JP2000305750A - On-vehicle electronic controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an on-vehicle electronic controller preventing the deteriora tion of precision or the decrease of a processing speed at the time of storing an arithmetic result in an accumulator, and operating the numerical calculation of three or more numerical data whose LSB is different according to four arithmetical operations. SOLUTION: This on-vehicle electronic controller is provided with a microcomputer. In this case, a signal from a sensor or a switch is inputted, and each kind of arithmetic operation is executed, and an actuator is driven based on the arithmetic result. The arithmetic result is stored in an accumulator, and the numerical calculation of three or more numerical data is operated according to arithmetical operations. The arithmetic result obtained in the middle of the numerical calculation of the numerical data is shifted so that the leftmost side bit in the accumulator can be '1' in a step 202. The number of times of shift is managed and held in a step 203, and shift adjustment is operated in a step 204 after the final arithmetic operation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an on-vehicle electronic control unit, and more particularly to a technique for enabling high-precision arithmetic processing.

[0002]

2. Description of the Related Art A 16-bit microcomputer is used as a vehicle-mounted controller (ECU) for performing engine control and the like. In many cases, the arithmetic operations performed by the engine control are complicated in the four arithmetic operations, and at the same time, accuracy is required. Conventionally,
In such a case, the intermediate calculation result is saved in the RAM, and
It is necessary to perform the byte operation or perform the operation in consideration of the order of the four arithmetic operations so that the number of shifts is as small as possible and the guard is not applied during the operation.

[0003]

The numerical data (captured values and constants) used in the engine control are the number of bytes, L
The SB, offset, range, and the like vary depending on the application. If the range of use of numerical data is very narrow, it is possible to secure a certain degree of calculation accuracy in consideration of the order of the four arithmetic operations. However, in the case of engine control, the value may vary greatly depending on the adaptation, and therefore, most do not have a range. In this case, the calculation on the program is performed for the entire area indicated by the defined number of bytes, that is, for one byte, $ 00
(Hereinafter, in order to indicate that it is a hexadecimal number, ＄
To make it a hexadecimal expression. Also, add ₍₁₀₎ after the numerical value to indicate that it is a decimal number, and ₍₂₎ after the numerical value to indicate that it is a binary number _). It is necessary to obtain the calculation accuracy for the area of ＄ FFFF.

As a method of increasing the operation accuracy, there is the above-mentioned prior art. However, when performing a 4-byte operation, there is a problem that the logic configuration is complicated and the processing time and the number of bytes are greatly increased. In addition, there is a limit in considering the order of the four arithmetic operations in order to minimize the number of shifts during the operation or to prevent a guard from being applied during the operation. There is a problem that the calculation error becomes very large due to the digit loss or rounding off.

[0005] Two parameters used in the four arithmetic operations are:
Each has an LSB, and the calculation result obtained from these two parameters also has an LSB. For example, for both parameters, the number of bytes is 2, the LSB is 2 ^-2 , and the value is HEX.
To add the value $ 8000, simply add $ 8000 + $
$ 10000 obtained from 8000 is 3 bytes and LSB
Results in ^2-2 .

If the result can be used as it is in the subsequent operation, the precision does not decrease, but the result on the register often needs to be 2 bytes. Generally, $ 10,000
Is set to 2 bytes, there are a method of guarding with $ FFFF, and a method of shifting once to the right to make $ 8000 of LSB2 ^-1 . In the former case, the accuracy does not decrease because the LSB does not change, but when guard is applied, the subsequent calculation results are greatly affected. If there are specification conditions where the commonly used values of the two parameters are small and the addition result is rarely guarded, the guarding method will be fine. In the latter case, since no guard is applied, the subsequent calculation result is not significantly affected, but LS is always used.
The precision of B is reduced by one, and a delicate error is given to the calculation result thereafter.

More specifically, as shown in FIG. 15, in an arithmetic expression K = ka × kb × kc, the substitution value ka = 1.00
003 ₍₁₀₎ , kb = 1.00012 ₍₁₀₎ , kc = 1.0
In the case of ₍₁₀₎ , the theoretical value K = 1.00064
₍₁₀₎ . In hexadecimal notation, the number of bytes of ka, kb, kc, K is all set to “2”, and the L of ka, kb, kc, K is
SB, ka is 2 ^-15 , kb is 2 ^-13 , kc is 2 ^-11 ,
When K is set to 2 ^-15 , ka = ＄ 8001, kb =
＄ 2001, kc = ＄ 0801, K = ＄ 8015.

