Miscellaneous active electrical nonlinear devices – circuits – and – Signal converting – shaping – or generating – Rectangular or pulse waveform width control
Reexamination Certificate
2002-04-19
2003-10-21
Callahan, Timothy P. (Department: 2817)
Miscellaneous active electrical nonlinear devices, circuits, and
Signal converting, shaping, or generating
Rectangular or pulse waveform width control
C327S170000
Reexamination Certificate
active
06636094
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to pulse-width-modulated (PWM) signals.
BACKGROUND OF THE INVENTION
Pulse-width-modulated signals are useful, for example, for the control of asynchronous motors or inverters used in household appliances, ventilation air control systems, pumping systems and the like. A pulse-width-modulated signal SC (
FIG. 1
c
) is a cyclical signal that is activated periodically (A
i
, A
i+1
), and deactivated at variable points in time (D
i
, D
i+1
) between two operations of activation. The time TC between two operations of activation is called a switching period. The ratio TC1/TC between the duration of the active signal and the switching period is called a cyclical ratio. The signal SC may be active at 0 or at 1.
When a pulse-width-modulated signal is used to control an asynchronous motor, the variations in time of the cyclical ratio of the control signal lead to similar variations in current in the phase or phases of the motor. For example, with an appropriate control signal SC, sinusoidal variations of the current can be obtained in the phase or phases of the motor.
A pulse-width-modulated signal SC is conventionally obtained by a generator that produces the signal SC from a reference signal SREF which is a sampled analog signal. The initial analog signal may have a sine (
FIG. 1
a
), trapezoidal, square or any other shape. The sample reference signal SREF (
FIG. 1
b
) is obtained by the sampling of the initial analog signal. It comprises a set of binary numbers E
0
to E
NBECH
whose value depends on the amplitude of the signal at the instant considered. The smallest value of the numbers E
0
to E
NBECH
, for example, equal to 0, is associated with the sample having the smallest amplitude. Conversely, the highest value of the numbers E
0
to E
NBECH
is associated with the greatest amplitude of the initial signal.
The number of signals chosen NBECH is a compromise between the desired precision (which increases with the number of samples) and the computation time needed to obtain the signal SC from the sampled reference signal SREF (which also increases with the number of samples). For example, in the case of an initial sine analog signal, it could be chosen to take 360 samples, i.e., every 1
0
on a period TREF=360° of the analog signal, or else 2
N
samples every TREF/2
N
on a period TREF of the analog signal, with N being the size of a register of the circuit.
A known generator of pulse-width-modulated signals is shown in FIG.
2
. The generator has a counter CPC, a reference memory MR, a comparison register RC and a comparator CPT. The numbers E
0
to E
NBECH
are stored in the reference memory MR of the generator. The M-bit counter CPC counts pulses of a clock signal CP and gives a number of counted pulses NB varying between 0 and NBMAX=2
M
−1. When the number NBMAX is reached, the counter returns to zero and then starts counting again.
The samples E
0
to E
NBCH
of the sampled reference signal SREF contained in the reference memory MR are successively loaded into the comparison register RC. A sample E
i
is loaded at each return to zero of the counter CPC. As the case may be, the same sample E
i
may be loaded several times successively. In every case, a loading is done at each return of the counter to zero.
The comparator CPT continuously compares the number NB given by the counter with the sample E
i
contained in the comparison register RC. The comparator CPT gives the control signal SC having the following properties. For every value of i ranging from 0 to NBECH, SC is active when NB<E
i
, and SC is inactive when NB≧E
i
.
FIG. 1
c
shows the development of the number NB, the samples E
i
and the signal SC resulting in one example. The signal SC is active at 1 and inactive at 0. The signal SC is thus activated at each return to zero of the counter, and then deactivated when the number NB given by the counter exceeds the value E
i
of the sample contained in the comparison register RC.
Referring to
FIG. 1C
, a pulse signal is obtained. In this signal, the width of the active pulses (and hence the cyclical ratio) varies in time as a function of the value E
i
of the samples of the reference signal SREF, and hence as a function of the initial analog signal. The precision of the generator depends on the range of variation of the cyclical ratio and on the minimum variation of the cyclical ratio.
The cyclical ratio R of the control signal SC can be computed as follows. For each period TC of the counter (since the signal SC is activated at each return to 0 of the counter):
R
=
(
E
i
*
TCP
)
/
(
TC
)
=
(
E
i
*
TCP
)
/
(
(
NBMAX
+
1
)
*
TCP
)
=
E
i
/
(
NBMAX
+
1
)
.
NBMAX=2
M
−1 is the maximum value of the number NB, M is the size of the counter, and TCP is the period of the clock signal CP.
The minimum variation in the cyclical ratio is equal to &Dgr;Rmin=1/(NBMAX+1). The precision of the generator is directly proportional to the number NBMAX, namely the size of the counter. The precision of the generator increases also with the range of variation of the cyclical ratio. It is preferable to have available a generator producing control signals whose cyclical ratio varies from 0 to 100% to have as wide a range of control as possible. In this way, unnecessary losses and deterioration are avoided in the electronic control circuits of the motor.
The minimum value Rmin of the cyclical ration is equal to 0%. This corresponds to E
i
=0.
The maximum value of the cyclical ratio is equal to:
R
max=
X
/(
NB
MAX+1).
X is the maximum value of the numbers E
0
to E
NBCH
. Rmax can reach 100% only if X can reach the value NBMAX+1=2
M
, with M being the size of the counter. To attain a cyclical ratio of 100%, the numbers E
0
to E
NBCH
should be encoded on a number of bits at least equal to M+1 to be able to reach the value 2
M
.
If the register RC used has a size N (for example, N=16) greater than the size M (for example, M=12) of the counter CPC, this does not raise any problems. It is possible to use 13-bit numbers E
i
which are loaded into the register RC and then are compared with the 12-bit numbers NB given by the counter. It is also possible to use 16-bit numbers E
i
(enabling higher precision to be obtained on the sampled signal). These numbers are loaded into the register RC, and only the 13 most significant bits of the numbers E
i
are compared with the 12-bit numbers NB given by the counter CPC.
A problem arises, however, when the registers and the counter have an identical size. A first known approach uses two comparison registers, the first to store the N least significant bits of the numbers E
0
to E
NBCH
, and the second to store the most significant bits of these numbers. This approach, however, is not worthwhile because it implies the loading of two registers at each return of the counter to zero. This increases the time for loading the numbers E
i
, and therefore the time for computing the signal SC. Furthermore, the size of the circuit is increased.
In a second approach, only the M−1 least significant bits of the counter CPC are used to produce the M−1 bit numbers NB, and the register RC sized N=M is used to store the numbers E
0
to E
NBCH
. It is thus possible to attain a cyclical ratio Rmax equal to 100%. This approach, however, is not worthwhile because it impairs the performance of the generator by reducing its precision (the number &Dgr;Rmin increases).
Thus, if the registers and the counter of the generator are identical in size, it is not possible, with the prior art approaches, to obtain an optimum generator that has both maximum precision and minimum computation time.
SUMMARY OF THE INVENTION
In view of the foregoing background, an object of the present invention to provide an optimized generator making the most efficient use of the capacities of its components while at the same time maintaining high precision and limited computation time.
This and other objects, advanta
Allen Dyer Doppelt Milbrath & Gilchrist, P.A.
Callahan Timothy P.
Cox Cassandra
Jorgenson Lisa K.
STMicroelectronics SA
LandOfFree
Method for the generation of pulse-width-modulated signals... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for the generation of pulse-width-modulated signals..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for the generation of pulse-width-modulated signals... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3124802