JP2896004B2 - Calendar iC correction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for correcting a time measurement error of a calendar iC in a system including a calendar iC for measuring time and a microcomputer for reading the time. In the drawings, the same reference numerals indicate the same or corresponding parts.

[0002]

2. Description of the Related Art FIG. 3 shows an example of the configuration of a conventional system of this kind. In the figure, reference numeral 2 denotes a calendar iC, 4 denotes a crystal oscillator for a calendar iC which gives a clock frequency to the calendar iC (hereinafter, the crystal oscillator is also simply abbreviated as crystal), and 1 denotes the time measured by the calendar iC2 when necessary. It is a CPU (also called a microcomputer) to issue.

By the way, adding a calendar function to a computer is generally widely performed. In many cases, the calendar function performs time measurement using a calendar iC2 independent of the CPU 1 that performs main processing as shown in FIG.
Then, the CPU 1 obtains the time by reading the time register in the calendar iC2. In the system as shown in FIG. 3, the accuracy of the calendar depends on the oscillation frequency characteristics of the crystal oscillator 4 connected to the calendar iC2. The generally used crystal oscillator for calendar iC has the temperature characteristics shown in FIG. 4, so that when the temperature changes, the oscillation frequency changes relatively largely, causing an error in time.

A crystal oscillator having a temperature characteristic as shown in FIG. 4 is used because of the following three requirements. Keep oscillating even at low voltage such as battery backup. The oscillation frequency must be low to minimize battery consumption. Inexpensive.

[0005] Quartz crystals that can be selected to satisfy these requirements have only the temperature characteristics shown in FIG. For this reason, the conventional system has a problem that the accuracy generally changes when the temperature changes. In addition, a crystal resonator whose accuracy changes little due to temperature due to special crystal processing is also known, but is expensive. Next, another conventional system that solves such a problem will be described with reference to FIGS. 1 and 2. FIG. FIG. 1 is another conventional system configuration diagram corresponding to FIG.

FIG. 2 is a flowchart showing the operation of the main part of the CPU shown in FIG. In FIG. 1, calendar iC
Reference numeral 2 has an alarm function and can output an interrupt signal iNT to the CPU 1 when the time set in the alarm register is reached (in this example, every one minute).
Reference numeral 3 denotes a crystal oscillator for driving the CPU 1 (also referred to as a crystal for a microcomputer). Reference numeral 5 denotes a free-run counter that counts with a CPU clock CLK generated from the crystal oscillator 3 for the CPU, and has a counting cycle of 1 msec. Reference numeral 6 denotes a memory read and written by the CPU 1.

Next, the operation of the main part of the CPU 1 will be described with reference to FIG. In the following, reference numerals S1 to S10 indicate steps in FIG. The CPU 1 reads the time from the calendar iC2 and stores the time in the memory 6, and also reads the time from the counter iC2.
Set to output an interrupt signal iNT after one minute (S
1). However, one minute notified by the calendar iC2 includes an error of the quartz crystal 4 for calendar. Immediately after the setting, the CPU 1 reads the current free-run counter value 5a from the free-run counter 5, stores it in the memory 6 (S2), and waits for an interrupt from the calendar iC2 (S2).
3).

Next, when an interrupt from the calendar iC2 occurs, the processing proceeds to the next (S3, with branch). That is, after the interruption occurs, the free-run counter 5 is read (S4), and the previous read value is subtracted from this read value, and the time during which the calendar iC2 is 1 minute is obtained based on the CPU clock (S5). ). For example, when the frequency deviation of the calendar iC crystal 4 is 0, the counter value becomes 60,000 counts in one minute interrupt interval caused by the calendar iC2 (1 minute = 60,000 msec). Assuming that the error of the calendar crystal 4 is -50 ppm, the counter becomes 60003 counts. Similarly, when the error is +50 ppm, it becomes 59997 counts.
Whether the count value is more or less than 60000, it detects whether it is advancing or lagging.