Then, as shown in FIG. 16, as a calculation pattern 1, ka × kb is calculated first, and the calculation result is multiplied by kc. FIG. 17 shows the operation status on the register at this time. In FIG. 17, ka
= $ 8001 and kb = $ 2001 multiplied by ka x
kb = $ 1000A001, and in order to make it 2 bytes, the lower 2 bytes of the operation result are rounded off to $ 1001. Then, the value kc (= ＄ 08
01), ka × kb × kc = ＄ 80180
It becomes 1 and if this is 2 bytes, $ 8018 = K is obtained. The value thus obtained is the theoretical value K = ＄ in FIG.
8015 has $ 03 as an error.

Thus, three parameters (ka, k
b, kc) have different LSBs, and if a numerical value is determined so that the weight of each 1 LSB can be accurately reflected in the final result, it is necessary to always consider the LSB, and furthermore, the maximum parameter Since the value also needs to be taken into account, when the first multiplication result is held in 2 bytes, a carry due to rounding occurs, causing a large error in the final result.

Compared with the calculation procedure of FIG. 16, first, as shown in FIG. 18, as a calculation pattern 2, kb × kc is calculated, and the calculation result is multiplied by ka.
FIG. 19 shows the operation status on the register when the order of the operations is changed. In this case, kb × kc = ＄ 10028
01, which is 2 bytes, which is $ 100. Further, $ 100 × ka = $ 800100, which is 2 bytes, which is $ 8001. As a result, ka × kb ×
kc = ＄ 8001 = K is obtained. At this time, $ 100280
When 1 is set to 2 bytes and set to $ 100, significant digit loss occurs. Therefore, the value obtained in this way has ＄ 14 as an error with respect to the theoretical value K = ＄ 8015 in FIG. 15, and as can be seen by comparing FIGS. I will.

SUMMARY OF THE INVENTION It is an object of the present invention to prevent a decrease in accuracy and processing speed when calculating three or more numerical data having different LSBs in accordance with four arithmetic rules while storing the calculation result in an accumulator. It is an object of the present invention to provide an in-vehicle electronic control device capable of performing the following.

[0012]

According to the first aspect of the present invention, the operation result in the middle of numerical calculation of numerical data is such that the leftmost bit in the accumulator is "1". And the number of shifts is managed while being managed, and the shift is adjusted after the final calculation.

That is, when selecting the bits that can be stored in the accumulator among the intermediate calculation results, the higher-order bits of the intermediate calculation results are stored in the accumulator, and the subsequent calculation is performed.

Therefore, an increase in a calculation error due to loss of digits during calculation or rounding is suppressed. Further, the logic configuration is simplified and the processing time and the number of bytes are suppressed from being increased as compared with the case of performing the conventional 4-byte operation (the method of continuing the calculation while saving the data in the RAM).

In this way, the accuracy and the processing speed can be prevented from being reduced when calculating three or more numerical data having different LSBs in accordance with the four arithmetic rules while storing the calculation result in the accumulator. Become.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. In the present embodiment, the present invention is applied to an automobile engine control system. FIG. 1 shows a vehicle-mounted engine control E as a vehicle-mounted electronic control device.
1 shows a configuration of a CU (ECU; Electric Control Unit) 1.

The ECU 1 includes an A / D converter 2, an input processing circuit 3, and an MPU (16-bit microcomputer).
4, an output processing circuit 5, and a power supply circuit 6. MPU
Reference numeral 4 captures analog signals such as a battery voltage, an engine coolant temperature signal, an engine intake temperature signal, an engine intake pressure signal, and a throttle opening signal via the A / D converter 2. Further, the MPU 4 receives a crank position signal, an engine rotational speed signal (pulse signal), an O ₂ signal via the input processing circuit 3.
It takes in digital signals such as sensor signals, vehicle speed signals, idle signals, air conditioner switch signals, starter signals, ignition switch signals, and the like. Here, the input processing circuit 3
MPU4 measures the pulse period of the engine speed signal (pulse signal) and generates a signal corresponding to the engine speed.
To send to.

