I have been writing this in the past tense because this inverter modification project has been happening over the last 6 months not the last 6 days...

It's been abit of a marathon project, but i've working on getting it all working properly before publicising any info...

Good guess Tracker - but the inductors in the last photo are actually the final ones.

Also regarding just reducing the 4.7uF cap, by all means try it, but i don't think no load current will drop that much as poor design of the original transformers mean the magnetising inductance is not enough and they draw to much current...

To fill in the background:

I worked out by experimentation and calculation that approximately 60uH would do the trick in terms of filtering the switching waveform before the transformer.

My first attempt was actually this:

3x 2.5mm diameter enamled copper wire wound on an F6 neosid ferrite, which is a low Mu ferrite that is specifically used for filtering and power suppression applications. (We use them at work for loudspeaker crossover inductors)

Actually this worked fine with the reduced 0.47uF capacitor the inverter could easily carry a 1kW load without any problems and the no-load current was about 0.34A (approx 18W)

The next refinement was to try a 110uH inductor on each of the 30V 50A winding which would reduce the current and losses thorugh each inductor (approx 2x 50uH)

Such that now i could use the inverter to charge the battery at full power (70A) using 3100W from the grid supply.

Actually these inductors were not too bad. However they buzzed quite badly at full power and because the magnetic circuit is open on these inductors the radiated field was quite large and the metal case was would affect the inductance, vibrate and buzz ect

The resistance of these inductors in critical as it directly eats into the voltage delivered to the transfomer reducing the headroom available to the inverter to regulate it' output voltage as covered previously.

Also the inductive reactance at the fumdemental frequency of 50Hz also does a similar thing but slightly differently

Z = 2 x PI x F x L

= 2 x 3.1416 x 50 x 50E-6

= 0.0157 Ohms

The voltage across the inductor at full power is then V= I x Z

= 0.0157 x 50

= 0.785V

Howvwer this voltage is at 90 degrees to the voltage across the transformer and has to be added as a vector wise

Pythagoras theorem http://en.wikipedia.org/wiki/Pythagorean_theorem another brilliant mathematician from another time...

= sqrt (30^2 + 0.785^2)

= 30.0102

So this doesn't really effect the voltage overhead too much

Anyway wondering where i could find some 50A inductors that were appropriate i found this site:

http://www.coilws.com/index.php?main_page=index&cPath=208_212_229_113

Which are designed for power applications like welding output ect using the similar types of low Mu ferrites that specifica to these applications.

At the spec'd resistance of (only) 4 miliohms will have a voltage drop of 50 x 0.004 = 0.2V at full power which OK and won't compromise the inverter's ability to regulate it's output voltage

I ordered 2x of the 110uH 60A units part No:ES55246-111M-60AH

Datasheet:http://www.coilws.com/images/ES55246-111M-60AH%20Rev%20B-Web.pdf

These work superbly allowing the toroidal transformer to perform correctly, i have made an aluminium bracket to hold them in place to ensure there is no magnetic interaction with the inductor

The last piece of the puzzle is some final output filtering, as any additional capacitance place on the output (say the power factor correction capacitor in a fluro lamp) of the inverter would serve to short circuit the switching harmonics increasing the current draw

Basically a 0.82mH inductor in series with the output and another 0.47uF capacitor across the output form a filter whose pole is tuned BELOW the switching frequency of 20kHz.

F = 1/(2 x PI x sqrt (L x C))

= 1/ 2 x 3.1416 x sqrt (0.82E-3 x 0.47E-6)

= 8.1kHz

This is very important, if the filter happened to be tuned near the switching frequency large currents would flow, large voltages appear across the filter, and nasty things might happen

With this filter any additional capacitance or inductance placed on the output of the inverter will mean the filter pole is moved down increasing the filtering effect.

In other words, at the switching frequency the filter remains inductive blocking any switching harmonics and delivering a smooth sine wave!!

The 0.82mH inductor is another Harbuch product, actually a 1.6mH filter inductor wound for use in a lighting dimmer rack, that i unwound a few turns off to get the desired value.

This only has to handle 12.5A which is max output current of the inverter at 240VAC

Lastly Grounding the negative terminal of the DC power connection to the metal case and adding ferrite chokes to the mains output leads keep the radio interferance under control, which is actually quite bad without those measures.

Here's a pic of the approximately finished unit:

The last checks i did on this unit was to load test the unit with different loads:

1500W heater - OK

700W transfer pump - OK

Lastly our 2600W Grunfos CR4 high head pump located at the dam down 100m of 6mm2 cable, this is a SERIOUSLY taxing load, the 2 pole induction motor uses 12A when runnning but least 3X the current is required to start it.

When i switched on the overload light flashed and the inverter beeped but it just kept going, let it go for a couple of minutes and all seemed to work fine, although the fan on the MOSFET heastsinks turned on the keep things cool.

So you can't argue with that

The final result is good all round:

Cost: Approx $1800

Loads: 3kVA + more short term

No load power: less than 20W

Here a pic of inside a LATRONICS PVE 1200 (my grid tie inverter) that uses a toroidal transformer.

Guess what that other small looking toroid is below the main transformer is?? (filter inductor)!!

I'd done all this modification development on the POWER STAR W7 before i brought the PVE1200, when i took the lid off i was veeeery happy to know that i was on the right track...

Basically a now V12 motor that idles using only 0.2L of fuel an hour but can deliver 3000hp when required.

Enjoy

Andrew

PS: i might try to put together a more concise list of the changes so that anyone can follow them