another inverter project


Author Message
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 3338
Posted: 10:20pm 20 Jan 2020      

A few points to ponder...

If you connect up a pwm inverter with little or no series choke, the transformer voltage waveform will follow fairly closely the desired sine wave shape. The output waveform should look quite reasonable.

The big disadvantage of this will be the HUGE peak currents through the mosfet bridge. Every time it switches it has to charge and discharge all the capacitance of the transformer and any additional capacitance that you have connected across the secondary, through a very low series impedance.

While this can and often does work, it will be very stressful for the mosfets.
So people say "never fitted a choke and my inverter works fine". I really cannot argue with that.

As you add more series choke inductance, the switching peak current spikes greatly reduce, and the mosfets work under much less stress.  But adding series impedance between the switching bridge and the transformer tends to decouple the transformer from the pwm drive waveform slightly.

The transformer then tends to do its own thing with regard to any self resonances or core saturation problems. So we may start to see wiggles and bumps on our waveform.

The natural reaction to that is that the big new high inductance choke is bad because it creates all this new waveform distortion, and its not an improvement. We have fixed one problem (the huge destructive mosfet killing current spikes) and replaced it with a different problem (waveform distortion).

The solution is to keep the big choke and fix the problems that the transformer is creating at the transformer.

The biggest problem with the transformer is the high voltage secondary winding which has a lot of capacitance down to the steel core because its one very wide layer. It also has a very high inductance, due to the massive core size and the high permeability of the really good grain oriented steel.

This results in a low self resonant frequency in the low Khz region, something we can do nothing about. Harmonic distortion in our 50Hz can excite this resonance and create stationary ripples in the waveform.  The more choke we add, the higher the amplitude of these ripples become.

The solution surprisingly is to deliberately add more capacitance across the secondary and lower this resonance. The lower we go, the better the waveform becomes, but this introduces yet another problem !!

This all assumes that the series choke is large enough to allow the adding of this extra extra capacitance. Without the choke, extra capacitance might kill our mosfets from lethal current spikes.

Anyhow, as we tune our transformer resonance closer to 50 Hz, the wiggles go away, but we might find that the voltage feedback now has problems controlling the output voltage at light loads. There can be a massive buildup of resonant energy which can cause the voltage to run away beyond the ability of the feedback to control.

There is a solution to that problem too...  If the transformer is deliberately tuned to exactly 1.5 times the inverter frequency, any tendency to build up resonant energy is defeated by the stored energy being out of phase.  Its a kind of magic self damping effect.  So tune your transformer secondary to 75Hz +/- 1Hz (90Hz in America) and you will have a very clean waveform.

There is another issue that is completely separate. Some people have observed a kink in the waveform at or soon after the zero crossing, but in one direction only. This appears to be transformer staircase saturation caused by the negative and positive half cycles not being of identical area. It could be due to many things, unequal pwm due to unequal delay in the gate drive circuitry, faulty pwm waveform due to software, or even unequal rds on of the upper and lower mosfets. The flux in the transformer ratchets up to saturation in one direction.

Maximum flux is at the zero crossing point, not the voltage peak, so we see the effect at or soon after the zero crossing.  There is not a lot you can do about it, so unless there is something seriously wrong, it can probably be ignored.

So in summary:

1/ fit a big high inductance non saturating choke, which will probably introduce some distortion.

2/ tune your transformer to exactly 75Hz (90Hz in the US) to get rid of the distortion.
Cheers, Tony.