E3 - 3 Phase Inverter

Introduction: Experiment 3 is a detailed analysis of a full-bridge inverter using power MOSFET as the switching devices. Both square wave inverter and sinusoidally modulated pulse width modulator inverter(with unipolar switching also) were investigated. The purpose of an inverter is to take a DC supply and produce a AC current or voltage waveform. Inverters find common use in backup power supplies and motor control.
The following experiments have introduced inverter circuits and the effects of modulation frequencies and cross-over protection scheme on the inverters output wave.

Basic Operation: Inverters essentially take a DC source and switch it across a load in alternating polarities. Half-bridge using a centre tap supply and full-bridge in a similar construction as to a full-bridge rectifier. Since a DC source is being switched across a load, achieving a sinusoidal waveform is unlikely, but near sinusoidal characteristics can be achieved.
Special switching techniques such as SPWM can also be employed to control the output voltage of the inverter, which finds great use in motor control. SPWM switching involves using a saw tooth waveform to cut a sinusoidal of the desired frequency, where the two waves intersect triggers and releases the modulator, which in turn switches the transistors. Changes in the saw tooths amplitude change the number of intersections between the waveforms and subsequently the switches on time, which has a direct relationship to the output voltage. The longer the switches are on the greater the output voltage. Higher saw tooth frequencies will result in smoother approximations to sine waves and will shift harmonics to higher frequencies and incur greater switching losses.

Experimental Notes: Notes and figures taken during the experiment have been included at the end of the report.
Report Questions:
Answers and findings requested for the report are as follows.
5.1
Measured Values
Calculated Using E2
Vin = 146V
vout = 145v
Vout = 131
Iin = 3.33A
iout = 3.89i
iout = 3.5i

The measured and calculated values are different with the measure voltage reading much higher than the expected value. One obviously explanation for this is that the voltage meter provided to take measurements with was not infact a True RMS meter and was assuming a sinusoidal waveforms in its reading. When in reality the output waveform more closely resembled a square wave.


The harmonics in the above chart are a few amperes below the fundamental frequency, but a value of 1/3 of the fundamental for the 1st order harmonic is still a substantial contribution and most unlikely unacceptable in real applications.

5.2



Waveform 1: Vout
Waveform 2: Load current
Waveform 3: MOSFET current
Waveform 4: Energy return diode current
When the MOSFET in W3 is switched on negative current flows through the load. During switch off the inductor forces current to continue flowing at which stage the energy return diode, from the opposing MOSFET, carries the current until the next cycle. This behaviour is reflected during the next cycle by the next pair of MOSFET/diodes and is what give the current its shark fin shape.
The energy return diodes allow the stored energy in the inductor to return to the supply and prevent large transient voltages from damaging the MOSFETs. They also reverse bias the opposing MOSFET that is next to conduct and hence carry the inductor current for a short period of every cycle.

5.3
Increasing the switching frequency of the carrier wave made no changes to the harmonics around the fundamental frequency. It is however expected that the higher order harmonics were infact shifted by multiples of the carrier frequency. Unfortunately FFTs over a wide enough window were not recorded.
The following values were recorded for both 1khz and 10khz carrier frequencies and would be expected to be reflected around multiples of the carrier frequencies with reduced magnitudes.

The illustration above depicts the lower order harmonics, an almost identical FFT was produced for the 10KHz case.
5.4
Increasing the dead-time whilst fixing the switching frequency resulted in distortion of the current wave around zero, as depicted below, which is expected as some MOSFETS remained off for a longer period. Output voltage was also slightly effected due to short on time across the load.
Also increasing the dead-time counter acts the benefits produced by increasing the carrier frequency, raising the order of principle harmonics, by introducing low frequency subharmonics. A harmonic around 30Hz was introduced. Unfortunately no other measurable change was recorded and similar waveforms as to those reported in 5.3 were found.

The illustration above demonstrates the distortion around zero that become apparent as the dead time is increased.

5.5
The diodes need to be rated to carry the load current and a reverse breakdown voltage higher than that of the input DC voltage.
By using a dead-time to ensure that the MOSFETS are switched correctly, avoiding short circuiting the DC source, a period of time is created during which the DC supplies zero current. Attempting to reduce the supply current to zero and the inductances inherit to all wires will result in a large voltage generated at the top of MOSFET 1 & 3 and negative at the bottom of 2 &4. This could result in damaging the MOSFET and/or the diodes. One possible protection mechanism would be to place a clamping zener diode across the input to the inverter to ensure that the input voltage remains within an acceptable range, and send all others to ground, there by protecting the inverter.
5.6
Varying the depth of modulation will result in variation of the output voltage.


5.7
A unipolar scheme allows the stored energy in the load to continue circulating through the load, using the diode and 1 MOSFET, rather than returning to the power supply. This is achieve by switching off the MOSFETs in stages, leaving 1 leg to circulate the current. Uni-polar designs reduce the output voltage harmonics(it will not contain even harmonics), as current is allowed to flow through load and reduce switching losses.

Conclusion:
As observed square wave inverters have large harmonic components which would limit their use with out higher order filters to remove such noise.
A SPWM inverter on the other hand has reduced harmonics, as it shifts higher order harmonics by multiples of the carrier frequency. This allows designers to use lower order filters effectively as the harmonics can be shifted orders magnitudes away. The cost of higher carrier wave frequencies is increased switching losses.
Dead-times are dependent upon the switching times of the MOSFETs and should be as low as possible to reduce the introduction of harmonics components around the output wave. A uni-polar switching scheme can result in reduced harmonics and switching losses.