Miscellaneous active electrical nonlinear devices – circuits – and – Signal converting – shaping – or generating – Frequency or repetition rate conversion or control
Reexamination Certificate
2001-06-14
2002-05-21
Tran, Toan (Department: 2816)
Miscellaneous active electrical nonlinear devices, circuits, and
Signal converting, shaping, or generating
Frequency or repetition rate conversion or control
C327S117000
Reexamination Certificate
active
06392455
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the field of electronic circuits, and in particular to a circuit for generating lower-speed baud rates, such as used for serial communications, from higher-speed oscillators, such as microprocessor clocks
2. Description of Related Art
Digital systems typically include a clocking signal for providing periodic signals, and signals of known duration. That is, the clocking signal provides a time reference upon which other timed actions depend. Within the same system, multiple time references are often required. For example, in a microprocessor system, the microprocessor typically requires a high-speed clock that regulates the speed of the central processing unit (CPU), the speed and timing of memory accesses, and other high-speed events. Additionally, for communicating with external devices, such as modems or printers, the microprocessor requires a relatively low speed clock.
Because whole-cycle frequency dividers are relatively simple devices, compared to frequency generators, systems with multiple time references typically provide the multiple time references by dividing-down a high-speed clock.
FIG. 1
illustrates a microprocessor system that includes a conventional frequency divider
100
. The divider
100
is programmable to provide a lower speed clock signal, the “Baud Rate Clock”, from a higher speed clock signal, the “Xtal Osc Clock” (Crystal Oscillator Clock). The high-speed clock signal is preferably the highest speed clock within a processing system, which is generally the same clock signal used by the core processor, or CPU,
150
. The example divider
100
includes a re-loadable count-down counter
110
, and a register
120
that provides the re-load value to the counter
110
. In operation, the register
120
is loaded with a value corresponding to a ratio of the high-speed clock frequency to the desired low-speed clock frequency, discussed further below, and thereafter the low-speed clock is automatically generated at the desired frequency, without further intervention by the core processor
150
.
The count-down counter
110
is clocked by the higher speed clock signal, and generates an output signal each time the count-down counter
110
reaches zero. The output signal is also coupled to the reload input of the counter
110
, and causes the counter
110
to be reloaded with the contents of the register
120
at the next clock cycle of the higher speed clock signal. Thus, if the register
120
contains a value of N, the zero signal will be asserted at each N+1th cycle of the higher speed clock (one cycle to load the value of N, plus N cycles to reduce the count to zero), thereby providing a “divide-by-N+1” frequency divider that can be programmed for a particular division by loading the proper value of N into the register
120
. Illustrated in
FIG. 1
is a 12-bit register
120
and 12-bit counter
110
, thereby allowing the frequency divider
100
to provide a division of the high-speed clock by any integer multiple between 2 and 8192.
Note that unless the high-speed clock is approximately an integer multiple of the low-speed clock, the frequency divider
100
will not be able to provide an accurate low-speed clock frequency. Note also that the higher the ratio of the high-speed to low-speed clock frequencies, the finer the available resolution. Consider, for example, a 100 MHz high-speed clock, and a desired low-speed clock frequency of 18 KHz. A ratio of 5555 provides a low-speed clock of 18,001.8 Hz, and a ratio of 5556 provides a low-speed clock of 17,998.5 Hz. In this example, 5556 would be selected, and the low-speed clock will be within 1.5 Hertz of the desired 18 KHz (less than 0.01% error). Consider, however, a high-speed clock of 100 KHz, and the same desired low-speed clock frequency. A ratio of 5 provides a low-speed clock of 20,000.0 Hz, and a ratio of 6 provides a low-speed clock of 16,666.7 Hz. In this example, the best option (a ratio of 6) will produce an error of over 1,333 Hz. (over 7% error), and would likely be unacceptable. Generally, for acceptable frequency accuracy, a minimum ratio of 50:1 is required for supporting independent high-speed and low-speed clock frequencies. That is, the high-speed clock frequency is typically selected by the designer of the core processor
150
, or other components within the microprocessor system, and the low-speed clock frequency is typically determined based on existing communications standards, or based on characteristics of a device that is external to the microprocessor system. As such, the high-speed clock frequency and low-speed clock frequency are generally substantially independent of each other, and an integer-factor relationship between the high-speed and low-speed clocks cannot be assumed. To allow for the programming of an arbitrary low-speed clock frequency, a ratio of 50:1 between the high and low clock frequencies is generally considered a minimum requirement.
This minimum ratio of 50:1 for the general-purpose application of a programmable frequency divider had not been a significant constraint in typical microprocessor designs, because the master CPU clock has conventionally been substantially faster than the clocks required for serial communications, or other derived-clock applications. The increased need for low-power processors (lower processor clock speed), and the increased need for faster communications (higher interfacing clock speeds), however, has substantially narrowed the gap between CPU clock speeds and derived-clock speeds.
BRIEF SUMMARY OF THE INVENTION
It is an object of this invention to provide a frequency divider that can provide accurate low-speed clock frequencies with a minimal difference between the low-speed clock frequency and the clock frequency that provides the input to the frequency divider. It is a further object of this invention to provide a programmable frequency divider that has a resolution that is substantially less dependent upon the difference between the high and low clock frequencies than conventional frequency dividers.
These objects and others are achieved by providing a programmable fractional frequency divider that is configured to enable a finer resolution of output frequency than conventional frequency dividers. The programmable fractional frequency divider of this invention allows for the programmability of both an integer divisor as well as a fraction component. The average frequency of the output signal from the fractional divider is dependent upon both the integer divisor and the fraction component, thereby providing for a finer resolution to the average frequency of the output signal. This combination of integer and fractional frequency division is particularly well suited for the generation of signals for systems that are substantially jitter-insensitive.
REFERENCES:
patent: 4638498 (1987-01-01), Sinniger et al.
patent: 5361290 (1994-11-01), Akiyama
patent: 5754067 (1998-05-01), Komatsu et al.
patent: 6072534 (2000-06-01), Dell'ova et al.
Koninklijke Philips Electronics , N.V.
Nguyen Linh
Tran Toan
Tsiang Harold
LandOfFree
Baud rate generator with fractional divider does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Baud rate generator with fractional divider, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Baud rate generator with fractional divider will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2864226