Sinewave Inverter Design

[Karakacha]

Iam working on a transistorized sinewave inverter using a tranistor oscillator, a buffer amplifier tranformer coupled o/p and a push pull o/p all transistors are BJT's. one may ask why?.... I'm just fed up with noise from rectangular and modified sinewave inverters.
May one assist me on how to wind a coupling tx and what determines o/p driving power.

What ever looks simple might be difficult....

[Eva]

Linear or class D?

Linear is a bad idea (too much power wasted), class D requires some effort to learn for anyone not experienced.

[Karakacha]

that,s why I wanted one with an idea to assist me......

[Eva]

Linear and Class-D are essentially the two most common ways to amplify a sinewave

[TechGuy]

Originally posted by Karakacha
Iam working on a transistorized sinewave inverter using a tranistor oscillator, a buffer amplifier tranformer coupled o/p and a push pull o/p all transistors are BJT's. one may ask why?.... I'm just fed up with noise from rectangular and modified sinewave inverters.
May one assist me on how to wind a coupling tx and what determines o/p driving power.

Your best option is to buy a True-sinewave inverter instead of building your own. The cost to design, build and test will greatly exceed the cost of buying one. Efficient True Sinewave inverters are tricky to design and build. Thats why they are expensive and the majority of inverters are squarewave\modified sinewave output.

[star882]

You can actually do quite well by taking a modified sine or square inverter and adding filtering to it. Use inductance first as connecting capacitance directly to the output really loads down the inverter.

You can also delta sigma a sine wave to generate signals for driving the H bridge. Then just add a simple LC filter to get a sine wave. (The interesting part is, that is very common in motor control, but there is no separate filter. The motor windings act as the filter.)

If you're looking for a cheap and easy way to get a sine wave inverter, find a discarded UPS with sine wave output (it's usually the batteries that are bad) and use it as an inverter.

[TechGuy]

Originally posted by star882
You can actually do quite well by taking a modified sine or square inverter and adding filtering to it. Use inductance first as connecting capacitance directly to the output really loads down the inverter.

You can also delta sigma a sine wave to generate signals for driving the H bridge. Then just add a simple LC filter to get a sine wave. (The interesting part is, that is very common in motor control, but there is no separate filter. The motor windings act as the filter.)

If you're looking for a cheap and easy way to get a sine wave inverter, find a discarded UPS with sine wave output (it's usually the batteries that are bad) and use it as an inverter.

Using an inductor to smooth a squarewave into a sinewave does work:

1. A large enough inductor will just limit the current flow. A small inductor will just round off the edges but it waveform will still be square. An inductor will also increase the noise because the inductor will release its energy each time the output switches. It will create voltage spikes (ie inductance leakage).

2. You would need a low pass filter consisting of an inductor and a capacitor. But this would also filter out 30 to 40% of the power to the output device. You would basically reduce a 1KVA inverter into a 700 to 600 VA inverter. All that filtered energy is also going to make the filter cap run hot (very short life before all the electrolyte evaporates).

If it was as simple as adding a LC filter to a squarewave inverter to make it into a sinewave, then all of the manufacturers would have done it.

[star882]

Originally posted by TechGuy



Using an inductor to smooth a squarewave into a sinewave does work:

1. A large enough inductor will just limit the current flow. A small inductor will just round off the edges but it waveform will still be square. An inductor will also increase the noise because the inductor will release its energy each time the output switches. It will create voltage spikes (ie inductance leakage).

2. You would need a low pass filter consisting of an inductor and a capacitor. But this would also filter out 30 to 40% of the power to the output device. You would basically reduce a 1KVA inverter into a 700 to 600 VA inverter. All that filtered energy is also going to make the filter cap run hot (very short life before all the electrolyte evaporates).

If it was as simple as adding a LC filter to a squarewave inverter to make it into a sinewave, then all of the manufacturers would have done it.

The main problem is that a suitable filter will be too bulky. The inductor will be rather big, and so will the capacitor. Electrolytics cannot be used anyways as it's AC - use motor run capacitors, which are even bulkier. If you put that into a modified sine inverter to make it a sine wave inverter, the resulting inverter will be much heavier and bigger. Not exactly desirable for a device that is often designed to be portable. And a small microcontroller operating as a DDS is cheaper than a large LC filter anyways and will make a better sine wave. (The newer UPSes use microcontrollers to generate PWMed sine waves or a series of pulses that approximate a sine wave better than a modified sine.)

Therefore, a good starting point would be to just connect some inductors (about 500uH to 1mH) in series with each line and connect a 1-5uF motor run capacitor in parallel after that. You won't get a pure sine but it might be close enough.

[ChrisA]

Originally posted by Karakacha
Iam working on a transistorized sinewave inverter using a tranistor oscillator, a buffer amplifier tranformer coupled o/p and a push pull o/p all transistors are BJT's. one may ask why?.... I'm just fed up with noise from rectangular and modified sinewave inverters.
May one assist me on how to wind a coupling tx and what determines o/p driving power.

What you will be buildig is a class D amp followed by a large low pass filter. Given a class D amp you don't even need to really create a small alalog sine wave. All you need is to synthisize the digital switching pattern to send to the switch transisors. Think of the signal as a stream of binay numbers with as many bits as you have switches
The output filter can be simple too because you already know to cut off frequency. OK that is one way.

For those who need even cleaner power, use a "motor generator". this is just what it sounds like. You connect the shafts of a motor and a generator with a flexible coupling and a big flywheel.

Ferroresonant Transformers provide another good way to filter the output of the class D amp. Basically what yuo do is rather thaen us a simple low pass you use a use an oscilator with a inductor and cap. It's a transformer and capasitor that are tunned for 60Hz.

None of these are easy to engineer if you need more than a few watts of power

[jackinnj]

Might I suggest that you look into the Linear Technology low-noise switcher chips -- like the LT3439, LT1533 etc. -- there are probably 5 0r 6 slew controlled chips by LLTC -- you can also do this with discrete devices if you stare at the product folders long enough.

...and they do live up to their descriptions