If an error is detected in this way (S5, with branch), then the integrated value of the error is stored in the memory 6 (S6). Then, when the time error of the minimum unit that can change the time is reached (S7, branch Y), the time is changed. That is, the time is advanced if it is just before the time (S8 → S9), and the time is advanced if it is almost the last (S8 → S10). For example, if the time of the calendar iC being used can be changed in seconds, the time is changed when the integrated value reaches 1 second (1000 counts).

[0010]

The best correction effect can be obtained for a board which performs correction in the other conventional system described above, if a microcomputer crystal having a specified oscillation frequency can be mounted. However, although the crystal for the microcomputer has a small change in the temperature of the oscillation frequency, the expensive one has a small variation of the oscillation frequency of several ppm as shown in FIG. 8 and the cheap one has a large variation of about 100 ppm as shown in FIG. However, now, suppose the calendar crystal error is 0pp
If a microcomputer crystal of -50 ppm is mounted at m, the calendar generates an error of about -50 ppm.

This is for the following reason. That is, since the microcomputer crystal is the reference, even if the calendar iC notifies the time at a time interval of zero error by the crystal having the error of 0 ppm, the free-run counter 5 of the microcomputer side has 50 times.
Since the count is performed at a low ppm rate, the count number is smaller by 50 ppm. The microcomputer uses 5
It mistakenly reports that the time has been notified at a time interval shorter by 0 ppm, and concludes that the calendar has advanced too much. Therefore, the time adjustment that is not actually necessary is performed, and the calendar i
The time of C is delayed. That is, the calendar is delayed by about 50 ppm as a result.

Therefore, the present invention utilizes a low-cost crystal for a microcomputer, and even when a crystal for a microcomputer of -50 ppm is mounted, a calendar error can be correctly recognized and the calendar error can be minimized. It is an object to provide a calendar iC correction method.

[0013]

According to a first aspect of the present invention, there is provided a calendar iC correction method, comprising:
A calendar iC (e.g., 2) that performs a timing operation with an output clock of a crystal oscillator for C (e.g., 4) and a CPU (e.g., 1) driven by the output clock of a crystal oscillator for CPU (e.g., 3). A system comprising:
A means (such as a battery backup RAM 7) for storing frequency error data (7a or the like) of the PU crystal oscillator, wherein the CPU measures a time measured based on the output clock of the CPU crystal oscillator, and further stores the frequency error data. The time measured by the calendar iC (hereinafter, referred to as a second time) is corrected based on the time measured by referring to and corrected (hereinafter, referred to as a first time).

According to a second aspect of the present invention, there is provided a calendar iC correction method according to the first aspect.
The CPU accumulates an error (stored in the memory 6 or the like) of the second clock time with respect to the first clock time at predetermined intervals, and each time the integrated value of the error reaches a minimum unit of time change. The correction is performed. In the calendar iC correction method according to a third aspect, in the calendar iC correction method according to the first or second aspect, the first clocking time counts an output clock (CLK or the like) of the CPU crystal oscillator. It is assumed that the count value of the counter (such as the free counter 5) is corrected and the time is measured.

According to a fourth aspect of the present invention, there is provided a calendar iC correction method according to the third aspect.
The CPU reads the count value of the counter once each time the calendar iC synchronizes with an interrupt request (eg, one-segmented iNT) generated at a predetermined time interval measured by the calendar iC, and performs the correction. Things.

[0016]

The crystal oscillator connected to the microcomputer generally has a temperature characteristic with a small frequency change as shown in FIG. Since the frequency band required by the microcomputer is high and does not require operation during battery backup, it is easy and inexpensive to obtain a crystal oscillator having the temperature characteristics shown in FIG. Crystal oscillators are commonly used.

