Pulse Width Modulation (PWM) Basics
There
are many forms of modulation used for communicating information.
When a high frequency signal has an amplitude varied in response to
a lower frequecny signal we have AM (amplitude modulation). When the
signal frequency is varied in response to the modulating signal we
have FM (frequency modulation. These signals are used for radio
modulation because the high frequncy carrier signal is needs for
efficient radiation of the signal. When communication by pulses was
introduced, the amplitude, frequency and pulse width become possible
modulation options. In many power electronic converters where the
output voltage can be one of two values the only option is
modulation of average conduction time.

Fig. 1: Unmodulated, sine modulated pulses
1.
Linear Modulation
The
simplest modulation to interpret is where the average ON time of the
pulses varies proportionally with the modulating signal. The
advantage of linear processing for this application lies in the ease
of de-modulation. The modulating signal can be recovered from the
PWM by low pass filtering. For a single low frequency sine wave as
modulating signal modulating the width of a fixed frequency (fs)
pulse train the spectra is as shown in Fig 2. Clearly a low pass
filter can extract the modulating component fm.

Fig. 2: Spectra of PWM
2.
Sawtooth PWM
The
simplest analog form of generating fixed frequency PWM is by
comparison with a linear slope waveform such as a sawtooth. As seen
in Fig 2 the output signal goes high when the sine wave is higher
than the sawtooth. This is implemented using a comparitor whose
outputvoltage goes to a logic HIGH when ne input is greater than the
other.

Fig. 3: Sine Sawtooth PWM
Other
signals with straight edges can be used for modulation a rising ramp
carrier will generate PWM with Trailing Edge Modulation.

Fig. 4: Trailing Edge Modulation
It is
easier to have an integrator with a reset to generate the ramp in
Fig 4 but the modulation is inferior to double edge
modulation.
3.
Regular Sampled PWM
The
scheme illustrated above generates a switching edge at the instant
of crossing of the sine wave and the triangle. This is an easy
scheme to implement using analog electronics but suffers the
imprecision and drift of all analog computation as well as having
difficulties of generating multiple edges when the signal has even a
small added noise. Many modulators are now implemented digitally but
there is difficulty is computing the precise intercept of the
modulating wave and the carrier. Regular sampled PWM makes the width
of the pulse proportional to the value of the modulating signal at
the beginning of the carrier period. In Fig 5 the intercept of the
sample values with the traingle determine the edges of the Pulses.
For a sawtooth wave of frequency fs the samples are at
2fs.

Fig. 5: Regular Sampled PWM
There
are many ways to generate a Pulse Width Modulated signal other than
fixed frquency sine sawtooth. For three phase systems the modulation
of a Voltage Source Inverter can generate a PWM signal for each
phase leg by comparison of the desired output voltage waveform for
each phase with the same sawtooth. One alternative which is easier
to implement in a computer and gives a larger MODULATION DEPTH is
using SPACE VECTOR MODULATION.
4.
Modulation Depth
For a
single phase inverter modulated by a sine-sawtooth comparison, if we
compare a sine wave of magnitude from -2 to +2 with a traingle from
-1 to +1 the linear relation between the input signal and the
average output signal will be lost. Once the sine wave reaches the
peak of the transgle the pulses will be of maximum width and the
modulation will then saturate. The Modulation depth is the ratio of
the current signal to the case when saturation is just starting.
Thus sine wave of peak 1.2 compared with a triangle with peak 2.0
will have a modulation depth of m=0.6.

Fig. 6: Saturated Pulse Width Modulation
Copyright © G. Ledwich 1998.
|