Inverter Output Filter Design

[alisarhangpour]

Is there anything on the net, like a rule of thumb for designing output LC filter of a Power Inverter?
I'm wondering why I can't find something useful. Any kind of detailed approach is appreciated as well.

 

[Hero999]

That's such a vague question it's impossible to give and answer.

Are you talking about an inverter with a class D ouput stage? It will depend on the switching frequency.

 

[alisarhangpour]

No, I think you're talking about communication circuits.
The inverter I'm talking about is simply an inverter that inverts DC to AC, using PWM or so.
there should be millions of them all around the world. in UPS', etc.

 

[Nigel Goodwin]

No, I think you're talking about communication circuits.
The inverter I'm talking about is simply an inverter that inverts DC to AC, using PWM or so.
there should be millions of them all around the world. in UPS', etc.

No, he's talking about inverters - there are various different ways they work.

 

[evandude]

What Hero999 said makes perfect sense to me. It was fairly obvious you were referring to a DC-AC power inverter, and some inverters use a class D output stage.

Once again, it's hard to give a "rule of thumb" because the LC filter will depend on the circuit and the switching frequency. Presumably, anyone with enough knowledge to actually build an inverter should have no trouble designing such a filter for its output...

 

[Hero999]

No, I think you're talking about communication circuits.

No, I wasn't.

The inverter I'm talking about is simply an inverter that inverts DC to AC, using PWM or so.

That's exactly what a class d output stage is and it is what I'm talking about too.

there should be millions of them all around the world. in UPS', etc.

The answer to your question will depend on the switching frequency, current, output frequency and maximum and minimum duty cycle.

Here's a simplified schematic of a typical class D output stage in an inverter.

**broken link removed**.

 

[mneary]

I've taken apart a few of the more modern (lightweight) class-D inverters, and now they boost the incoming power to a DC voltage of (Vout x 1.5) and then use a class-D output stage.

I haven't seen a heavy 50/60 Hz transformer in about 10 years!

 

[Hero999]

I've taken apart a few of the more modern (lightweight) class-D inverters, and now they boost the incoming power to a DC voltage of (Vout x 1.5) and then use a class-D output stage.

I've seen those too.

I haven't seen a heavy 50/60 Hz transformer in about 10 years!

The transformer in my sketch design is a high frequency transformer. My idea is to integrate the step up function into the class d amplifier.

 

[alisarhangpour]

...

The answer to your question will depend on the switching frequency, current, output frequency and maximum and minimum duty cycle.

Here's a simplified schematic of a typical class D output stage in an inverter.

**broken link removed**.

The thing you're calling a "Class D" inverter is simply a Single-Phase Full-Bridge inverter. Of course I know what a Full-Bridge is and how it works. thanks for the diagram, it helped me to learn the other name of Full-Bridge.

I don't want exact figures for a certain inverter and of course, I know it depends on these parameters, I want to know HOW it depends.

 

[Styx]

Yes a class-D aplifier is effectivly a full H-bridge inverter, but the same rules apply.

Both will need to reject the switching frequency
* for an audio Class-D anywhere from 100kHz and higher
* while a power inverter would have around 10khz [dependant on power level).

Both have a low-frequency cutoff
*in the case of the class-D it will be ~30khz to reject switching and other components polluting the audio
* the power one will have one to reject freq higher then the fundemental (say a cutoff around 100Hz).

Both have a output quality to meet


So it's not really wrong for someone to post what is done for an audio amp which only differs from the power counterpart by just specifics.

The exact same design requirement's for the output filter applies to both, just numbers are different.

With the 1/2*pi*sqrt(LC) equation you can find out the L*C value for the cutoff you want then its all about power really.
The filter could be made up of really high-L and really low-C, but then you just won't be able to get much current through the inductor.
Likewise a really low-L and a really high-C will effectlivly do the same thing but from a voltage P.O.V. good old reactance

truth be told simulation is the quickest solution to this. try some values, sim it and see how it responds to static and step-changes in load. where needed re-ballance the L and C.

Then have a look at the output voltage & current quality to see if it meets what you want.

one piece of advice I would suggest a 2-stage filter (if you plan to draw any significant current that is). The first is tuned to really reject the switching-frequency (the 2nd major frequency component in the output spectrum), then do the one for the power fundemental.

This way the power filter doesn't have to dissipate the HF associated with the switching freq, and thus stop the inductor-core from heating up un-nessasarily (plus you cna make that switching filter really tight at that freq)

 

[alisarhangpour]

Thank you Styx, that reply should have taken you a considerable time. But I'm still surprised I can't find a lot on this topic, on the net.
Tough I should confess I'm not a talented Googler!

 

[notreally]

Dear everyone,

I also have problem with designing output filter for a power inverter.
The case is: PIC produces 40kHz PWM and the PWM is modulated by the required 400Hz sine wave. The PWM goes to a FET-bridge (IRFP460) through an IR2184. After the FET bridge we have the amplified PWM with 320V apmlitude. Then I want to get the 400Hz sine back. I think, an L-C filter could be good, but I have problems with it:
1.: the filter choke heats a lot. Styx wrote that a 2-stage filter could fix this problem, but I dont really understand this part of his post. May I make the first stage to filter at approx. 20kHz and than the second stage at 4kHz for exmaple? How will I know which is the "2nd major frequency component in the output spectrum"?
2.: there is a huge voltage drop on the filter. That means: only the 80% of the FET-bridge's output amplitude reaches the device I want to supply. This device is an old radar transceiver, so I dont know actually the impadance of it, but I guess, the best will be to make the filter's impedance as small as it is possible. But how? I mean how the filter "interferences" with the impedance of the supplied device? (At the input of the radar transceiver it starts with inductances)

Please help if have any idea (the best help with be something with concrete or approx. values)
Thanks,
Adam Pinter (adam@icicom.hu)