The MPU 4 calculates an optimum fuel injection amount, ignition timing (advance angle), and ISCV opening in accordance with the engine operating condition based on the signal indicating the engine condition, and outputs an output based on the calculation result. Actuators such as an injector, an igniter, and an idle rotation speed control step motor are driven and controlled via the processing circuit 5. MPU
Reference numeral 4 includes a CPU (central processing unit) 7, a RAM (read only memory) 8, and a ROM (random access memory) 9. The power supply circuit 6 converts the battery voltage to a 5-volt voltage and supplies it to each device in the ECU 1.

Thus, in the engine control system,
While performing input processing such as taking in switch inputs, A / D conversion of sensor signals, and pulse period measurement,
The advance amount and the ISCV opening are calculated, and output signals are sent to those actuators at a predetermined timing. These processing procedures are stored in a memory (ROM) 9 in the MPU 4 and are sequentially transmitted to the CPU 7 as instructions and executed.

FIG. 2 is a flowchart showing an outline of the control. First, in step 100, the initial functions are set,
In steps 101, 102, and 103, input processing of each sensor switch includes A / D conversion, input of a switch signal,
Calculate the engine speed. And step 10
In steps 4 and 105, an injection correction value is calculated and an injection value is output as a fuel injection process. Here, regarding the fuel injection processing, the fuel injection control is performed to obtain an optimal air-fuel ratio from signals from the sensors that detect the state of the engine for the purpose of improving the power performance of the engine, purifying exhaust gas, improving fuel efficiency, improving drivability, and the like. In this way, the fuel amount required for the engine is calculated and the injection signal is sent to the injector.

Further, in steps 106 and 107, an ignition timing correction value is calculated and an ignition signal is output as an ignition timing process. Here, regarding the ignition timing process, the ignition timing control is a control for determining an optimum ignition timing from signals of the respective sensors for the same purpose as the fuel injection control and sending an ignition signal to the igniter.

FIG. 3 shows a functional block diagram of the fuel injection control. The input signal processing unit 10, the injection correction calculation / injection amount calculation unit 11, and the injection output unit (timer control unit) by the MPU4
12 are configured. Analog / digital signal is A
The signal is input to the input signal processing unit 10 via the / D converter 2 and the input processing circuit 3, where the signal processing is performed, and then the injection correction calculation / injection amount calculation unit 11 performs a desired operation. Output processing is performed by the injection output unit (timer control unit) 12 based on the calculation result, and the injector is driven via the output processing circuit 5. As a result, an optimal amount of fuel is supplied to the engine from the injector.

In the calculation for the fuel injection control,
The calculation is performed using the parameters shown in FIG. Specifically, the final injection time ti is obtained from the effective injection time tau and the invalid injection time tv by ti = tau + tv (1), and the effective injection time tau is Is required. Here, tp is a basic injection time obtained from an output signal of an air flow meter or the like for realizing the target air-fuel ratio, ftha is an intake temperature correction coefficient, ftotal is a total increase correction coefficient, and kF
IXF is a correction coefficient for matching.

That is, the final injection time ti is obtained by adding the invalid injection time tv to the effective injection time tau, and the effective injection time tau is obtained by performing various corrections on the basic injection time tp. tv is the time from the start of energization to the injector to the actual injection of fuel. Further, the basic injection time tp is obtained as an injection time corresponding to a fuel injection amount per injection so as to obtain a stoichiometric air-fuel ratio, and the intake air temperature correction coefficient ftha is determined based on the temperature of the intake air. Since it changes, the correction is performed. Further, the total increase correction coefficient ftotal is based on the basic injection time tp, and various corrections of the injection amount are performed by changing the injection time for the purpose of improving the warm-up characteristics, improving the acceleration performance, purifying the exhaust gas, etc. based on signals from the sensors. Correction coefficient k for matching.
FIXF is a correction term used at the time of matching.

Here, the number of bytes, LSB, and offset are determined for each time and correction coefficient. Specifically, ti, tau, tv, and tp have the number of bytes “2” and L
SB is “4 μs”. ftha, ftotal "2" number of bytes, LSB is 2 ^{-1 1.} Furthermore, kF
The IXF has a byte number of “2”, an LSB of 2 ⁻⁸ , and an offset value of 0.5.

As described above, the number of bytes, the LSB, and the offset are determined for each of the plurality of parameters, and the equations (1) and (2) for obtaining the desired result are defined. FIG. 5 shows the configuration of the CPU 7.

