AN2154 datasheet(11/16 Pages) STMICROELECTRONICS

11 / 16 page

AN2154

ST7MC implementation of symmetrical SVPWM

11/16

3.2

Symmetrical SVPWM algorithm description

As discussed in Chapter 1, in order to generate SVPWM, vector V must rotate continuously

inside the space vector hexagon. This implies that time ta, tb and t0 must continuously vary

following Equation 10 - 12. Since these equations are only valid for

α in the range between 0

and

π/3, it could be very convenient to represent angle á by an 11-bit variable. This way, the

most significant 3 bits can be used to indicate the sector, and the least significant 8-bits can

contain the angle of vector V inside the identified sector. The resolution is given by

π/(3.256)

radians equivalent to 0.234 degrees. In practical implementation, to keep the frequency

resolution high, it is necessary to store angle

α in a 16-bit variable. According to the previous

discussion, the most significant 3 bits identify the sector, the successive 8 bits are used to

identify the angle inside the sector and the least significant 5 bits are not used for ta, tb and

t0 computation.

In order to reduce the CPU computational load, a look-up table has been used for storing

time ta and tb using Equation 10 and 11 with the maximum admissible value of modulation

index mi (that is 0.866).

Moreover, comparing Equation 10 and 11 it is possible to note that the ta table is exactly the

same as the tb table in reverse. This means that only one look-up table is necessary. During

runtime, the look-up times ta and tb must be scaled accordingly with the value of modulation

index imposed from the user. This is achieved in two steps: first, times ta and tb are

multiplied by an 8-bit variable (SineMag) read from potentiometer RV2 and, therefore, only

the result of the most significant byte is considered (that is the equivalent of a division by

256).

After ta and tb scaling, the three values to be loaded into the compare registers must be

calculated to generate the waveform shown in Figure 7. Let us suppose, for example, that

vector V is, in a well known instant, situated in sector 0 and, therefore, it has to be "time

weighted" between state 001 and 011 (this situation is exposed in Figure 7). The values to

be loaded into the compare registers are:

Equation 14

Equation 15

Equation 16

Moreover, as it is:

Equation 17

and

Equation 18

----

⎝⎠

⎛⎞

counter

⋅

Phase U compare register (MCMPU)=

----

⎝⎠

⎛⎞

counter

⋅

Phase V compare register (MCMPV)=

----

⎝⎠

⎛⎞

----

–

⎝⎠

⎛⎞

counter

⋅

Phase W compare register (MCMPW)=

⋅

()

-------------------

15.625

⋅

()

---------------------------------

µs

MCMP0

counter

1255

–

⋅

AN2154 Datasheet(PDF) 11 Page - STMicroelectronics

AN2154 Datasheet(HTML) 11 Page - STMicroelectronics

Html Pages

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Datasheet Download

Link URL