another inverter project


Author Message
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 3338
Posted: 01:47am 21 Jan 2020      

  renewableMark said  
Poida, it's very unusual how you and Warp have such different views on iron cores, perhaps re visit the idea when you have the Mad power board operating.


Ferrite is definitely the thing to use for high frequency transformers and chokes that operate at both very high voltages and very high frequencies.

You will find that most commercial switching power supplies use a ferrite cored transformer, and a powdered iron toroid for the choke. Powdered iron is a lot cheaper, much more lossy, but can take a lot more ampere turns before saturation.
Its the preferred material for chokes that operate at low voltages and very high dc current, even at very high frequencies.

Now the twin evils we are trying to avoid in our choke are core saturation and excessive core heating.
Ferrite saturates at around 0.3 to 0.35 Teslas.
Powdered iron saturates at around 1 Tesla.
Silicon steel about roughly 2 Teslas plus.

So right away we see that steel can run about six times the dc ampere turns than with a ferrite core of the same cross sectional area. Ferrite is also very expensive in the larger sizes.

Now we come to core loss, especially core loss at high frequencies. Core loss is composed of two parts, hysteresis loss and eddy current loss. Hysteresis loss is mainly frequency dependent, but is also flux swing dependent. It rises very steeply with frequency.
Eddy current loss is also dependent on flux swing and frequency, and also rises steeply with frequency.

We can burn up a choke from just core temperature rise by having both a large ac flux swings combined with a high frequency.

Now a transformer works from flux swing to generate voltage in the secondary. That is inescapable.  So a high frequency transformer MUST have an efficient core to do what it must do.

On the other hand, a choke can be designed to have an absolutely minimal ac flux swing by using a relatively lot of turns, and a comparatively large core cross section.

Now all of this depends on the current flowing through the choke. If its pure dc, core loss will be zero. All the heating comes from resistive heating in the wire.

Imagine we had a choke of infinite inductance, the current would always be pure dc, no matter how high the applied ac voltage or the frequency.
If our choke has enough turns, and enough core cross section, the inductance will be high enough to create a high Xl or inductive reactance.

The ac ripple current could be very low, like 100mA rms with 50 amps of dc flowing.
In fact as the frequency rises the inductive reactance rises too, and that can greatly offset the relative rise in core loss with frequency.

Its perfectly reasonable to use a laminated steel core in a high frequency choke provided its large enough in core cross section and has ample turns. It would never work in a high frequency transformer though !!

There is another bad effect of high frequencies, skin effect in the wire.
This too is a magnetic effect where the alternating magnetic field forces current to flow along the surface of the wire.  As with the core, we can build a high frequency choke with solid wire because the ac ripple current is very low, low enough to flow along the surface while the bulk of the dc current uses the entire wire cross section.

Again, high frequency transformers are very different to high frequency chokes.

Transformer windings at very high frequencies are often thin foil, or multistranded litz wire. Chokes can work perfectly well with solid round wire, because the choke predominantly carries only dc.

Its the high frequency ripple current the choke is there to limit in the first place.
If its designed and working properly, there should be very little ac component in the dc current the choke carries, so high frequency effects in both the wire and the core are nowhere near as severe as in a transformer.

All the commercial inverters run steel or sometimes powdered iron chokes. Never ferrite its just not cost effective in the very large sizes that would be required for a high powered inverter. It would certainly work, especially if you can get the ferrite free.

But a steel choke will do the exact same job at far lower cost and far smaller size. And that is what you will usually find in the larger commercial inverters, and sometimes powdered iron in the smaller commercial inverters.

*edit*

Here is a further thought.

The choke is directly in series with the primary of the inverter transformer, and both carry the exact same identical current.  Why is it that the choke must have a ferrite core, while the main inverter transformer can get away with a steel core ?  
Remember the current, and the magnetic effects produced by that current in the respective windings and cores will be the same in both.
.
Edited 2020-01-21 11:59 by Warpspeed
Cheers, Tony.