I'm making a PWM driver for a DC motor. After some questions were answered in this post, I ended up with this schematic:
And I have some new questions:
Do I need a resistor between the gate and the output of the gate driver (U?)?
Do I need a pull-down resistor at the gate of the MOSFET?
Maybe. The MC34152 datasheet pp.8 on shows a series Rg to damp oscillations, and reverse-bias Schottky diodes for catching negative ringing spikes at the driver. Wouldn't hurt to have these in your layout. You could stuff zero-ohm for Rg if you don't need damping, and no-stuff the diodes if you find the ringing isn't too bad. Have one resistor/diode per FET, don't share them. Place them near the gate.
No pull-down is needed at the gate drive. But you will want a pull-down on the driver input to make the default state 'off'.
Also, while we're discussing the inputs - tie them together and use both of the separate outputs, one for each FET. The way you have it - driving 2 FETS together - kind of defeats the purpose of the buffer.
Finally, if your motor is a normal brush type you will want to use a freewheel diode across it to catch the flyback spike when the switches turn off. For BLDC this isn't an issue. Yes, C2 does this too, but the diode is better.
Some FET circuits, with various parasitic capacitances (including inside the FET) and inductances (including the FET packaging) and lumped loads, WILL OSCILLATE.
Having gate resistances (as an option) provides a means to dampen the circulating energy, if the C_gate_drain or C_gate_source is in the resonant loop.
You might not need a gate resistor but if this is a custom PCB, you should make space for one and just jumper it if it is not needed. Gate resistors slow down the rise time which can result in EMI, ringing, and damaging spikes. Ferrite beads can also be used instead of resistors.
You want one resistor per gate and you want the resistor close to the gate. This is for ringing issues should they arise. You don't want to share gate resistors because parallel MOSFETs can ring with each other.
You should have a pull-down if your drive circuit doesn't sink it when powered is off since your MOSFET gate capacitance will just remain charged and keep the MOSFET on. It also helps keep the MOSFET in a default state if your driver is busy powering up if your driver doesn't handle that well either. I don't think you need one with the gate driver you are using.
You should have a flyback diode anti-parallel to your motor. C2 does help though. Your gate driver IC also needs a decoupling capacitor.
No, you do NOT requires a pulldown resistor for the gates. This is already built into the driver and the driver when powered on will immediately have a defined state.
You MAY require a series resistor since the the driver is rated for only 1.5A maximum. The series resistor is used to limit maximum current and reduce parasitic ringing which may increase dissipation in the output FETs.
As shown in the datasheet:
As mentioned in another answer you could put a diode across the motor to control back EMF instead of the capacitor, but neither of these are a good solution.
The best way to get rid of the stored energy as fast as possible is to use a Zener across your drive devices. If you do this you will definitively require a series resistor from the driver to the gates since the Zener will hold the drive FETs on to dissipate the back EMF..
Update: One of the comments exposed (@DKNguyen) exposed the potential for gate oscillations between the paralleled FETs, and linked to this paper.
From this it would appear prudent to include a separate series resistor for each FET gate to reduce the Q of any parasitic elements.
Now however you definately will only use one of the FETs (the one with the Zener Drain-Gate) to limit the back EMF, even if the are well matched.
simulate this circuit – Schematic created using CircuitLab