As shown in FIG. 5, the CPU 7 performs an arithmetic operation (ALU: Arithmetic Log) for performing arithmetic operation and logical processing on data.
ical Unit) 20, register group 2 for temporarily storing data
1. The control unit 22 controls operations such as decoding of program instructions. The arithmetic unit 20 numerically calculates three or more numerical data according to the four arithmetic rules while storing the operation result in the accumulator of the register group 21.

FIG. 6 shows registers in the register group 21 of FIG. Accumulator D and accumulator E
And an index register Y. Each accumulator D, E and index register Y are 2 bytes (1
6 bits). The accumulator D includes a 1-byte accumulator A and a 1-byte accumulator B.

7 and 8 show a general operation method.
In FIG. 7, parameter 1 is 2 bytes and parameter 2 is 2 bytes. When multiplication of 2 bytes × 2 bytes is performed to obtain a 4-byte operation result, parameter 1 is stored in accumulator E (E register) and , Parameter 2 are stored in accumulator D (D register), the upper two bytes of the multiplication result are stored in accumulator E (E register), and the lower two bytes are stored in accumulator D (D register).

Here, as shown in FIG.
Is 2 bytes, and the data usage range is $ 00
00 to $ FFFF, parameter 2 is 2 bytes, and the data usage range is $ 0000 to $ 08
If the value is 00, the calculation result becomes $ 07FFF800 at the maximum, and the upper two bytes of the multiplication result are stored in accumulator E (E register) and the lower two bytes are stored in accumulator D (D register). When this calculation result is expressed in a binary system, 0000011111111111111
1100000000000 ₍₂₎ , and the maximum value that can be taken as data is 1111111111111111.
It turns out that it is ₍₂₎ .

FIG. 9 shows a flowchart when this calculation method is adopted. In FIG. 9, an operation is performed in step 200. That is, multiplication (MUL) or division (D
IV), and temporarily holds the operation result on registers (E and D registers) (accumulators E and E for 4 bytes).
D). And. In step 201, it is determined whether or not the final calculation is completed.
2 performs the maximum effective numerical value. That is, the 4-byte operation result is stored as 2 bytes. At this time, in the D register of the E and D registers, as shown in FIG.
Shift so that the leftmost bit becomes 1. In the calculation example of FIG. 10, $ 8001 * $ 2001 = $ 100
0A001 is obtained, and the 4-byte data is converted into 2-byte data.

Further, in step 203 of FIG. 9, the left and right sides of the shift are positioned with positive and negative, and the number of shifts is shown in FIG.
As shown in FIG. 10, it is held in a 2-byte index register Y (Y register).
D is stored). Thereafter, the process returns to step 200 in FIG.

Steps 200 to 203 are repeated,
If the final computation is completed in step 201, step 20
4 and the operation result of 4 bytes in FIG.
As shown in 2, computing LSB (Figure 11 and 12 in 2 ^-39) and the target LSB (also 2 ^{-1 5),} the number of shifts (also 13), and finally the number of times of shifting (also X = 11) Is calculated, and the calculation result at the target LSB is obtained (LSB adjustment is performed). The operation LSB is an LSB resulting from a simple operation ignoring the number of bytes, and the target LSB is an LSB to be finally stored.

Thereafter, the result is output in two bytes in step 205 of FIG. Hereinafter, the processing of FIG. 9 will be described in detail. In step 200 of FIG. 9, the calculation is started in accordance with the rules of the four arithmetic operations, and the operation result is held in the E and D registers in 4 bytes. Convert to bytes.

At this time, as shown in FIG. 10, the LSB of the two parameters is not taken into account, and the effective data (maximum physical quantity) that can be held by 2 bytes of the D register is
Shift and convert byte results. In the case of FIG. 10, it shifts 13 times to the right. The number of shifts is stored in the Y register. In the case of FIG. 10, 13 _(2), that is, ＄ D
Is stored in the Y register.

If the operation is not the final one, the above calculation is repeated in step 203 of FIG. 9 by sharing the Y register storing the number of shifts. Therefore, the number of shifts indicates how many times the shift has been made to the left or right in total.

When the operation is final, in step 204, LSB adjustment is performed on the result of 4 bytes from the number of shifts, the operation LSB, and the target LSB, and the result is output in 2 bytes as shown in FIG.

That is, as shown in FIG. 11, the intermediate calculation result is calculated while holding the effective data (the maximum value of the physical quantity) in the E and D registers without considering the LSB, and finally adjusting the LSB. Thus, the calculation accuracy is maintained to some extent, the execution processing time is relatively short, and it is not necessary to consider the calculation order.

