Hello

I never succed with PIC programing, so I've assembled a schematic of a simple version pwm sine inverter.

Anyones can look at my schematic and tell me if it would work ok ?

Thank you

Bye

Gaetan

 

Hello

I've put a 1k resistor (from gate to ground), to remove gate charge of power mosfet´s.

Here's my new schematic.

Is my new schematic are ok ?

Thank

Bye

Gaetan

 

Don't think it will work for you Did you look at the data sheet and application notes for the SG3524? It has the oscillator built into it, the ZR2206 is not even needed, and if it was it would be the square wave out-put not the sine wave you would use.

The SG3524 is usually used as a DC-DC converter so you will get a square wave out-put not sine wave, even when running it through a transformer.

 

Hello

I have few SG3524 but no SG3525, that's why I use a SG3524 in my schematic.

Btw, there is a typo error in my schematic, the Ic should be SG3524.

Is the SG3525 would do the job ?

Thank

Bye

Gaetan

 

It is going to output square waves at the PWM chip's output rate (many kHz), and the amplitudes will be modulated by the sine wave. Its nothing like the output you expect.

 

What are the transformer ratings?

 

What are the transformer ratings?

Hello

Since the oscillating frequency will be arround 4khz, I would need to find a suitable 12/110 volt transformer.

Thank

Bye

Gaetan

 

Hello

I can't find any SG3525 here around, so I've redone my SG3524 schematic for a better turning off of the mosfets, as you suggested.

Is it better in my ver.3 schematic ?

I know that I won't get a very good sine at the transformer output, but maby adding a negative feedback from the transformer output to the SG3524 input will do a better sine, wich side of the transformer output I use for a nfb ?

Thank you

Bye

Gaetan

 

It is going to output square waves at the PWM chip's output rate (many kHz), and the amplitudes will be modulated by the sine wave. Its nothing like the output you expect.

Not so. The circuit is intended to modulate the duty-cycle, not the amplitude of the square-wave.

In principle, the circuit you posted should work by modulating the square-wave output pulse width at the sine wave frequency so that the average value, after being filtered by the transformer, will be the sine wave output. Whether you circuit values are correct, I don't have time for that analysis.

You may require some additional filtering, (inductive before or capacitor/inductive after the transformer) to minimize any residual square-wave signal in the output.

A 1kΩ resistor from gate to source on the transistors gives you a switching time that is a little long (I did a rough calculation of about 10μs) and that may give excess dissipation in the transistors. It's much better to add a push-pull
gate driver for that purpose.

Edit: The PNPs in the push-pull driver you show will not turn on since there is no negative voltage to do so.

 

Hello

So it's better to do biasing with a 10k resistor at the base of the bjt transistors.

Here's the new circuit.

Thank you

Bye

Gaetan

 

So it's better to biasing with a 10k resistor at the base of the bjt transistors.

You likely need less than a 1kΩ resistor to get reasonable switching speeds.

I would suggest that you use a Spice analog simulator to simulate at least the output stages of your circuit to see how they will operate at the speed you need. LTspice is a good free simulator from Linear Technology that you can download. If you don't simulate your circuit to determine the correct component values and configuration, it is unlikely that it will operate properly when you build it. Designing such a circuit without a simulator is rather like trying to build and test it without an oscilloscope.

If you do that we can help you through the simulation.

 

Hello

I have LTspice but I can't find a working model of the SG3524

Thank

Bye

Gaetan

 

I have LTspice but I can't find a working model of the SG3524

And there may not be one available. But you don't need that to test the output driver and output stage response to optimize the part values for that part of the circuit. That's the part of your circuit that needs work.

 

One thing you need to get rid of, are the 100nF capacitors on the mosfet gates. They are just adding to your gate switching time, since the gate is a capacitance on it's own.

 

Hello

I've get rid of the 100nF capacitors on the mosfet gates in my last version of my circuit schematic, and I ad a nfb to my circuit.

I can even use one of my big standard EI silicon steel transformer, but it would better to bring down my SG3524 oscillation frequency to 200 hz or 300 hz, because at higher oscillation frequency the Eddy current would be quite high in my output transformer.

At those low oscillation frequency (200 to 300 hz) the LC low pass filter would need a big L coil, but it would be easyer to keep my SG3524 circuit very stable.

Here is my my last version of my circuit schematic.

Thank

Gaetan

 

Not so. The circuit is intended to modulate the duty-cycle, not the amplitude of the square-wave.

In principle, the circuit you posted should work by modulating the square-wave output pulse width at the sine wave frequency so that the average value, after being filtered by the transformer, will be the sine wave output. Whether you circuit values are correct, I don't have time for that analysis.

You may require some additional filtering, (inductive before or capacitor/inductive after the transformer) to minimize any residual square-wave signal in the output.

A 1kΩ resistor from gate to source on the transistors gives you a switching time that is a little long (I did a rough calculation of about 10μs) and that may give excess dissipation in the transistors. It's much better to add a push-pull
gate driver for that purpose.

Edit: The PNPs in the push-pull driver you show will not turn on since there is no negative voltage to do so.

How do you figure the switching times?