another inverter project

Author Message

Joined: 09/08/2007
Location: Australia
Posts: 3338
Posted: 10:09pm 21 Jan 2020      

  wiseguy said  
I see the choke and transformer performing quite different jobs even though they share the same series current, the choke is essentially integrating the HF energy into a LF fundamental content of the PWM. The transformer then changes the LF voltage level to the desired voltage. I see the PWM driving squarewaves into the choke resulting in a sawtooth triangular current ramp (in the choke) as energy is briefly being stored and then returning that stored energy to the series choke/transformer circuit. The same 20kHz waveform applied to the transformer meets the reflected very low impedance of the tuning capacitor (resembling a short at 20kHz) so the resultant HF energy dissipation & storage at 20khz of the toroid is very low.

Yes you are quite right. The choke operates in a totally different mode because the inductance is "free" to store and release energy and allow the voltages on each end to be completely decoupled at high frequencies. Its a kind of series high frequency isolator.

The transformer is "nailed down" by the very tightly coupled secondary, and has to supply significant power into the inverter load, which effectively loads it right down.

But the fact is, the current through both is absolutely identical. Its the ampere turns through the wire that controls the flux in both magnetic cores, and also influences any skin effects in the wire. This current will be a 50Hz sinusoid with some slight imposed triangular ripple at the switching frequency. How much ripple amplitude there is depends on the total series inductance.

If the inductance is made high enough, the ripple amplitude can be made so low we can pretty much ignore it as far as core losses in both the transformer and choke are concerned.

The voltages across the choke and the transformer are an entirely different matter. High frequency voltages applied across a large inductance just see an open circuit (if the choke is operating below its parallel self resonant frequency).

I said it a year ago that for my modelling purposes the inverter really resembles an overgrown synchronous bipolar (2 quadrant) buck converter.  I dont agree that the magnetic effects of both will be the same though, the transformer has a tuned secondary (50/75Hz) with reflected impedance back to the primary, and the choke is just a choke.  The transformer is dealing with the resultant integrated HF energy ( a 50Hz fundamental) the choke is doing the integration work of the 20kHz HF PWM waveform.

Yup, agree totally.

A sine wave inverter is just a fancy buck regulator where the output bounces up and down like a yo-yo at 50Hz in two quadrants. A suitable choke is a vital part of the action in both cases, and its the ac ripple current component determines the core loss in the choke.

The only real difference is a dc buck regulator usually has a massive energy storage electrolytic across the dc output, so we can use a smaller less costly choke designed for a much larger ripple component. That choke will be more lossy so we need to use reasonable material.

We obviously cannot use a huge electrolytic in our sine wave inverter! so the choke needs to be made larger, and that comes with the added bonus of reducing the ripple current through the choke so we can get away with using an even more lossy type of choke without having it burn up.

Not trying to bait anyone, just trying to explain why the commercial inverters use steel cored chokes. I think most of us have grave robbed a dead grid tie inverter at some time, and salvaged the steel U cores from the original choke. It worked for them, and it will work perfectly well in your inverter too.

Switching power supply design, especially the design of the magnetics is a fascinating field of study, and something very few people know a lot about. Just trying to light one small candle to illuminate the darkness.

Why don't we see steel chokes used in commercial switching power supplies ?

Because its smaller and cheaper to use a big electrolytic for smoothing dc, and use a lower inductance choke with a powdered iron core.
Edited 2020-01-22 08:10 by Warpspeed
Cheers, Tony.