The fan in the switch is a AFB0412VHB 3 pin, a 9500RPM fan. At 'idle' speed it is driven at about 6V average, max 12V (via PWM on 31kHz, as it should). You will need 12V fans. At startup, and at higher temps, the fan WILL get the full 12V.
A noctua nf-a4x20-flx (a VERY silent 5000 RPM fan) will NOT work as is, as 6V is too low for it to run. I have not tested other fans yet.
You MAY be able to get other fans to run, but it is likely that the 'fan fault' will be triggered. Why? Because of this:
The PWM controller is a TC654 (using a MOSFET output stage), so it does NOT use the RPM signal from the fan. The fan speed given by 'show system fans' is derived by the TC654 from the current consumed by the fan, and fan fault is determined by software, based on a minimum and a maximum value set at run time. If you do not use a fan with the same RPM range (or if you use a fan that has a significantly different internal architecture than the original fan), you WILL get a fan fault.
There are very few 9500RPM fans out there of the same quality and lower noise than the Delta fan. If you find them, please tell me.
BUT:
Even with 'fan fault' triggered, the switch will work though. So if you don't care about that alert, go ahead and hard-wire a fan on 12V. There is a big and a small power connector inside, find the right pins and you are good to go. If you use both 10gbt (copper) ports and some power hungry SFP+ modules, you may want to monitor the device temperature though.
Silencing the fan fault by simulating the fan on the 3 pin header will be very hard. I've read something about alarm silencing via SNMP, but haven't tried that (yet).
Note: Fan fault state will change after X (about 30) seconds, so do not expect immediate feedback if you test fans.
Note 2: the SG350X and SG350XG very probably use different hardware. The 350X board is re-used for all the PoE variants when looking at the board, but I do not expect XG to use the same board. The port layout is completely different already.
A noctua nf-a4x20-flx (a VERY silent 5000 RPM fan) will NOT work as is, as 6V is too low for it to run. I have not tested other fans yet.
You MAY be able to get other fans to run, but it is likely that the 'fan fault' will be triggered. Why? Because of this:
The PWM controller is a TC654 (using a MOSFET output stage), so it does NOT use the RPM signal from the fan. The fan speed given by 'show system fans' is derived by the TC654 from the current consumed by the fan, and fan fault is determined by software, based on a minimum and a maximum value set at run time. If you do not use a fan with the same RPM range (or if you use a fan that has a significantly different internal architecture than the original fan), you WILL get a fan fault.
There are very few 9500RPM fans out there of the same quality and lower noise than the Delta fan. If you find them, please tell me.
BUT:
Even with 'fan fault' triggered, the switch will work though. So if you don't care about that alert, go ahead and hard-wire a fan on 12V. There is a big and a small power connector inside, find the right pins and you are good to go. If you use both 10gbt (copper) ports and some power hungry SFP+ modules, you may want to monitor the device temperature though.
Silencing the fan fault by simulating the fan on the 3 pin header will be very hard. I've read something about alarm silencing via SNMP, but haven't tried that (yet).
Note: Fan fault state will change after X (about 30) seconds, so do not expect immediate feedback if you test fans.
Note 2: the SG350X and SG350XG very probably use different hardware. The 350X board is re-used for all the PoE variants when looking at the board, but I do not expect XG to use the same board. The port layout is completely different already.
Last edited: