Page 3 of 3

Re: Possible low speed fan firmware?

Posted: Sun Jan 27, 2019 10:59 am
by sirhc
lligetfa wrote:
sirhc wrote:With the new scheme the fan always starts at full speed or 8K for 1 second then slows down to 4K. We changed this behavior to help insure the fan does not get stuck on startup...

Rather than an arbitrary 8K for 1 second, since the fan has a tach, drop the speed the moment you see the desired RPM.


There are a lot of limitations on controlling the fans, for 1 we do get real time stats on the fans, we get 1 second intervals.

Again this is a small MIPs 24K CPU running everything. We also have limited GPIO pins so that means in multi fan units there is a MUX involved to get RPM readings.

We did the best we could, every issue your guys have brought up had been debated and argued and compromises made.

4K is simply the lowest RPM that is reliable on the new fans

Sending pure 24V (no modified pulse at 100 Hz or 25 KHz) to the fan before switching to modified pulse was determined most reliable to prevent a possible stalled fan. This was not needed with the old fans but as I have explained that fan was EOL and we were told this fan was drop in replacement which it was not.

We did the best we could, we had 3 people working on this for almost 3 weeks trying a lot of different things.

Re: Possible low speed fan firmware?

Posted: Tue Jan 29, 2019 8:01 pm
by scracha
Popped 1.5.0->1.5.1 on an 8 port DC switch and the fan is no longer switching on and off every 15 seconds. Will see of librenms reports anything cooking before I pop it on other towers.