Square wave inverter?

I am building an inverter that will use a square wave pulse to drive MOSFET's at 50-60Hz. I have read and heard a lot of stuff that running square wave into a transformer is not a good idea, what I want to know how bad is it?

Also will adding capacitors as seen in my image be able to smooth the square wave a bit to help the problem?
If so what capacitance should I use? (About 200A input)

(The 12V square wave is coming from paralleled MOSFET's)

Picture of Square wave inverter?
sort by: active | newest | oldest
Have you considered Optocouplers to drive your Mosfets?
Optocouplers protect the signal circuit from the power in the inverter.
If you would like to check out my Modified Sine Wave Signal Generator, it might be just what you are looking for.
iceng5 years ago
I would caution against putting capacitors as drawn.
They the capacitors will allow Very High ( Damaging ) Currents to flow
through your Power MOSFETs every cycle, Heating the FETs a lot faster
then the transformer.

No offense and in all friendliness, we just got you to abandon bipolar
devices in favor of MOSFET devices and now you launch into a new
scary circuit adventure.

I have built a Square wave inverter for a Neonatal transport that could
use power from 12VDC, 24VDC, 110VAC, 220VAC, stand-alone internal
batteries and 400cycle military aircraft. 
Among many sub systems being provided square wave power was a
Tektronix baby breathing monitor scope which worked perfectly on the
Square wave power. ( and to the pleased interest of the Tektronix Co. ).

Just build the inverter, you can do it twice as you gain experience.

If you are concerned about making a mistake, start by limiting the power
to your design by putting the best resistor series in this world
( a 40 Watt incandescent light bulb, then a 60W, then 100W, )
( then two 100W in parallel ) in series with your battery or DC power.
I use this whenever, I am not sure of the odd circuit that Im working on !


Most of the inexpensive inverters you can buy generate what they call a "modified sine wave". If you put a scope on the output, the output votlage waveform has the shape shown in the diagram.

A waveform that is symetrical about zero volts will have only odd harmonics. The most pronounced harmonic of the simple square wave will be the third harmonic. The benefit of using the modified sine wave waveform versus a simple square wave is the elimination of the third harmonic. This all has its roots in the Fourier Series representation of signals.

It would be beneficial to make the RMS value of the modified sine wave be the same as a sine wave. For the modified sine wave, the peak value should be 0.866 times the peak value of a sine wave. So, for 120 VRMS used in North America, the peak voltage of the sine wave is about 170 volts, and thus the peak value of the modified sine wave should be about 147 volts to have the same RMS value.

There are more complicated waveforms that can eliminate additional harmonics as well. Search on "Harmonic Cancellation" to find more on this. If you are using a microcontroller to generate the drive pulses, then it is easy to generate the more complicated waveforms needed to reduce harmonics.

If you use a different waveform then you will need to use a different factor to determine the peak voltage needed to keep the RMS value the same as that of a pure sine wave.
Modified Sine Wave.gif
Its bad, because it significantly increases the iron losses in the transformer, caused by the higher frequency harmonics. If you understand that, you will find that providing you are prepared to accept the transformer will give less power than it did on a sinewave input before it gets too hot to use, you can still carry on.

The MadScientist (author)  steveastrouk5 years ago
So basically it's simply less efficient and will get hot?
Yes., so you will have to derate the output by about 30 %