0
\$\begingroup\$

Can somebody explain to me why we need MOSFET gate driver in class D amplifier like in the picture below(common class D schematics)?
What I don't understand is if we for example have 9V rails, then the comparator(LM311 for example) would output PWM waveform with 9V level. If we then use MOSFET doesn't it just becomes equivalent to wire? So, wouldn't it be the same if we just connect it neglecting MOSFET driver?
I understand the concept of MOSFET amplification if we have weak signals, but for me we just converting 9V pwm to the same 9V pwm here. I was trying to see how the current changes if we place MOSFET, but also don't see the difference...
One reason I can think of is that comparator has high output resistance which limits the current, I wonder if that is true, but still if we connect 8ohm load to it shouldn't matter so much.
class d block diagram
(source: soundonsound.com)

\$\endgroup\$
  • \$\begingroup\$ Because impedance matching is a thing. \$\endgroup\$ – Ignacio Vazquez-Abrams Jan 4 '15 at 0:14
  • \$\begingroup\$ But how do we change load impedance with mosfet? it's just a wire, comparator doesn't have more impedance on the output with mosfets \$\endgroup\$ – ScienceSamovar Jan 4 '15 at 0:16
  • \$\begingroup\$ What is MOSFET gate drive capability and why do I care about it? \$\endgroup\$ – Phil Frost Jan 4 '15 at 0:42
  • \$\begingroup\$ Some of your comments below lead me to ask - are you aware that a 'MOSFET' and a 'MOSFET Driver' are two different things? Which one are you actually asking about? \$\endgroup\$ – brhans Jan 5 '15 at 7:22
5
\$\begingroup\$

Power MOSFETs typically have a fairly large gate capacitance, thousands of pF. The output current drive of a "standard" comparator will be in the mA. Combine the two and you get a level transition of the gate voltage changing slowly, where slowly means less than tens of nanoseconds. Slow gate drive means slow switching of the MOSFET load, and that causes the MOSFET to heat up, which negates the value of the class D approach. A MOSFET driver, on the other hand, can typically source and sink pulse currents on the order of an amp. This provides the speed needed for class D to work properly.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I'm not sure what you are talking about in the first part, I suppose you described the situation with 1 mosfet? I talk about not using mosfets at all, since we already have pwm signal from comparator and why is it bad. \$\endgroup\$ – ScienceSamovar Jan 4 '15 at 0:50
  • 2
    \$\begingroup\$ @ScienceSamovar If you are wondering about why the MOSFETs are necessary at all, the reason is simple: this is an amplifier. The MOSFETs provide the capability to drive a lot of current, and that's what makes this thing an amplifier. You aren't going to run a 500W speaker directly from LM311. You need an amplifier. \$\endgroup\$ – Phil Frost Jan 4 '15 at 0:57
  • 1
    \$\begingroup\$ @ScienceSamovar A MOSFET is in no way a piece of wire. \$\endgroup\$ – Dzarda Jan 4 '15 at 2:53
  • 1
    \$\begingroup\$ @ScienceSamovar - that is only true once the gate has been "charged up". Until then, it looks to the comparator like a fairly large (several thousand pF) capacitor. And it is the time it takes to charge up the gate which determines how fast it switches. Regular comparators simply won't do the job - their output impedances are several orders of magnitude too large. \$\endgroup\$ – WhatRoughBeast Jan 4 '15 at 3:47
  • 1
    \$\begingroup\$ @ScienceSamovar - In addition, "in ON state it makes no load on comparator(resistance and reactance is negligible)" is true but irrelevant. The RESISTANCE of a MOSFET gate is very high, but since it behaves like a capacitor its REACTANCE is frequency-dependent. That is, the reactance is very high ("negligible", in your terms) at very low frequencies and DC, but very low during level transitions, which are high-frequency. And that's why you need a low-impedance driver. A comparator will work fine for very low frequencies, but not at the >100 kHz needed for class D audio. \$\endgroup\$ – WhatRoughBeast Jan 4 '15 at 13:56
3
\$\begingroup\$

It's true that the comparator has a relatively low current capability relative to the capacitance of the MOSFET, which would lead to slow switching times. See What is MOSFET gate drive capability and why do I care about it? This is one reason you will see gate drivers in class D amplifiers (or really any application involving power MOSFETs).

There is another reason. If you look at the schematic you posted, both of the MOSFETs are N-channel devices. This means to turn on the top one, something needs to generate a voltage that floats with the output, because the source of the top MOSFET is connected to the output. If we didn't have this voltage, we'd have no suitable way to drive the gate of the top transistor.

One way to get around this is to use a P-channel MOSFET for the high side. However, this introduces two problems:

  • the two MOSFETs are now not identical, potentially introducing distortion into the output.
  • all else equal, P-channel MOSFETs have a higher on resistance than N-channel MOSFETs. The reason has to do with semiconductor physics and the relative mobility of holes vs. electrons.

Many integrated half-bridge drivers are available which include a charge pump to generate that floating gate drive voltage, and all the level shifting logic to make interfacing with them trivial.

There are other subtle problems with half-bridges that integrated gate drivers will usually help solve. One is that you never want both transistors to be on at once, as it would effectively short the power supply. This is called shoot-through. Integrated gate drivers usually include logic which makes this impossible, or at least they guarantee equal propagation delay and gate drive current on the high and low sides which makes life easier.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Removed - bad fingers. \$\endgroup\$ – WhatRoughBeast Jan 4 '15 at 4:53
  • \$\begingroup\$ You might want to edit to change "subtle problems with half-bridges" to "subtle problems with any bridge configuration", or something like. After all, shoot-through is just as big a problem for full bridges. \$\endgroup\$ – WhatRoughBeast Jan 4 '15 at 4:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.