Hmm sucking air vs pushing air, which one would keep the boards more cool with the least amount of noise? This is an interesting experiment i'd like to know the results, for science of course
You know in the other side without fans, there is a grill you can take off?
I believe the S11, S15/T15 and S17 will all share the same physical size, someone will build things for them.
In my situation, since everyone has different setups and needs, I don't really care much about noise. And yes I can remove the grill, but still would need something to attach there so I can connect the ducts. About the same size, the metal enclosure that house the hash board has the same size, BUT I'm noticing few differences in the grill design and screws position for example between the last batch of S15 and the older one (attached pic, new batch is on top). Not sure if there is any difference between S11/T15 and S15.
https://i.imgur.com/DjE4MO5.jpgThe main issue I'm trying to solve is that right now I'm unable to properly remove the hot air from the room. Because of the room design (no natural ventilation nor windows), the system has been running for the past 2 years (with S9s) removing the air using ducts attached with 3D printed adapters, so the hot air was thrown to a garage through several holes in the wall. This system has allowed me to mine under acceptable temperatures but right now the S15s are throwing the air directly to the wall, and temps are getting higher. Right now the miners are taking air that is warmer than it should, since it's not properly removed from the room; I'm in a heat vicious cycle.
Honestly, I wouldn't expect better cooling. Mainly if the fans could cool the chips better with that configuration the machine would have been built that way. Efficiency is the name of the game, and a better cooling design would reduce the work required by the fans, directly contributing to the machines overall efficiency numbers.
I also think that this setup is best unless Bitmain machined new heatsinks similar to what Canaan has had in the past. Pushing the air into the machine and creating that pressure build up through resistance to air flow and heat, pays off on the exhaust side. The air picks up velocity as it leaves the machine back to ambient room pressure.
In the end though, it's your machine and if you want to experiment, go at it slow like Phil suggested.
My point with inverting the fans is that I will be able to attach ducts again with the S9s 3D printed adapters and properly remove the hot air from the room, thus lowering ambient temp and (hopefully) cooling the miners. I know that in "normal" ambient conditions inverting them around will not make an improvement in temps, as you said if they were designed this way there must be for a reason
run it on lower power setting normal fan position with no ducts check temps and fan speed do it for a hour or more
run it on lower power setting with inverted fans and ducts check temps MY guess is it will work for sure on lower setting. do it for an hour or more
if it works on low setting and temps/fans hardly changed upward then try higher setting. My guess is it may not work on higher setting.
Right now I'm running them in low power mode to keep things under control. I'll try it this week and let you know.
Thanks for the replies guys!