Issues 'silencing' Supermicro JBOD

[SubnetMask]

I've read a bunch of posts about making supermicro enclosures quieter, and rather than necroposting, figured I'd start a new thread.

I have several CSE-826 enclosures that are all set up as JBODs with JBPWR2 control boards and BPN-SAS2-826EL2 backplanes, with the fans connected to the backplane, and of course, the stock 94L4 and 126L4 fans which are quite loud, even when they spin down to their lower speed after everything starts up. From my reading, the FAN-0074L4 fans were said to be much quieter, so I ordered three to give them a whirl. Once I swapped the fans into the tan 'caddies' and installed them, when I powered the unit on, they powered on at full speed, which I'd say is roughly the same noise level as the other fans at their reduced speed, but the problem is they never ramp down like the other fans. I was able to get them to run at the lower speed by powering the chassis up with the noisier fans installed, then after they ramped down, swapped them one by one and they continued to run at the lower speed. This of course is not a feasible solution. At the lower speed, the 74L4s are quite a bit quieter. Around 10db it seems. Although interestingly, when I connected them to a SAS-826-EL1 (SAS1) backplane, they DID ramp down. I also found that if I had one of the louder fans in along with two of the 74L4 fans, it ramps down, but this pretty much negates the entire point as the louder fan still makes quite the racket.

I could go with something like the Noctua fans, but the problem with them is they're running half the airflow at full speed (I think they're rated at 34CFM and the 74L4s are rated at 68CFM full speed), which I think should be enough, but if they're running at full speed all the time and an environmental or load change causes some higher temps, they cannot move any more air, whereas if the backplane would ramp the 74L4s down, if it needed more airflow to keep things cool, it's got headroom to increase the speed. I could also go with some sort of external fan controller, but that has the same drawback as the Noctua fans as if more airflow is needed, the backplane can't increase the speed.

That being said, has anyone else had the same issue with the 74L4 fans not dropping to a lower speed after initialization, and if so, was there any fix that would allow them to work right?

 

[TXAG26]

Is it possible to adjust the IPMI fan thresholds on the JBPWR2 board (like you can with IPMI Tools on Supermicro motherboards)?

 

[SubnetMask]

I wouldn't think so. There's an i2c port on it that I believe is intended to connect it to the backplane, but there's pretty much no intelligence to it. The CSE-PTJBOD-CB3 might be my only possible option here.

 

[TXAG26]

I think I've seen some fan controllers that have a temperature sensor/probe that control the fans based upon temperature settings.
Maybe something like this would work?

Coolerguys 12v PWM Fan Thermostat Controller

This is perfect for an audio visual, home entertainment, or other enclosure where keeping noise to an absolute minimum is vital. Unit has been tested to work for up to 2A worth of devices.

www.coolerguys.com

 

[SubnetMask]

That's a neat little board... too bad they don't seem to have one with three or more outputs... Not sure how it would work driving three fans off of the one output. I'd be somewhat surprised if it worked right. I wonder if someone else makes one with more outputs. I'll have to look around.

Edit: Search and you shall find lol: https://www.amazon.com/Temperature-Control-Speed-Controller-Module/dp/B019P0FLHW

Although I have to admit these cheap little boards make me a little nervous... Would hate for one to croak and cause my array to cook. But I might have to check one out.

Edit2: I wonder if the tach signal can be split off and sent to two places: The cheap control board and the JBPWR board, so if the JBPRW sees a loss of speed, it'll scream... That would probably make me feel a little better.

 

[kapone]

Combine either of the two fan controllers above with...

Voila! TEN outputs.

 

[SubnetMask]

So it seems like my only real solution to this problem will be to get the SuperMicro CB3 power boards, which give more control over the fans.

SuperMicro was of no help - the agent I was communicating with suggested that the reason the stock fans spun down was because they draw too much power, but he couldn't explain why they spun down but the 74L4 fans did not (other than suggesting that the higher speed fans draw too much power). My best guess is that the backplane is looking for a particular speed on startup, and if it doesn't see that speed from the fans, it doesn't ramp them down, and the 74L4's don't hit that mark (again, a guess). The only other possibility might be modifying the .FW file for the backplanes as looking at it with a text editor, there is some readable data there and references to 'fan', so maybe, if someone knows the layout in the file, it might be able to be modified with a hex editor, but that might be a long shot, but modifying the file incorrectly could potentially brick the backplane, which would be no good..

So I ended up ordering one of the three port boards to give a whirl. Besides the fact that I apparently got a bad one because no matter what, fan1 would never run at anything other than 100%, the other problems with that board are only fan1 is temperature controlled and the speeds for fan2 and fan3 are controlled by pots on the board, and they will never run at anything other than what is set by the pots. So if you have them all running at 40%, for example, and the temp goes up and fan speeds should increase, ONLY fan1 will increase speed. Then only fan 1 & 2 are monitored for failure, so you could theoretically end up with two fans failed before you're alerted.

The problem with the single output board and the 10 fan splitter is kind of similar - My suspicion, without looking at the bottom of the board, is that only one fan, most likely the one labeled 'CPU', is monitored for failure, so while all fans connected would be running at the same speed, and in theory, they all would increase speed if the temp increased, you could, in theory, end up with two fans failing without knowing it until the third and last one fails.

With the discussions around about quieting these, I was hoping someone had succeeded in doing something like this, but perhaps most of the people that have swapped fans like this were running systems with motherboards to control the fans, which is a different scenario.