Project 2: Open Loop H-Bridge Inverter Design Using Traditional PWM Method and One Cycle Control
Project 2: Open Loop H-Bridge Inverter Design Using Traditional PWM Method and One Cycle Control
Project 2
Open Loop H-Bridge Inverter Design using
Traditional PWM Method
and
One Cycle Control
Problem Statement.............................................................................................................. 3
Conclusion ......................................................................................................................... 17
3
H-Bridge Specifications
Switching Input Voltage Output Voltage Output Output Power
Frequency Frequency
fs = 20kHz VG = 300VDC V = 100VRMS 60Hz POUT = 1kW
(Ts = 50µs) (Vpeak =
141.41V)
The four switches have the non-idealities: on-resistance, RON = 25mΩ, and a body diode
(though VON is neglected).
An H-bridge is like two Buck converters: One converter operates during the positive half-
cycle, and the second during the negative half-cycle. Therefore the H-bridge inductor
current and output voltage ripple should be similar to that of a Buck converter.
V 100
For an H-bridge: V = VG*d during (+) and V = -VG*d during (-). ∴ d = = = 0.33
VG 300
Simulation output filter parameters:
Inductance was desired to be large, but not too large. L = 5mH, with resulting inductor
current ripple of 33.3%. Output voltage ripple was held much tighter at 2.1% using C =
100µF. Simulation took place at full load.
Special Explanation: The reason we have two separate references, one for (+) and the
other for (-), is because of ORCAD Lite convergence issues; That is, adding inverters to
two of the gates inputs resulted in convergence problems.
5
Driving Signals:
The driving signals Mpos and Mneg are generated by PWM. To be specific, we compare s
sawtooth signal with the sinusoidal signal using an Op Amp to create each driving signal of
the four switches. The resulting driving signals, Mpos and Mneg, will have a duty cyle directly
related to the amplitude of the sinusoidal input.
As one can see from the image above, the two control signals, Mpos and Mneg, are
complimentary to each other at all times. Thus, switch S1 and S2 would never be ON at the
same time, so the input voltage would not be shorted.
The output signal tracks the ideal sinusoid well with an addition of a phase delay.
6
The table indicates that in the positive half cycle, switch M3 is always ON and M4 always
OFF. M2 is the same as duty ratio (d2), and M1 is the inverted value of the duty ratio d2.
During the negative half cycle, M1 is always OFF and M2 is always ON. The M3 gate signal
is triggered by the duty cycle d2, M4 is triggered opposite to d2. In addition, the M2 gate
signal should be the compliment of M1’s at all times, otherwise the input voltage Vg is
shorted. Likewise, the M3 gate signal should take the complimentary value of M4’s.
Logic Optimized
ORCAD Simulated Logic
Our logic circuit (above left) could be logically reduced to the circuit on the right. However,
we consistently experienced unknown component errors in the ORCAD simulation using
the NAND gate reduction to the right. For this reason, we chose to simulate the circuit to
the left even though it is not minimal.
8
The following image is the full schematic for a Pulse Width Modulated H-bridge with
the energy saving gating scheme.
9
Energy Saving Mode Pulse Width Modulation Controlled H-bride gating Signals
Energy Saving Mode PWM H-bride Output Voltage compared to an ideal sinusoid The
large inductor causes a phase delay (lag). ORCAD Lite does not give us the ability to
create a delayed reference sinusoid.
(Contains Input Voltage with perturbation, reference sinusoid, and Output Voltage)
10
The unipolar output voltage signal is very similar to the bipolar one. However, the VA
voltage swing is only half that of the bipolar case. VA is defined as the voltage differential
just prior to the output filter. This makes the unipolar PWM case “energy saving.”
11
should be close to, but a bit shorter than the switching period TS. Since
1 1
TS = = = 50µs , an acceptable integration time constant ( τ = 40µs ) is given by
f S 20kHz
R = 100kΩ and C = 400 pF .
The open-loop circuitry special to OCC is shown below. The upper left section of the
schematic shows the SET-RESET latch. The integrator output is fed back to RESET, and a
clock corresponding to the switching period drives SET.
13
The following image shows the OCC process:
Each time the integrated signal reaches the sinusoid reference, RESET is activated, and
the integration process begins anew. Consequently, the duty ratio, d2, increases with
increasing sinusoid and vice versa.
The following image shows the open-loop OCC controlled H-bride output voltage compared
with an ideal sinusoid. Note: The ORCAD Lite software only permits a limited number of
logic transitions, so the OCC case, which requires more logic transitions, has a shorter
simulation time. The input voltage is 300 V.
The large inductor causes a phase delay (lag). Closed-loop control is required to appreciate
OCC’s fast tracking. Therefore, the above waveform is comparable to PWM’s open-loop
response.
14
OCC more faithfully tracks the response even with the input perturbation in the closed-loop.
Remember the inductor causes a phase lag.
17
As expected, the Open-loop control schemes had comparable tracking error. The Closed-
loop OCC provided greatly improved performance. If time permitted, it would have been
nice to design a PID controller to close the PWM loop. This speaks to the advantage of
OCC and its simplicity.
Conclusion
Through this exercise we analyzed the design and implementation of two open-loop control
schemes: Pulse Width Modulation and One Cycle Control. These schemes were further
specialized to Bipolar gating and Energy-Saving (unipolar) gating. Energy-Saving gating is
preferred since it consumes much less power. One Cycle Control, though not meant for
open-loop control, is great in a closed-loop since it eliminates the need to design a PID
controller. Since duty cycle is recalculated each switching period, waveforms are more
faithfully reproduced, especially when an input disturbance is introduced.
It would be desirable to analyze the efficiency of both Bipolar and Unipolar cases.
Additionally we would like to quantify cross-over and total harmonic distortion as well.
However, the ORCAD Lite software wattmeter does not function properly, and distortion
analysis is not available. Future students using the recently donated Saber software should
be able to accomplish this.