Why does my Inductor Get Warm? (Inverter Output Filter)

Question

I have a H-Bridge based Inverter that modulates a 350VDC source into 220V RMS @ 50 Hz. Right now I'm working at just 70VDC but my output inductor is getting quite warm. Here is a schematic:

The output inductors, L1 and L2, both get quite warm when the DC source is at 70VDC. I haven't gathered the courage to go higher voltages. The switching frequency for my PWM is 16KHz. I have no load on the inverter. By adding a 1.95 Ohms shunt resistor I was able to measure the RMS (I assume mostly reactive) current, which was approximately 400mA. Note that the output waveform looks extremely good and relatively free of ripple, even under some load.

The particular inductor I'm using is 2320-V-RC. The DC Current rating is 3.8A. I don't have enough experience with filters meant for power to know if I should be using a different type of filter or a different sized inductor.

Answer 1

Almost certainly your inductor core is saturating. The currents into the 6.5 uF will be quite high and this is the reason. You have basically a series tuned circuit formed by 2x 470uH and a 6.5uF capacitor.

This will have a resonant frequency of: -

f = \$\dfrac{1}{2\pi\sqrt{LC}} = \dfrac{1}{2\pi\sqrt{0.001\times 6.5\times 10^{-6}}}\$ =1974 Hz.

Ideally, you want the resonant frequency to be logarithmically half way between 50 Hz and 16 kHz and this would be more like 894 Hz so maybe you are a little close to the switching frequency. You could probably go lower - maybe towards 300Hz to improve this but, all the time the inductor or the cap is getting bigger.

Basically if you could raise your PWM frequency by 3 or 4 times you'd see an improvement I reckon.

Answer 2

First of all, note that that family of inductors has their current rating set by a 30°C temperature rise, and that they also lose nearly half their inductance at that current. I think that for this application, you're going to want much beefier inductors to start with.

As far as the filter design itself, you basically want the inductors to block as much 16 kHz energy as possible, while having a minimal impact at 50 Hz. For example, if you made them 1500 µH each, they would add less than 1 Ω of total reactance at 50 Hz, but would have an impedance of about 300 Ω at 16 kHz, passing only about 733 mA(RMS) at the full voltage. I just pulled the 1 Ω number out of the air; the key point is that you want this value to be less than the effective source resistance of the H bridge, but about the same order of magnitude.

But the key point is, the inductors need to maintain this level of inductance even while they're carrying the peak value of the 50 Hz load current plus the peak switching current, and it is going to require a physically large inductor to accomplish this.