Mosfet driver options

[Pozar]

Hi!
I want to make D-class amp with aprox. 100..200W/8ohms and
I draw schematics with semi discrete triangle wave oscillator and
PWM modulator. Runnig frequency will be about 400kHz. I thinking to use IRF450 mosfets on end of amp.
But tricky thing is following:
do I use special IC like HIP drivers or discrete circuit with pulse ferrite core tranformers for galvanic isolaton.
Do any of You guys have some expirience with ferrite core
pulse transformers and their making?
Please help me on this matter....

[Kenshin]

ferrite core transformers can't transform the DC components in
PWM signal, so the level of gate driving signal will vary with the
duty cycle.

[Jaka Racman]

Hi,

here is one transformer coupled driver that restores DC level, so drive level is independent of duty cycle. It does not like sudden changes of duty cycle, which is true of all transformer coupled gate drivers (excluding drivers using pulse transformers that differentiate control signal). Because of emiter follower output it works even when primary is driven by 6 paralleled AC04 inverters.

Best regards,

Jaka Racman

[phase_accurate]

IR's AN 950 shows quite an elegant and simple solution for transformer coupled gate-drive which is capable of high duty cycles.

http://www.irf.com/technical-info/appnotes/an-950.pdf

OTOH an IR 2110 would still be cheaper than a pair of good impulse transformers.

Regards

Charles

[ekaerin]

yep, pulse x-formers can be tricky. But you could use one, then you
wont need to get a driver with level shift. However a single supply
driver will need ac coupling cap on the primary just like Jaka showed.

With 400k switching the core can be quite small, but make sure
that the volt-second product stays within limits.

Also be carefull when using IC drivers at those frequencies together
with large die FETs. I know they tend to get pretty warm at even
50k.
That is not only due to the load but also internal crossconduction
in the output satage of the IC.

/ Mattias

[tiki]

Hi,
there were already very interesting posts by JohnW and others, regarding output devices, and these: diyAudio Forums > Top >Amplifiers >Class D >Output devices?. This thread ended too early, unfortunately.
Do you think, you need couplers for that power stage really? It will complicate the design a lot.
You may also be interested in the Thunderball schematic by Felix Schill.
Good luck!
Regards, Timo

[markrav]

Hi Pozar,

I am using the HIP2106 it works at 1MHz, but 500KHz is better.
However, despite the spec sheet saying that it is ttl compat, I found that it needs at least 8-10v logic inputs, so I drive it with a
TC4427. This setup lets me drive it from 74HC logic, no problems at all, as long as you don't try to drive high continuously, as the bootstrap cap will run out of juice, so you must switch at least 10 times a second - which is not hard for class D

[Pozar]

Hi!

Thanx for a lot of tech info. I definetly want to use ferrite
transformers for galvanic isolation between driver (modulator)
and output stages because I wand to buid some kind of
prototype board with separate circuits for specified functions (one PCB for master clock, one PCB for modulator and one PCB for output stage with isolalating ferrite stage).

I have one qwestion regarding modulator solution:
on Yours opinion what is better
1. using PWM chip
or
2. using clasic book schematics with saw wave generators and
comparators.

With best regards....

[Kenshin]

Pozar:

self oscillation is perhaps the best,unless you can get some control chips dedicated to class D audio applications.

general PWM chips are for switching power supply use,they usually have a single-ended control method.

triangle wave modulator is something too complex.

you can search for SODA on this forum -- a simple & promising design.

[ekaerin]

Pozar,
I assume you don't need isolation for safety. You could try to
get some low current common mode filter inductors and try to
use as pulse transformer. I once did that on a prototype for a
Buck converter.

/ Mattias