The frequency deviation of the crystal for the microcomputer with respect to the temperature change is only about one-tenth to one-tenth that of the crystal for the calendar iC. Therefore, the time can be corrected by reading the time register in the calendar iC based on the time measured by the microcomputer, extracting the error, and rewriting the time. As a result of such correction, the time of the calendar iC always falls within the error of the microcomputer crystal regardless of the accuracy of the calendar crystal. Then, the error is reduced from a fraction to a tenth before correction.

When using a crystal for a microcomputer which has a small frequency deviation with respect to a temperature change but deviates from a specified oscillation frequency, a non-volatile memory 7 is added, and the oscillation frequency of the crystal for the microcomputer is added to the non-volatile memory 7. Stores the error. In the time correction process, the microcomputer refers to and corrects the error data of the microcomputer crystal stored in the memory 7 from the collected calendar iC error, and finally obtains the calendar error. ).

When the time error of the calendar iC is specifically determined, the count value of the free-run counter 5 in which the microcomputer counts the microcomputer clock is synchronized with the calendar iC, for example, in response to an interrupt request generated every one minute. Is read once each time, thereby preventing the time correction processing time of the microcomputer from intervening in the clock error.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the invention according to claims 1 to 3 will be described with reference to FIGS. The invention according to claims 1 to 3 uses a low-cost crystal for a microcomputer, and
An object of the present invention is to provide a calendar iC correction method capable of correctly recognizing a calendar error and minimizing a calendar error even when a 0 ppm microcomputer crystal is mounted.

That is, in the present invention, in this case,
Until the count of the free-run counter is reduced by 50 ppm, it is not different from the conventional one. However, before making a final judgment whether the calendar is advanced or delayed, the error data of the microcomputer crystal is read from the non-volatile memory, and the error of the free-run counter itself is -50 pp.
This is corrected because there is m (step S4A in FIG. 7). Then, the result that the calendar error is 0 is obtained, and the time is not adjusted. However, while the frequency error of the crystal for the microcomputer exists as an analog value, the memory for writing the error data and the error calculation value are digital, so that a rounding error occurs at the time of digitization. This rounding error finally becomes the remaining calendar error in the calendar iC correction method according to the present invention. FIG. 6 is a system configuration diagram as one embodiment of the present invention, corresponding to FIG. In FIG. 6, a 4-bit RA battery-backed up in calendar iC2 in FIG.
The difference is that M7 is incorporated. However, generally, this RAM 7 may be outside the calendar iC2. The calendar iC2 has an alarm function as in FIG. 1 and can output an interrupt signal iNT to the CPU 1 when the time set in the alarm register is reached (every minute in this example).

The backup memory 7 stores the frequency error of the crystal 3 for the microcomputer. This storing operation is performed at the time of the shipping test of each microcomputer board, and the error value obtained by measuring the oscillation frequency is written. Since the written error data is backed up in the RAM 7 by a battery, the error data remains even if the power is turned off unless rewritten thereafter. Free run counter 5
1 is a counter that counts with a CPU clock CLK generated from the CPU crystal 3 as in FIG. 1 and has a counting cycle of 1 msec.

FIG. 7 is a flowchart showing a first embodiment of the correction processing of the CPU 1 of FIG. 6 and corresponds to FIG. FIG. 7 is different from FIG. 2 in that step S4A is added. Next, referring to FIG. 6, referring to FIG. 7, the CPU 1 sets the calendar iC2 to issue an interrupt after one minute (S1). However, one minute notified by the calendar iC2 includes an error of the quartz crystal 4 for calendar. Immediately after the setting, the CPU reads the current free-run counter value (referred to as (A)) from the free-run counter 5 and stores it in the memory 6 (S2).
Wait for an interrupt from C2 (S3). Next, when an interrupt from the calendar iC2 occurs (S3, with branch), the process proceeds to the next.