Hereinafter, description will be made with reference to specific numerical values. When it is necessary to continue the calculation after the calculation illustrated in FIG. 10, it is necessary to store data of 2 bytes or less on the E and D registers for the next operation. As described with reference to FIG. 8, when the calculation result (theoretical maximum value) in the case where the maximum value is included in each of the two parameters is considered, each of the parameters is assumed to be $ FFFF.
If there is a range up to, 4-byte operation result is $ FF
FE0001. In this case, the only way to convert the 4-byte operation result to 2 bytes is to take the upper 2 bytes (when guarding is not performed), but there is almost no such case (in terms of control specifications).

Considering that this will be the case in software calculations,
If the upper two bytes are fixed for use in the next operation, the operation result is very small with the values of the parameters normally used, and large information in the lower byte is discarded.

Therefore, in order to make maximum use of data (information), "1" standing at the leftmost position in the E and D registers in FIG. 10 is replaced by the most significant bit (bit 15) of the D register. Then, the number of shifts is stored in the Y register. In the example of FIG. 10, since "1" is set at the 12th bit of the E register, the data is shifted 13 times to the right to bring it to the 15th bit of the D register.

The Y register storing the number of shifts is
It is used continuously until the final calculation, and the total number of shifts and
And save. Explanatory drawing of the calculation method after the final calculation of FIG.
, The operation result is $ 0400A805,
LSB is 2^-39And the number of shifts is thirteen,
The target LSB is 2^-15It is. In other words, the operation LSB is calculated by
Is obtained simply regardless of the number of bytes according to
The result of the byte operation is stored in the E and D registers.
Is the LSB of the calculation LSB shifted to the right by 13.
In the case of FIG. ^-26Becomes Furthermore, the target LS
B (finally desired LSB) is the top 2
The result is the litter.

Based on the description of FIG. 11, the final shift number for converting the 4-byte operation result into a 2-byte result (D register) is calculated from the operation LSB, the target LSB, and the shift number. Become like That is, since the operation LSB is ^2-39 and the shift number is 13,
From −39 + 13 = −26, the LS of the 4-byte operation result
B = 2 ^-26 is obtained. Further, in order to calculate the final shift number X, the solution of the equation −26 + X = −15 is obtained.
11 and the number of final shifts X = 11, that is, “＄ 0
Shift 400A805 right 11 times ". Therefore, when shifting right 11 times from the state of FIG.
2 and the final result of 100000000001
0101 ₍₂₎ is required.

FIGS. 13 and 14 show the results of error analysis with respect to theoretical values when specific numerical values are set and calculated by the present calculation method. The error with respect to the theoretical value between the case where the calculation is performed by the conventional calculation method of FIGS. 15 to 19 and the case where the calculation is performed by the present calculation method of FIGS.

As shown in FIG. 15, three parameters (k
a, kb, kc) have different LSBs,
Numerical values are determined so that each 1 LSB weight can be verified to be accurately reflected in the final result. 15 to 17
In the conventional method (non-four bytes), it is necessary to always consider the LSB and further consider the maximum value of the parameter. The carry due to drop or rounding occurs and causes a large error in the final result.

On the other hand, in the case of the present calculation method, as shown in FIG. 13, since the value is held as large as possible in the holding stage of 2 bytes, the effect of the digit loss is very small, and the result is hardly affected. Not. FIG. 18 also shows the results when the order of the operations is changed (operation patterns 1 and 2). In the conventional method, the results differ depending on the order of the operations. The method gave the same result. Based on the rules of the four arithmetic operations, there is no need to consider the order of the operations. In other words, in the conventional method, the carry due to rounding occurs, which causes a large error in the final result. When $ 100281 is converted into 2 bytes to $ 100, a significant digit drop occurs and the error occurs. ＄ 14, and the result differs depending on the order of operation. However, in this calculation method, the error is reduced and the result does not differ depending on the order of operation. In other words, in the conventional method, the calculation is continued while taking the LSB into account every time the multiplication is performed. In the present calculation method, the LSB may be checked at the end of the calculation, so that the calculation load is reduced. be able to.

