. 60 minute temp was 333. Temp went to 343 after an hour of idle. So I decided to put the extra fans in while the machine was still hot. Immediately temps started to come down.

Rear Shroud Behind Backplane with 2 Extra Case Fans:
You know this set up from the above paragraph. Start temperature was 273. 25 minute temperature was 281! Woot! This was the solution I have been looking for. Final temperature was 282 and was back down to 275 after 20 minutes.

Printing Instructions:
PLA is fine and no support should be needed.