After the occurrence of the interrupt, the value of the free-run counter 5 is read, and when the previous read value (A) is obtained, the CPU i.c.
It is determined by a value based on K (referred to as (B)) (S4).
For example, when the frequency deviation of the calendar iC crystal 4 is 0 and the error of the microcomputer crystal 3 is 0, the calendar iC
The count value (B) of the counter 5 at the 1-minute interrupt interval caused by 2 becomes 60,000 counts (1 minute = 60,000 msec).
When the error of the quartz crystal 4 for the calendar is -50 ppm and the error of the quartz crystal 3 for the microcomputer is 0, the count value (B) becomes 60003 counts. Also, the error of the crystal 4 for the calendar is + 50ppm,
When the error of the crystal 3 for the microcomputer is 0, the count value (B) is
59997 counts. Whether the count value (B) is more or less than 60,000 is detected to determine whether it is advancing or lagging.

The detected count value (B) may include an error of the microcomputer crystal 3 itself. For example, when the error of the calendar crystal 4 is +50 ppm and the error of the microcomputer crystal is also +50 ppm, the count value (B) is 60000. Therefore, the count value (B) is corrected by reading the error 7a of the microcomputer crystal 3 from the battery backup RAM 7 and subtracting the error 7a from the count value (B) (S4).
A). Then, the remainder becomes an error of the calendar iC crystal 4 and is used for updating the error integrated value of the memory 6 (S
5, S6). When the integrated value reaches the time error of the minimum unit that can change the time (S7, branch Y), the time is changed. If it is just before the time, the time is delayed (S8, S9), and if it is just behind, the time is advanced (S8, S10). For example, if the time of the used calendar iC2 can be changed in seconds, the time is changed when the integrated value reaches 1 second (1000 counts).

Next, referring to FIGS. 10 to 13, C of FIG.
A second embodiment of the correction processing of PU1 will be described. When the correction processing (correction processing (2-1) for convenience) of the first embodiment of the CPU 1 in FIG. 6 described in FIG. 7 is performed, if the execution time of this correction processing cannot be ignored, an erroneous correction is performed. There is a problem to do.

FIG. 10 is a time chart for explaining this problem, which is based on execution of the main program of the CPU 1, interruption of the calendar iC2, execution of the correction processing program of the CPU 1, and correction processing (2-1) in order from the top. The timing of reading the count value of the free-run counter 5 is shown. Here, the interrupt from the calendar iC2 is raised to the CPU 1 every minute, and the time is corrected by comparing the value with the value of the free-run counter 5 to obtain an error. Here, in hardware with a resolution of 1 msec of the free-run counter 5 and a minimum unit of time correction of 1 sec of the calendar iC2, the calendar i
Assuming that oscillation occurs with an error of C2 being zero, the CPU 1 must recognize that one minute notified by the calendar iC2 is 60,000 counts in order to perform correct correction. If it is long, it will be less than 60,000 counts. This is calendar i
This is because the CPU 1 reads the value of the free-run counter 5 at the end (read value An) and at the beginning (read value Bn) of the correction processing program, while C2 continuously counts the time. . Here, the subscript n for the read values A and B generally indicates the total number of executions (execution number) of the correction processing program up to that time. Therefore, if the program execution takes a long time, it will not be counted.

For example, assuming that the execution time of the correction processing program of the CPU 1 is Texe msec, one minute of the calendar recognized by the CPU 1 = Bn−An = 60000−Texe + αn−αn−1 = (60000−Texe) + (Αn−αn−1) ≒ (60000−Texe) msec. Note that αn−αn−1 is the interrupt response time difference shown in FIG. 10, which may be a positive value or a negative value. There is no effect on error detection. Therefore, a transformation from the equation is possible.

After all, the CPU 1 mistakenly recognizes the correction processing program execution time Texe as the time during which the calendar is delayed, and
This Texe msec is accumulated as an error. When the erroneously recognized error reaches the minimum time correction unit of 1 second of the calendar iC, erroneous time correction is performed.
FIG. 13 corresponds to FIG. 10 showing the read timing of the free-run counter 5 based on the correction processing (correction processing (2-2) for convenience) of the CPU 1 of FIG. FIG.

