Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
In case anyone else struggles with fan control on similar X11 boards, the approach I'm taking now is to use a Nocuta NA-FC1 to reduce the PWM signal to reduce the amount of noise the chassis produces. I got a bunch of fan extension cables, and plan to cut the PWM wire from each fan extension at the motherboard (female) side of the connector and join these together at the PWM controller, and have FAN1's tachometer reading pass through the NA-FC1 so that it can handle the "low RPM" solution that Noctua provides. This has the advantage of not driving all the fans off the same motherboard header, and avoids the need to do some crazy Molex->Sata->NA-FC1->fan splitter to three (w/ two fans) -> fan splitter to three (w/ two fans) to power four fans.
I'll try to post some pictures of the wiring harness once complete.
I'll try to post some pictures of the wiring harness once complete.
Have you tried rebooting (the BMC and server) after setting it to FULL?
I have the same problem and this is my last hope but I can't try it until I'm back home.
Yes, I've pretty much exhausted my options. I ended up installing an NA-FC1 between the FAN1 PWM pin and the PWM pin for all the fan connectors. I can't upload the picture for some reason, but essentially what I ended up doing was taking a 5 pack of 4 pin fan extension cables, and removing the PWM signal wire from each fan header, reconnecting all the extension cables between the FANx header and appropriate fan, and joining all the PWM controllers with the NA-FC1 inline on FAN1. This allows the NA-FC1 to see the fan speed of FAN1 and control all of the fans, without needing 4 NA-FC1s or pulling all the power from a single fan header. As an extra bonus, since the tachometer wire from each fan still goes back to it's appropriate header, the BMC still sees each fan's speed. The system runs near silently now, and even has better thermals than my 2U and 3U chassis...
Just to add to this. For me on a X10 board (X10SRW-F) it only worked after a reboot of the unit and the ipmi. Then the commands would stick. It will default back to FULL after a reboot or messing with the fan settings again. It's now set on 20% speed and scripted it to set it back to 20% after a reboot.
I knew, but I was reading contradicting things. Just wanted to be 1 more voice if anyone is searching about this.
Was wondering if anyone had a similar experience. The fan control does work on an X10SRH-CF, but after a while the fans in the zone revert to the standard profile. I'm only setting a speed for peripheral fans. It doesn't seem to be a lower critical threshold issue and there aren't any corresponding logs for it in the IPMI.
It doesn't seem to be rebooting. So GD annoying. Why can't this sh!t just work and stick?
I've read the article about three times and still don't get it.
These commands do or do not let me set fan speeds? What does "PWM duty cycle" mean anyway?
I have one of the case rear fans going too fast and it's noisy. I presume this won't help me in any way?
These commands do or do not let me set fan speeds? What does "PWM duty cycle" mean anyway?
I have one of the case rear fans going too fast and it's noisy. I presume this won't help me in any way?
In a nutshell, no, you can't set individual fan speeds directly. The commands let you set either a) alert thresholds or b) duty cycles.
Duty cycles are a PWM thing and aren't done on an rpm basis but rather what your PWM-capable fan works at, it's basically a percentage-grade of how fast the fan will run. 10% duty cycle should run a fan at 10% of its maximum rpm, 100% duty cycle should run a fan at its max rpm, so obviously it depends greatly on the type of fan you use.
Supermicro IPMI basically supports two "fan zones" as far as I'm aware; one zone for the CPU fans, another for the system fans (which as far as I can make out means "not a CPU fan").
Duty cycles are a PWM thing and aren't done on an rpm basis but rather what your PWM-capable fan works at, it's basically a percentage-grade of how fast the fan will run. 10% duty cycle should run a fan at 10% of its maximum rpm, 100% duty cycle should run a fan at its max rpm, so obviously it depends greatly on the type of fan you use.
Supermicro IPMI basically supports two "fan zones" as far as I'm aware; one zone for the CPU fans, another for the system fans (which as far as I can make out means "not a CPU fan").
So this is only useful if you mostly use fans with the same or similar rpm I guess?
I used a some low noise adapter I got with some old Noctua fans in past and that mostly solved the problem, but it's still a workaround at best.
I'm wondering wherher I should use variable speed fans with PWM connectors instead of regular 3pin fixed speed ones in this particular case.
I used a some low noise adapter I got with some old Noctua fans in past and that mostly solved the problem, but it's still a workaround at best.
I'm wondering wherher I should use variable speed fans with PWM connectors instead of regular 3pin fixed speed ones in this particular case.
Duty cycle only works with PWM (4 pin) fans, and yeah you'll want the fans to have similar characteristics. Duty cycle changes won't affect 3-pin fans and it sounds like that's what you're using - you regulate the speed on those by changing the voltage which is basically what those low noise adapters do.
Might take a look at this:
https://www.amazon.com/gp/product/B072M2HKSN/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1
https://www.amazon.com/gp/product/B072M2HKSN/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1
I've recently done a few more builds with the NA-FC1 and asked Noctua if they can either make a smaller version or adjust the wiring to make it easier to fit in a server chassis. If this interests you as well, you might want to let them know. AFAIK, they're the only ones offering a PWM reducer solution like this that doesn't involve sticking a resistor in series with the fan.
Noctua on Twitter
Noctua on Twitter
Is it possible to run ipmi tool under a virtual environment like debian? I can´t seem to load ipmi_si because of this.