We have a fan driver circuit from a customer that doesn't work. I think the design intent was to filter the gained-up (logic level) pwm signal. The pi-filter LC section has a cutoff of 72Khz, so I assume it was to be driven with a 100 or 200 KHz pwm signal. It does not work. The mosfet switches too slowly and it really never turns off completely. I am thinking this circuit could be fixed with some sort of improved gate driver for the mosfet.
We need to retain the high-side position of the switching mosfet because the customer fans have "locked rotor" detection circuits that require the fan to be ground referenced. Just replacing the circuit with a buck convertor was considered, but the customer board has separate drivers for 5 different fans, and so the driver circuit needs to be cheap, simple, and small.
I am looking at the tps2819 gate driver, but I'm not sure it will work in this circuit. Having to operate from 24V supply limits available options. Anyone know of a creative way to use this gate driver chip (or some other TI driver) ??
Hi Lee,
I'd reduce value of R41 to 1k or less to have more than 12 V on the gate from U31 when Q3 switch LOW. The Gate-Source Capacity must be always charged and discharged through 10kOhm and it could be too much for your PWM signal.
That change does indeed speed things up a bit, but the real problem here is that I misinterpreted the intended operational mode. They want to drive the fan directly with the switched signal and let the fan inductance do the "filtering" such as it is. At low pwm frequencies this circuit works ok. I was trying to run it at too high a frequency. I think the matter is resolved and I won't be needing any kind of gate driver.
Thanks,
Lee
You stated that the customer wants the inductance of the fan to "filter" the current. For that to work, the capacitor that is connected to the FET's Drain should be replaced with a freewheel diode of sufficient current and voltage rating. Be careful not to over drive the FET gate; 24 Volts is a bit close to its limit.