I'm trying to build a DC light controller and I'm having all kinds of problems. My setup is a 12 volt car battery and a headlight, I have the CSD18535KCS MOSFET on the low side and an Arduino send a PWM signal directly to the gate.
IF I send pure 5V DC (from Vcc on the Arduino) to the gate the light of course it works and the temp stays around 100 deg F--i have a good size heat sink. If i send a PWM (32,250 Hz) let's just say a 50% duty cycle to the gate, the light dims, but the MOSFET gets super hot after a few minutes.
I have an LCD read out hooked to the Arduino to display the duty cycle and it goes crazy once PWM is turned on. If I disconnect the PWM from the gate, the LCD works fine. LCD not in the attached wiring diagram
Do I need to somehow filter the PWM to send pure DC to the gate? Is my frequency too high?
You have choosen a mosfet with a rather high gate capacitance (~6nF).
And you want the Arduino pin (without current limiting) to charge/discharge that cap 64,500 times per second.
Charging/discharging a cap takes time, and during that time the mosfet is not fully "off" or "on", resulting in heat.
The 10k resistor mentioned is a bleed resistor that is only functional during Arduino's bootup.
You need to add a 220ohm (>125ohm) current limiting resistor between Arduino pin and gate.
That protects the pin, but could heat up the mosfet even more.
Drop that PWM frequency to the default ~500hz, or even 200hz as I told you before.
Leo..
I tend to agree. Simply turning it on and leaving it on results one filament warming cycle, but repeated "On's" means it's hitting that low-ohms transition really hard, a real workout.
Depending on how much current we're talking about, like a car headlight, that could be a big deal.
Where are you getting this nominal 32kc from?
The frequency (from the Arduino literature is PWM frequency of 31372.55 Hz I figured 32K would be good enough for discussion. And I really want to avoid low frequencies due to the hum i'll get.
The current draw in 10amps. I'm now wondering of the light bulb (which is a coil of wire) is creating some inductive load and sending spikes back to the MOSFET--I guess a diode across the bulb may fix that.