Need Help: Class D Cross Over

[Genomerics]

Like what others have said the thing you are trying might be done in a better way...

However, having had a look at your circuits and being an old fart it reminded of this one.

http://focus.ti.com/lit/an/slua137/slua137.pdf

Yah yah yah, it is a DC motor controller and therefore plorp but the whizz gives a circuit good to a whole 30KHz that might plug in with what you are trying to do.

DNA

[munky]

Hello, THANKS for all the replies, I've tried to implement the circuit according to what you guys have told me. When I've added the resistors to the emitters of the transistors, the response is not squre anymore, will that cause a big problem? At least it improved the switching. I haven't tried the comparator idea yet, will do that very soon. Also, I made a very stupid mistake, I'm using a 4ohm speaker, so I should of put 8ohm load. Sorry for the confusion. By the way, for the project, I'm only allowed to use discrete components, so that's why I didnt have any high side/ low side driver ICs. Attached is the output with the resistors on the emitters. I will try the other ideas given to me really soon, and will keep you up to date. Thanks for the help, I appreciate it.

[munky]

Originally posted by mag
What is your PWM frequency?

The PWM frequency is 170kHz, so I just rounded it up to 200kHz in the hbridge circuit.

Originally posted by mag
use two comparator in place of U2E

I'm not too sure on how to put the comparators, so one is after the inverter, but what is it comparing with? Is the other comparator just right after the first one? Sorry, I'm a little slow on this.

[lumanauw]

You can put fast turnoff addition before mosfet gates (1 dioda+ 1transistor)

[mag]

Hi Munky,
look at my schematics.
It is just a suggestion, not a real schematics.

I have changed the mosfet gate drivers with push-pull BJT pairs to
have fast turn on and turn off of the mosfets.

If you are using a 4ohm speaker you should put a 4ohm load for simulations.
Every half of the H-bridge will see 2ohm impedance towards GND, so the output filter shall be designed coordingly. In my shematics you will find an output filter with 30khz cutoff when loaded with 2ohm (ie with a 4ohm speaker).

Two comparators are used to drive the two branches of the H-bridge, they receive the same input signal but two different triangle wave ramps. One of the two ramps is shifted trough a constant offset voltage (V_offset) allowing to implement dead-time.

The fact that now the waveform are no more square is probably because you are driving the mosfet gate too slow.

[mag]

Sorry for the mistake in the shematics,
of course you need to put an inverter between the lower left
mosfet and the upper right mosfet gate drive...
The same applies in the other direction.

[munky]

Hey munkey, please look at the following changes. Other members please let me know what you think?? I changed the values of the pull up resistors allowing for faster roll-off and added capacitors to slow down the response of the switching.

[Pafi]

As mag said, 2 uH is too small!

I don't know why did you change Rl to 8 ohm!

You don't have to slow down switching, it was already too slow!

I also don't know what do you want with these measurements. Where do you suppose to see the existence of dead time? And what do you measure actually? Y Axis is scaled in volts, and it seems to be real, but you have written I2_2 and I1_1 (currents) next to the curves.

Dead time is invisible with these measurements! Dead time means that Q1 have to switch off before Q2 switch on. You have to watch drain voltage of Q1 and gate voltage of Q2, or drain currents! Time relation between A and C is not really important!

[Pafi]

http://users.hszk.bme.hu/~sp215/elek...%20C%20MOS.GIF

[munky]

Hey, thanks for the site, I'm going to implement my circuit some more. I'm an idiot, making many stupid mistakes. Sorry about that, and thanks again.