Electronics Questions & Answers ← browse Q&As

I'm trying to turn on and off a 100W LED very fast, on the order of 10micro seconds, using a microprocessor. The microprocessor is working fine for lower powered LEDs. The 100W LED takes 30V to turn on so I got this MOSFET to do the work, but its Slow. Are some MOSFETS faster than others or is a 100W LED not going to switch on and off fast?

I'm trying to make a strobe scope for fast moving machinery so the switch on and off times must be very fast or the object will just blur.

What are your two scope channels connected to? Are you driving the MOSFET's gate directly from the microcontroller pin? Could you please draw a schematic?

The IRFZ44N datasheet shows gate charge Qg of 63 nC. If you want to push 63 nC in 10 us, you need a gate current of 63e-9/10e-6 = 6.3 mA. This should be achievable, so I'm guessing your delay has something to do with how you're driving the gate (or some unexpected behavior in the LED itself).

Here is the circuit diagram:

Note On my oscilloscope I inverted channel 2 so I could use the measure functionality to show the delay (Dly_B=19.60ms on the scope screen capture)

When I replace the LED with a 4kOhm resistor, the response time is much Faster. The I put the LED back and it seems to run Faster. I don't know what I was doing wrong the first time???

Great! Looks like maybe it's taking 400ns to turn on or so?

Not sure, but it looks like you're now being limited by only driving the gate with 5 volts -- 10 would be better, not just by a little bit but by a lot.

Adding a gate driver IC might be a solution. This would use the 0-5V voltage from your microprocessor to control a higher-voltage, high-current signal to drive the gate more swiftly if you had a 10V source available.

Notice how CH1 goes from 0V to 4V basically instantly, but then takes a few microseconds to climb that last volt? You're being limited by the ability of the microprocessor's pin output transistor to push current when it's near its supply rail. If you have extra pins free on your microprocessor, a cheap workaround might be to wire several pins together to all drive the gate, and be sure to switch them all simultaneously in software. (Make sure you can switch them all in one atomic CPU operation, so they aren't ever fighting each other.)

Note all the plots are using function generator with a 50Ohm output impedance. I assume this is stronger of an output compared to a microprocessor. Doing some quick simulations the 10V driving is much faster

Plot

I don't think I could use two pins combined as I'm using match interrupts to control predefined pins. I assume adding the 5-10V converter time would be negligible. Overall its seems to be working good enough to keep pushing forward on other aspect of the project (making a case exc.) Thanks for your input.

Great! I really like your turn-on time simulation with the sweep over gate voltage. In any case, glad to hear it's working now.