Full Bridge / Class D Gate drivers

[ElecDesigner]

Hi,

It seams quite common in these circuits to control the MOSFET Vds dV/dt by slowing down the leading edge gate drive on the complementary MOSFET. My understanding is that this is principally required due to the much reduced dV/dt rating of the MOSFET when the built in anti-parallel diode is in reverse recovery. On the falling edge gate drive there is no such limitation and a faster edge saves power.

Turning attention to the gate driver circuitry, it seams common to implement a series resistor (between the gate driver and the MOSFET), and a diode - cathode towards the driver (possibly with a smaller resistor in series).
I am looking into the details of a couple of existing company designs where quite a bulky schottky diodes have been used for this purpose, eg V8PAN50, 15MQ040N. These devices are several amp capable, but that is to be expected right? (gate drivers are several amp capable too).
Looking at the signals with a scope, it is quite clear that the capacitance of the diodes is having a large effect, effectively posing as a short circuit for about 3-5V of the rising edge (sharp edge then step then slower edge). Removing the diodes proved this is independent of and is having a much larger effect than hitting the miller plateau.
Has anyone else seen similar? Is a shottky the right choice here? What about standard fast recovery diodes (is there any disadvantages of using those)?

 

[FvM]

I agree with the analysis as far it's referring to high voltage transistors that have the problem of tight dV/dt constraints.

Asymmetrical gate drive through RD circuit makes only sense with integrated gate driver ICs, supplying gate currents up to 5 or maximum 10 A. Diodes must be capable to withstand the peak currents for a few 10 ns, an operation that's rarely specified in data sheets. V8PAN50 is just overkill, diodes with up to 1 or 2A continuous rating like 15MQ040N can be fine. Diode capacitances are probably only a fraction of Cgs in this range.

For higher gate currents, you'll use discrete driver transistors with asymmetrical series resistors, no need for diodes.

 

[ElecDesigner]

Your thought were the same as mine regarding the V8PAN50 but like you mention the high peak current for a few ns is not normally in the datasheets.

Looking at the MOSFET and 15MQ040N datasheets, you are right about the diode capacitance being a fraction of the MOSFET Cgs. The effect I am seeing however is real.
Any opinion on using fast recovery diodes instead of schottky - I'm tempted to try them.....

 

[mtwieg]

Hard to comment on your results when you don't show them, or specify your component values. But keep in mind that diode capacitance is highly nonlinear, and the singular value put at the top of the datasheet will underestimate the capacitance near zero volts.

More modern gate drive ICs have separate output pins for sourcing and sinking current, allowing you to independently control the rising and falling edges without using diodes.