DC-AC FULL BRIDGE TOPOLOGY INVERTER

Plan:
Simulation and optimization for efficiency in LT SPICE
Make a working prototype of the whole inverter first, in a small scale
Add extra features like producing variable frequency

Achieved:
Simulation of the circuit in LT SPICE
A working prototype of one stage of the inverter in a small scale


I worked on a stand-alone DC-AC full bridge topology inverter for my final project. There were few topologies to chose from, but since I was working with high power, I planned to use a two stage DC-AC full bridge inverter consisting of a DC-DC converter stage, and then a DC-AC inverter stage. The inverter would take in a 12 V input, and would give 120 Vrms, with 15 A of Irms, at 60 Hz of frequency, with 1.8 kW of power. Basically I wanted to be able to drive any AC load using this inverter. First of all, I wanted to make a running simulation of the full-bridge inverter, and then work on designing an inverter that had an efficiency of 90 %.

The above is a simple block diagram of my two stage inverter. At the end of the DC-DC full bridge converter, I wanted a DC voltage of 170 V, which is equivalent to 120 Vrms, and then I would use this to generate AC voltage at the load. So, first of all, I worked on the individual stage, and designed them accordingly in LTSpice, and at the end combined both of the two stages together at the end. Below are some simulation results that I got from individual stages and the whole inverter.

fig. spice simulation of the DC-DC full bridge converter stage 

fig. DC-DC full bridge converter

fig. spice simulation of the DC-AC inverter stage

. DC-AC full bridge inverter

fig. spice simulation of the whole DC-AC inverter

fig. the whole DC-AC inverter 

At the end of the simulation, I expected a Vp of 170 V, however, my spice designed suffered in efficiency from choosing very large inductor values for transformers. This was one of the things that I learned, since in spice simulation, I had thought only the turns ratio mattered, however, inductor sizes affected the efficiency. Also, my DC-AC inverter, might not have been running in full duty cycle, which probably has to do with the PWM, I was generating with the triangle and sine wave while driving the MOSFETS. These two things are what I plan to work more in the future.

I was also able to make a very crude working prototype of the DC-AC inverter stage using AVR micro-controller. For this, the issue was to figure out how to generate 60 Hz, by using a PWM. So, for each alternate diagonal pair of MOSFETS, I needed to generate a PWM, and the different points of the sine wave would correspond to different duty cycles of the PWM. I found a sine wave table on the internet, which I used in my code, to assign its values to different PWM levels. One issue with that, because I had 256 samples in the sine wave table, and the frequency of the PWM being generated was very limited by the prescaling factors of the pins, I was able to generate a 22 Hz sine wave, not 60 Hz. One way to get around it, was to store larger number of samples, since f(sinewave) = f(of PWM) / no. of samples. I calculated to use 766 samples in the sine wave but didn't have enough time to complete it. However, by using a 2V of DC supply, I was able to generate a 1.7 Vpp of AC(because I clamped or changed my duty cycle so as not to go all the way upto 0 or 256). I used the half bridge modules that we had in the lab, and 10mH(but with very small current capacity) and 4.7uF capacitor to filter it. So, at the end, across a 10 ohms of 1/4 watts resistor load, I was able to get a AC sine wave, that looked like the following:

                              

fig. AC sine wave generated from the 2V power supply

I think moving forward, I would first of all like to work on my simulation and optimize for efficiency, by first using smaller inductor for my transformer, and fix my DC-AC inverter stage as well. Then, I would probably want to design a transformer that can handle very large current, since the designs that I found on the internet, were limited by their current capacity. Also, using a FPGA board would be a good idea to implement better features in the inverter. To comment more on this project, this was a great learning experience in using LT Spice and coding the AVR micro controller, personally.