JPS5870055A - Knock control device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a knock control device, and particularly to a knock control device that does not affect engine/engine characteristics due to knock-like noise.

Knock that occurs in an engine is accompanied by a knocking sound that reduces running performance, and also causes a reduction in engine output due to generation of reverse torque or damage to the engine due to overheating. This knock has a close relationship with the ignition timing, and it is known that engine output can be maximized by setting the ignition timing, that is, the ignition advance angle, just before the knock due to the characteristics of the engine. Therefore, reducing the ignition advance angle in order to avoid the occurrence of knock will conversely reduce the engine output, so it is required to control the ignition timing immediately before the occurrence of knock. In particular, turbocharged engines have high compression ratios and require optimal ignition timing to maintain maximum efficiency.

In the conventional knock control device, a method is adopted in which the ignition noise caused by the ignition is masked in the signal from the knock sensor, the knock signal is detected, and the knock number 1g is used to retard the engine by a certain amount. . In other words, it retards a certain amount (for example, 0.4 degrees) due to the knock that occurs between the ignition timings, and determines whether or not to retard depending on whether or not there is a knock signal, and retards the engine by a certain amount in steps when the knock occurs. (so-called step retard) method was adopted.

However, when knocking occurs due to coupling, the conventional method has the disadvantage that the control response to knock is poor due to the small amount of retard (one step), resulting in extremely poor running performance. was. On the other hand, a method can be considered in which the amount of retard is increased per step, but the amount of retard is increased with only one knock signal.

It had the disadvantage of not using enough torque.

Therefore, in recent years, knock control devices have been developed that appropriately control knocking by starting the number of knocking pulses of a knocking signal in proportion to the total pulse width.

This latest knock control device will be explained next.

FIG. 1 shows the overall configuration of a conventional knock control device.

In the figure, the knock control device of the present invention includes a knock sensor 100 for detecting a knock signal, and a knock sensor 10 for detecting a knock signal.
The ignition coil 135 is activated by the knock signal input from 0.
0 A knock control device 101 outputs a control signal to control the ignition timing, a recovery coil 105 detects the spark timing of the ignition coil 135, and an ignition coil is ignited by the output from the pickup coil 105 and the knock control device 101. It also includes a non-contact ignition device 103 that sends a feedback signal to the knock control f-releasing device 101.

The knock control device 101 takes in the detection signal of the knock sensor 100 and the output signal of the non-contact ignition device 103, and controls the non-contact ignition device 103 in response to knocking to perform advance or retard control.

The knock control device 101 is a knock sensor 1o.

a fail-safe device 102 that detects a failure in the engine and sends a signal to forcibly retard the ignition timing; an AC coupling circuit 112 that removes the DC component; and an AC coupling circuit 112 that cuts ignition noise in synchronization with the spark timing. ignition noise cut circuit 1/3 having a gate; band pass filter (BPF) 114 for bandpassing the knock signal;
An automatic gain control circuit (AGC) controls the gain of its own amplifier in proportion to the input signal ratio using the output of the BPF114.
circuit) 160, a mask circuit 150 for masking a section at a predetermined timing with respect to the AGC output, a background level (BGL) for obtaining the average value of the knock signal from the input from the AGC circuit 160 via the mask circuit 150;
Detection circuit 119, an amplifier 161 for amplifying the output of the BGL detection circuit 119 and feeding it back to the AGC circuit 160, an amplifier 151 for increasing the output of the mask circuit 150, BGT, the output voltage of the detection circuit 119 and the amplifier 1
a comparator 118 that generates a retard signal proportional to knocking by comparing the output signal of the comparator 118 with the output signal of the comparator 118; a mask circuit 1 that masks the output of the comparator 118 at a predetermined timing and outputs the signal;
53. Integrator 15 that divides the output of the 7 times g153
4. A retard signal setting circuit 126 for setting a retard signal proportional to knocking based on the output of the integrator 154. The ignition coil 135 is activated by a signal from the non-contact ignition device 103.
(i.e., power transistor 134
monostable circuit 128. which generates a signal of constant pulses 128. While the output pulse of the monostable circuit 128 is output, the ignition coil 13 is activated at a constant voltage/period rate.
A lead angle signal setting circuit 127 outputs a signal to advance the ignition timing of No. 5, and a retard signal is set in the number setting circuit 126. The output from the fail-safe device 102 causes the ignition coil 135 to generate a proportional DC voltage.
An output circuit 125 and a reference voltage generation circuit 120 send out an output for advancing or forcibly retarding the ignition timing of the
Consists of.

The reference voltage generation circuit 120 includes an operational amplifier OP3, a capacitor C5% resistor 15. Consists of R, 11, R12°R14. By appropriately selecting resistors R14 and R15, the outputs R and V of the operational amplifier OP3 are fixed to the reference voltage. This reference voltage V is, for example, 3V, and is used for one variety of reference voltages.

