# Selecting a MOSFET to drive a coil with a square wave

I need help with finding an alternative to the IRF1310 in the circuit shown. The circuit is going to be controlled with a microcontroller with 3.3 V logic.

The simulation of the circuit in LTspice shows the signal forms shown in the figure and was successful in doing what was intended, but the current at the gate is high (1 A peaks every cycle).

The input signal is a square wave, simulated as a source, but in the application it will be microcontroller output with a frequency of around 3 MHz.

The reason I am not using a more complex circuit is that I need to fit everything on a coin-sized PCB with the least number of components used (cost reduction). SMT is preferred if available. I need to keep the gate current to a level that can be driven directly by the microcontroller.

• What you are looking for does not exist. In order to get a low Rds-on, the mosfet has to have a relatively large junction which means it has a large gate, which in turn means there is significant gate capacitance and charge exchange associated with switching. The faster you switch the more current you need to drive the gate to saturation. You did not specify the required output current, assuming it is non-trivial you will have to use a gate driver or a mosfet with an integrated gate driver. Commented Oct 23, 2017 at 19:24
• Also be aware that while the IRF1310 might turn on at Vgs of 3.3V (4V is the upper limit of Vth in the datasheet so it also may not), it will not be fully saturated and you may end up with significant conduction losses in the mosfet. You will also have significant switching losses @ 3Mhz. Commented Oct 23, 2017 at 19:28

1 A peaks every cycle

As is often the case with simulations: garbage in, garbage out. The circuit you're simulating isn't anything you can actually build, but it points you in the right direction.

You're driving an almost ideal capacitor - the gate capacitance - from an ideal voltage source with an ideal square wave of infinite bandwidth. "Infinite" current peaks are a reasonable expectation - the simulation would be doing a bad job otherwise :)

Are those peaks necessarily bad? You want to drive the gate with fast edges, the multi-ampere peaks are a fact of life. That's why gate driver chips are made: because most digital logic can't quite drive power MOS gates directly. If the driver can't supply such peaks, you won't get fast edges. It's one or the other. Most likely you don't need super fast edges. It all depends on how you apportion power losses in the circuit: no part is ideal, so the inductor windings will dissipate some energy due to ohmic losses, the inductor core will dissipate due to magnetic losses, the load rectifier will, the mosfet gate capacitance, the mosfet drain impedance during switching, the drain impedance during conduction (fully switched on), etc. You have to put real numbers to all of those and decide on what's acceptable. Make a power budget, and let that dictate what's important and what isn't.

Just for reference, 1A peaks lasting 50ns at 100kHz rate would average to a 5mA current. Is that a problem somehow? This would be the gate loss current in your circuit. To lower it, choose a part with lower gate capacitance, or lower the operating frequency. No other way around it.

In the simulation, you need to drive that gate from a source more representative of the real world. If the source is an I/O pin, then 5mA/V source resistance may be a good start, there should be some parasitic inductance of the trace from the MCU to the gate, and the source signal bandwidth should be limited as well - e.g. only the first 10 harmonic components of the square wave.

There are plenty of N-Channel FETs that might suffice for your task.

You are searching for a FET with low RDS(on) but also low C(gate).

You can search on Digikey or the like for low C(gate) ... <50 pf, and low RDS(on) (though this can be part of your 2.5 &ohml resistor in reality) and low VGS(threshold)

Here's one that might fit your need, the PMZ390UN. This has under 50 pF gate capacitance and is able to support your ID needs of 1.4 A at 3 V drive.

The RDS(on) is under 500 mΩ, so you could drop your 2.5 Ω resistor to 2 Ω to allow for it.

Turn on/off times would easily meet your 3 Mhz requirement, but remember that delay will impact any phase relationship you might want to maintain.

Since you are driving an inductive load, there will be no problem with turn on characteristics, but you might want to add a very fast Schottcky diode across the inductor.

• Maybe not. 1.4 amps and .5 ohms gives 1 watt, and at 50% duty cycle that's going to be .5 watts (plus switching losses). On a coin-sized circuit which is probably tucked away somewhere I'd worry about heating. Commented Dec 4, 2018 at 2:00