Although the result of the conventional 4-byte operation (a method of continuing the calculation while saving data in the RAM) is not shown, when the parameter is a very small value, for example, when 2 bytes are $ 0001 However, in most cases, no matter how many bytes are used to hold the calculation result, useless processing is performed in consideration of the number of bytes as a request. Furthermore, a multiplication / division instruction takes a particularly long execution time, and in the case of performing a 4-byte operation, it is necessary to perform multiplication / division several times, which causes a problem of processing load. On the other hand, the shift instruction has the shortest execution time, and does not take as much processing time as the 4-byte operation even if it is performed a number of times.

As described above, the present embodiment has the following features. (A) In step 202 of FIG. 9, the operation result during the numerical calculation of the numerical data is shifted so that the leftmost bit in the accumulator D becomes "1" as shown in FIG. In step 203 of FIG. 9, the number of shifts is managed and maintained.
Since the shift is adjusted as shown in FIG. 12 after the final calculation in step 04, it is possible to suppress an increase in calculation errors due to loss of digits in the middle of the calculation or rounding, and to reduce the conventional 4-byte calculation (RAM In this case, the logic configuration is simplified and the processing time and the number of bytes are suppressed from increasing as compared with the case of performing the calculation while saving data. As a result, the accuracy and the processing speed can be prevented from being reduced when the three or more numerical data having different LSBs are numerically calculated according to the four arithmetic rules while the calculation result is stored in the accumulator.

In the above description, the shift is to the right, but the shift may be to the left. That is, in the case of FIG. 10, since the D register (register corresponding to the lower bit) was used to convert the 4-byte multiplication result into 2 bytes, the data was shifted to the right. When using the E register (register corresponding to the upper bit) to make the result 2 bytes, the shift may be to the left. However, it is necessary to determine whether the left or right during the left shift is positive or negative.

[Brief description of the drawings]

FIG. 1 is a vehicle engine control ECU according to an embodiment;
FIG.

FIG. 2 is a flowchart showing an outline of control.

FIG. 3 is a functional block diagram of fuel injection control.

FIG. 4 is a diagram for explaining calculation using parameters.

FIG. 5 illustrates a structure of a CPU.

FIG. 6 is a diagram showing registers in a register group.

FIG. 7 is a diagram for explaining multiplication of 2 bytes × 2 bytes.

FIG. 8 is a diagram showing a data range and calculation results.

FIG. 9 is a flowchart of the present calculation method.

FIG. 10 is a diagram illustrating the present calculation method.

FIG. 11 is a diagram for explaining the present calculation method.

FIG. 12 is a diagram for explaining the present calculation method.

FIG. 13 is a diagram showing an error analysis result with respect to a theoretical value according to the present calculation method.

FIG. 14 is a diagram showing an error analysis result with respect to a theoretical value by the present calculation method.

FIG. 15 is a diagram illustrating an operation.

FIG. 16 is a diagram illustrating an operation.

FIG. 17 is a diagram illustrating an operation.

FIG. 18 is a diagram illustrating an operation.

FIG. 19 is a diagram illustrating an operation.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... In-vehicle engine control ECU, 2 ... A / D converter, 3 ... Input processing circuit, 4 ... MPU (microcomputer), 5 ... Output processing circuit, 6 ... Power supply circuit, 7 ... CPU
(Central processing unit), 8 ... RAM (read only memory), 9 ... ROM (random access memory), 10 ...
Input signal processing unit, 11: injection correction calculation / injection amount calculation unit,
12: injection output unit (timer control unit), 20: calculation unit (A
LU), 21: register group, 22: control unit.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Tomoyuki Kashiwagi 1-1-1, Showa-cho, Kariya-shi, Aichi F-term in DENSO Corporation (reference) 3G084 BA00 BA03 BA06 BA13 BA17 DA06 EA07 EB03 EB07 FA00 FA02 FA03 FA05 FA07 FA10 FA11 FA20 FA29 FA33 FA35 FA36 FA38 5B022 AA00 BA02 CA01 CA03 DA01 FA06 9A001 BB02 HZ34 KZ32

Claims (1)

[Claims] A microcomputer is provided, which performs various calculations by inputting signals from a sensor or a switch and drives an actuator based on the calculation results, and stores the calculation results in an accumulator according to four arithmetic rules. An on-vehicle electronic control device that numerically calculates one or more numerical data, wherein the leftmost bit in an accumulator of a calculation result in the middle of the numerical calculation of the numerical data is “1”. The electronic control unit for a vehicle according to claim 1, wherein the shift control is performed while the number of shifts is controlled and the shift is adjusted after the final calculation.