As shown in FIG. 13, in the correction process (2-2), the value A of the free-run counter read every minute is read.
Instead of n (n ≠ 0; the same applies hereinafter), for the second and subsequent times, the difference from the current value Bn is obtained using the previous value Bn−1,
Finally, a calendar error is determined. According to this method, the value of the free-run counter 5 is read only at the timing synchronized with the interrupt request notified by the calendar iC2. Regardless, the calendar error can be correctly recognized. Therefore, the time correction is not performed erroneously.

FIG. 11 shows the overall processing flow of the microcomputer system of FIG. 6 when the correction processing (2-2) is executed, and FIG. 12 shows the detailed flow of the correction processing program. That is, in FIG. 11, every time there is an interrupt from the calendar iC2, the correction processing program in FIG. 12 is called. After executing the correction processing program, the process returns to the main processing.

Next, referring to FIGS. 6 and 13, FIG.
Referring to FIG. In the following, reference numerals S21 to S23 indicate respective steps in FIG. 11, and reference numerals S31 to S44 indicate respective steps in FIG. Immediately after the power is turned on, first, the CPU 1
In the first main process, the interrupt is set to be issued to the calendar iC2 after one minute (S21). Immediately after the setting, the CPU 1 reads the current free-run counter value A0 from the free-run counter 5 and stores it in the memory 6 (S23). During the execution of the main processing (S23), the CPU 1 interrupts from the calendar iC2 coming one minute later. Wait for. When an interrupt occurs, the correction processing program of FIG. 12 is called.

In FIG. 12, first, the value B0 of the free-run counter 5 is read (S31), and when the difference C = (B0-A0) from the previous read value A0 is obtained, the calendar iC2 becomes 1
The minute time is obtained by a value based on the clock by the CPU crystal oscillator 3 (S32). Here, the free-run counter value B0 is set to the next value A1 of A for the next interrupt (S33). Since the value of C includes an error of the crystal oscillator 3 for the microcomputer, the error value 7 of the crystal for the microcomputer stored in the backup memory 7 in advance is used.
Correct by subtracting a. The result is stored in D (S34).

Next, 60000 is subtracted from the value of D in order to determine the error of the calendar crystal. The subtracted remaining E is the error that caused the calendar iC in the one minute (S35).
Here, 60000 means [exact one minute = 60000 counts]
, That is, if calendar i
If C2 operates with zero error, It is. Therefore, if there is this error E (S36, branch Y), this error E is integrated and set as F (S37). Then, when the error integrated value F reaches the minimum unit in which the time can be changed (S38, branch Y), the time is corrected.
As a procedure of the time correction, the time is advanced if the time is too late (S39 → S40), and if the time is almost late, the time is advanced (S39)
39 → S42). For example, if the time of the calendar iC being used can be changed in seconds, the error integrated value is 1 second (10
(00 count), the time is corrected.

Thereafter, the minimum time change unit is added or subtracted from the error integrated value F (S41, S43). That is, if the time is delayed by 1 second and the time is corrected, subtract 1000 and advance 1 second and time 4) If corrected, add 1000. An error less than the minimum unit is left as 5) in the error integrated value. Then, the time is read from the calendar iC2, and an interrupt is generated at the time of the next "minute" carry (S44).
From the correction processing program of FIG.
Return to 3). The correction processing program of FIG. 12 runs again at the next interruption at the time of “minute” carry.

