Quote from: kreativling on February 06, 2015, 09:56:29 PM

Quote from: dogie on February 06, 2015, 09:47:01 PM

Quote from: netcoingeek on February 06, 2015, 06:46:42 PM

Quote from: MrTeal on February 06, 2015, 06:35:34 PM

Quote from: netcoingeek on February 06, 2015, 06:29:49 PM

Back to fans. Noctura not-f12 is supposed to be a quiet 120mm with speeds up to 3000rpm. From reading all this, noise is the issue but you have to decide if you want speed or noise. The faster the s5 hashes the louder the fan because in order to keep the s5 in working temps (below 60) you have to push a lot of air and have the s5 is an ideal location. If you want a quiet fan then you need to sacrifice hash speed because the air displacement from fan has to exceed the heat generated to keep the s5 cool. This has been my struggle.

But I can't figure out why the s5 needs to be below 60c? Most computers run at or over 60c. The over temp cutoff is software set at 80c. I do get some hw but over a 12 hr period and from what Ive read some hw is ok but when does it become an issue?

Unlike the Spondoolies or Hashfast chips (or your CPU) the S5s don't have a thermal diode right on the chip. The temperature sensor is on the back side of the board, and so is really measuring the temperature of the PCB. It's unknown (at least to me) what a good rule of thumb is for mapping die temperature to the temp sensor reading, but it's pretty much guaranteed to be a decent bit higher than what the unit says.

Ok, so like most temp sensors, there is a variance and to error on the side of hot is better. Ok so my s5 running at 55 and 58 are probably running more like 60 and 65. But hardware errors? What causes them to occur and why would they occur at temps in the safe zone with no overclock, underclock say at 275Mhz? The miner still functions with hardware errors and they are less frequent than with my antminer U3. I guess the goal is 0 hardware errors =stability but 0 hardware errors = cooler temps in the below 60c range and that is where fans come in.

Probably in the 80s, its hard to tell. I'll see if I can measure the external chip temp at least when I'm next in there.

I would think the actual temperatures may be even higher, as the data sheet for the BM1384 states a max. operating temperature of 125° Celsius.

That's a crit temperature so there's likely to be a safety margin still within that. Say, 15C. So 80C = 110C and 55C = 85C. It could even be lower as the 80C threshold is arbitrary and more of a 'something is wrong if its got this hot' limit.