The failsafe device 102 has a failsafe function that detects various failures or abnormal modes of the knock sensor 100 and performs retard control according to the mode. This device 102 includes a constant current circuit 107, at least a sensor short detection circuit 108, and a sensor open detection circuit 10.
The sensor abnormality detection circuit 102A has a sensor abnormality detection circuit 102A having 9 and 9, and supplies the output of each circuit to an integrating circuit 125 to perform a fail-safe function. BGL detection circuit 119.

It consists of a half-wave rectifier 116, an integrating circuit 117, and an amplifier 117A, and is sent out via a mask circuit 150.
Rectifying and integrating the output signal of the GC circuit 160,
Averaging of the entire signal/No. 1a is obtained. This averaged signal becomes the BGL signal. Comparison circuit 118 includes at least two comparators 118A.

It has 118B. The comparator 118A compares the BGT output with the output of the amplifier 151, and the comparator 118B compares the output of the amplifier 151.
The output of 1 is compared with the reference voltage.

Non-contact ignition bag u6” 103 is pickup coil 1
An amplifier 131 that shapes the waveform of the output signal of 05 adjusts the ignition timing to 11ilI according to the output voltage of the knock control circuit 101
The retard circuit 132 includes a power transistor 134 that generates a high voltage on the secondary side of the ignition coil 135.

Next, the operation of the above configuration will be explained based on the waveforms shown in FIG. Figure 2 (1) shows one ignition timing waveform, and in reality, this waveform signal is
This is the base signal of the power transistor 134. II
The power transistor 134 is on (ON) at the level,
Power transistor 134 is OFF at L level
becomes. Sparks in the ignition coil 135 are generated during the switching process from ON to OFF. The signal s2 in FIG. 2(2) is a constant width pulse output signal of a monostable circuit 128 which receives the base signal as input and generates a pulse signal of a constant width (tl) by being triggered when turning from ON to OFF.

Figure 2 (3) shows the upgraded knock sensor 100.
shows the output of The signal detected by the knock sensor 100 is a signal that goes positive or negative by one level with the DC zero (0) level as a reference. The detection signal includes a knock signal. On the other hand, in order to enable failsafe, the detection circuit 108
, 109 are provided.

In the failure mode of these detection circuits, the mode cannot be detected if the DC zero level signal remains. As a countermeasure against this, a constant current circuit 107 is provided. A constant current from the constant current circuit 107 is superimposed on the output of the sensor 100, thereby providing a constant positive bias.

This constant current circuit 107 plays an important role. That is, the output of the knock sensor 100 changes between 100 mV and 600 mV. The range of change is 120 times. 1201 When attempting to adjust the knock sensor output to pRH using a general bias means using resistor partial pressure with respect to the change width of the sound, for example, if the knock sensor output is close to ±5 mV and the knock sensor output is attenuated by the bias means. If the ratio is reduced to 1/20, the output after biasing will drop to 0.25 mV. This obviously leads to a decrease in detection accuracy, which in turn makes it impossible to detect the failure mode. By providing the constant current circuit 107, the present invention improves knock detection accuracy and facilitates failure mode detection.

Figure 2 (3) is constant iR times! The output is biased by +8107.

AC coupling circuit 112 is provided to remove DC components. Furthermore, it may play a role in causing resonance in the knocking frequency region. By removing the DC bias in the AC coupling circuit 112, it makes a significant contribution to improving the noise cutting performance in the ignition noise cutting circuit 113 at the next stage.

This is because ignition noise cannot be cut unless the ignition noise is cut after removing the DC bias. The ignition noise cut circuit 113 is the second monostable circuit 128.
It is controlled by the pulse output as shown in the figure.The noise cut circuit 113 is a type of AND gate, and inputs the output pulse of the monostable circuit 128.Therefore,
Only in the t3 period when the output pulse of the monostable circuit 128 is 0'1'', the output obtained via the AC coupling circuit 112 is almost unchanged from the ignition noise cut circuit 113.
The output is This output is shown in FIG. 2 (4).

Ignition noise is cut by this.

The BPF 114 emphasizes the knock signal (attenuates other signals) and outputs it, and has a characteristic of being slightly attenuated at frequencies higher than the knock signal of knocking.

AGC circuit 160 connects BGL circuit 1 via amplifier 161.
19, it changes its own gain in inverse proportion to the feed bank signal, ie, the BGL output. The mask circuit 150 masks the output of the AGC circuit 160 at a predetermined timing.

This masking is performed by the pulse signal S2 shown in FIG. 2(2). In response to the output of this mask circuit 150, the BGL detection circuit 119 performs horizontal use of BGL.

The comparison circuit 118 is 13GT of the BGL detection circuit 119.

A comparator 118A compares the output (voltage) with the output of the amplifier 151, and a comparator 118B compares the output of the amplifier 151 with a reference voltage. The state of both signals at this time is the 2nd m
Shown in (5). /The comparator 118A of the comparison circuit 118 is
Only the output of the amplifier 151 that is larger than the BGL output is shaped and output. Further, the comparator 118B removes the output of the amplifier 151 at a level higher than the reference voltage applied to the negative terminal (this voltage is a reference value for limiting the abnormal amplitude of the knock signal). This clamps off the abnormally high voltage in the knock signal. This comparator 11
80 mask circuit 153, delay signal output circuit 126
The knock signal is converted into a DC level signal, and the DC level signal becomes the knock signal to be obtained.