[0036]

According to the present invention, a calendar iC2 which performs a timekeeping operation by an output clock of a crystal oscillator 4 for a calendar iC,
2. A system comprising: a CPU 1 driven by an output clock CLK of a U crystal oscillator 3;
According to the inventions according to the fourth to fourth aspects, when the error of the frequency of the CPU crystal oscillator 3 with respect to the reference oscillation frequency cannot be ignored, the battery backup RAM 7 that stores the frequency error data 7a of the CPU crystal oscillator 3 is provided. The CPU 1 corrects the time measured by the free counter 5 for counting the output clock CLK of the crystal oscillator 3 for the CPU with reference to the time measured by referring to the frequency error data 7a. Was corrected, so that a calendar error could be reduced. It has been confirmed that the microcomputer frequency corrected for the oscillation frequency error of the calendar crystal 4 of -56 ppm (day difference 4.8 seconds) has been improved to +3.8 ppm (day difference 0.3 seconds). According to the correction method of the present invention, the error after the correction is the error of the crystal for the microcomputer,
Depends on the rounding error when writing.

In general, there are microcomputer systems of various performances, but the execution time Texe does not become zero no matter how high the microcomputer system. Some microcomputer systems have a long execution time due to cost constraints.
However, according to the invention according to claim 4 included in the present invention, the count value read from the free-run counter 5 once each time is synchronized with a one-minute interrupt request by the calendar iC2. Since the calendar error is obtained by using the microcomputer, the calendar error can be correctly recognized regardless of the length of execution time of the microcomputer. It can be applied to microcomputer systems.

[Brief description of the drawings]

FIG. 1 is a conventional system configuration diagram.

FIG. 2 is a flowchart for explaining an operation of a main part of the CPU of FIG. 1;

FIG. 3 is a block diagram of another conventional system.

FIG. 4 is a diagram showing general frequency-temperature characteristics of a crystal oscillator for calendar iC.

FIG. 5 is a diagram showing general frequency temperature characteristics of a crystal oscillator for a CPU.

FIG. 6 is a system configuration diagram as an embodiment of the invention according to claims 1 to 3;

FIG. 7 is a flowchart showing a first embodiment of a correction process of the CPU in FIG. 6;

FIG. 8 is a diagram showing an example of variation in frequency temperature characteristics of an expensive CPU crystal oscillator.

FIG. 9 is a diagram showing an example of variation in frequency temperature characteristics of a relatively inexpensive crystal unit for a CPU.

FIG. 10 is a time chart showing the timing of reading a free-run counter based on FIG. 7;

FIG. 11 is a flowchart showing a main process as a second embodiment of the correction process of the CPU in FIG. 6;

FIG. 12 is a flowchart showing details of an interrupt process started by the main process of FIG. 11;

FIG. 13 is a time chart showing the timing of reading a free-run counter based on FIGS. 11 and 12;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 CPU 2 Calendar iC 3 Crystal oscillator for CPU 4 Crystal oscillator for calendar iC 5 Free-run counter 6 Memory 7 Battery backup RAM 7a Crystal error for CPU iNT interrupt signal

Claims (4)

(57) [Claims] 1. A system comprising: a calendar iC for performing a timekeeping operation by an output clock of a crystal oscillator for a calendar iC; and a CPU driven by an output clock of a crystal oscillator for a CPU. Means for storing frequency error data of the oscillator, wherein the CPU corrects the time measured based on the output clock of the crystal oscillator for the CPU with reference to the frequency error data (hereinafter referred to as the first time). A time measurement of the calendar iC (hereinafter referred to as a second time measurement) on the basis of the time measurement of the calendar iC. 2. The calendar iC correction method according to claim 1, wherein the CPU accumulates an error of the second clock time with respect to the first clock time at predetermined intervals, and the integrated value of the error is time. The calendar iC correction method, wherein the correction is performed each time the minimum unit of change is reached. 3. The calendar iC correction method according to claim 1 or 2, wherein the first clocking time is equal to the CP time.
A calendar iC correction method, wherein a count value of a counter for counting an output clock of a U crystal oscillator is clocked by performing the above correction. 4. The calendar iC correction method according to claim 3, wherein the CPU is configured to synchronize the calendar iC with an interrupt request generated at predetermined time intervals measured by the calendar iC.
A method for correcting the calendar iC by reading the count value of the counter once each time.