Class D - LC output filter - C ripple current

[Skeebopstop]

Hey all,

I'm designing in the 2nd order LC filter onto my class D design now and am unsure what level of ripple I should expect in the C. I don't want to heat it up and/or blow it.

It will be 200 Watts of output power, switching at 400kHz into a 4 Ohm speaker.

Any advice would be greatly appreciated.

Thanks

[wakibaki]

Have you read this?

http://sound.westhost.com/articles/pwm.htm#6

w

[Skeebopstop]

Thanks for the link. I have been following something very similar to that, however that link seems to summarize things up better. I'll add it into my handbook!

It 'nearly' answered my question. In the LC output filter, the C component is going to have to reactively store and re-supply the carrier frequency switching energy in order to provide the smooth output audio signal. So I was looking for how to approximate this energy in comparison to the signal.

Should I just assume that Class D is 90% efficient and therefore 10% of the switching energy is a good starting point of what the capacitor may need to store? It is really just ripple current I care of.

[wakibaki]

Originally posted by Skeebopstop
therefore 10% of the switching energy is a good starting point

This strikes me as a not unreasonable approach.

You could simulate your filter and speaker with downloadable software (pspice?) and put in a 400kHz square wave of variable duty cycle and ~80+ Vpp. You could use an ideal cap and inductor and put in real values for the series parasitic resistances. This would give you a way to approximate the answer or find a peak value without learning too much electronics.

w

[Skeebopstop]

I would rather learn the electronics behind it

I have a big tutorial on snubbers I need to read, perhaps it is buried in there somewhere but I just thought maybe there was a way to think of it in energies, and if so energy per carrier cycle time would be the power and at a given voltage one can derive the current.

Am I incorrect in focusing in on that ripple of the filter capacitor so much? Should I just stop being so anal and throw something in which 'should' do the trick?

[Eva]

The capacitance values required are usually between 0.47uF and 4.7uF. Electrolytics are ruled out due to high ESR. Films and ceramics have inherently high ripple current ratings.

You should focus more on inductor design and in the selection of the L and C values.

Anyway, you can calculate ripple current easily, you can assume that the capacitor is a short and calculate the amplitude of the resulting current triangle wave on the inductor. Then calculate the RMS value of a triangle wave (0.577 of amplitude).

[Skeebopstop]

Very intuitive solution, thank you very much!

[speakerguy79]

The use of a film cap is the ubiquitous choice for an LC output filter. Any mylar or polypropylene cap will be fine so long as it is rated for the correct voltage. Selecting inductors is much more critical.

The only time you should worry about ripple current is when you use an electrolytic as a DC blocking cap in half-bridge PWM amps.

[eclectic2k]

Originally posted by speakerguy79
The only time you should worry about ripple current is when you use an electrolytic as a DC blocking cap in half-bridge PWM amps.

I wouldn't say that's the ONLY time. Ripple current can be an issue when using a capacitor in a filter/storage capacitor as well. (energy is being transferred, current is flowing)

Though, yes, issue would mostly be with electrolytics. (due to the aforementioned higher ESR & Z) A properly selected film or ceramic should not have much issue.