Here, the mask circuit 153 has the role of removing noise similar to a knock signal generated from the intake cylinder during high-speed rotation.

This mask signal 810 is generated by the constant time width signal generation circuit 157.
is formed. Furthermore, masking is also performed by S2. The output S6 of the comparator 118 is in the form of a single pulse, and the retard signal output circuit 12 has a function of equalizing this single pulse.
6 has.

FIG. 2(6) shows the knock signal detected by the comparator 118. FIG. 2 (7) is an enlarged view of a portion thereof.

The integrating circuit 125 manually performs predetermined integration on the output signal of the retard signal output circuit 126. Therefore, an integral value is outputted from the selection circuit 125 according to the number of pulses.

The retard angle control of the retard circuit 132 is performed using this integral output.

FIG. 3 shows a specific circuit example of the configuration from the knock sensor 100 to I)I)F114. In the figure, a sensor 100 is an inductive sensor using a magnetostrictive element, and equivalently, it is a series circuit of an inductance L and a resistor R2. Generally, the value of the inductance L is III and the value of the resistor 2 is about 840Ω.

The constant current circuit 107 includes a resistor R1 and a Zener diode Z.
DI, ) transistor T1. Power for the circuit 107 is supplied from a power supply section 120.

The power supply unit 120 includes a resistor 11. 12.几14゜几15
, a capacitor C5, and an operational amplifier OP3. The operational amplifier OP3 has the role of a buffer. operational amplifier OP
The voltage of the B power source is divided by resistors 14 and 15 and applied to the positive end of the switch 3.

The output of operational amplifier OP3 is set to 3(v) and PNP
The collector current of transistor T1 is set at approximately 1.7 mA. Therefore, the DC bias voltage at point a is approximately 1.
6 (V), and the output signal of the knock sensor 00 overlaps with the 1.6 (V) voltage.

The reason for adopting the constant current circuit 107 will be described below.

The output impedance Z of knock sensor 00 is resistance 1
”(2, determined by the value of inductance L. Therefore, the impedance Z at the knocking frequency (approximately 7KH2)
The value is Z=2XfL+R, 2 =2XX7000X1+84O -45(KΩ) ・・・・・・・・・・・・
...(1). Therefore, if the input impedance of the knock sensor connection portion of the knock control device is low, the knock sensor output signal will be significantly attenuated. Knock sensor output is ±
It fluctuates between 120 times the range of 5 (mV) to 600 (mV). A constant current circuit 107 is provided to increase input impedance.

In particular, in the configuration shown in the figure, since the DC bias circuit is of a constant current type, the impedance of the DC bias circuit has a value close to infinity.

On the other hand, if the input impedance of the knock control device is increased, disturbance noise is likely to be superimposed. A typical disturbance noise is ignition noise (1g noise) that occurs in synchronization with one ignition timing.

Next, ignition noise will be explained.

Pace control of the power transistor is performed by pulses as shown in FIG. 2(1). When the pulse is at H level, the power transistor is turned on (ON) and when it is at L level, it is turned off (OF'F). In the process of switching from ON to OFF, or at the time of turning OFF, the secondary voltage of the ignition coil rises rapidly, generating primary noise. Furthermore, this rise in secondary voltage breaks down the insulation of the air layer between the plugs, causing ignition. Secondary noise occurs during this ignition. The second-order noise includes noise due to a capacitive eccentric current flowing at the beginning of one ignition, and noise due to an induced discharge current flowing at a subsequent stage. Among the second-order noises, the former noise is a major noise source. When the input impedance is increased.

The first-order noise and the second-order noise (the former noise) are superimposed on the knock sensor output as disturbance noise that adversely affects knock signal identification.

It is necessary to remove such disturbance noise. This disturbance noise continues for a period of 50 to 60 μs. Therefore, during this time, it is sufficient to mask the way the knock sensor is output. In order to achieve this purpose, an AC coupling circuit 112 and a noise cut circuit 113 are provided. However, the actual mask section is set to a time width that is sufficiently larger than the above noise duration time, for example, 0.8m5e08 degrees.

The AC coupling circuit 112 consists of a capacitor c1 and a resistor R3. The noise cut circuit 113 has a resistor I (4, 5.a
6. R8, R223, capacitor C2, transistor T
2. Consists of operational amplifier OPI. AC coupling circuit 112
is provided as a means for extracting the knocking signal from the knock sensor output signal in a good manner, and by passing the knocking signal through this circuit 112, the DC bias current riding on the knock sensor output signal is removed. There is. If it is attempted to extract only the knock signal from the output of a knock sensor formed by superimposing a DC bias component, and if it is attempted to apply the above-mentioned noise mask, the processing becomes extremely complicated. Although the idea of cutting off direct current is simple, it is an extremely practical technical means for correctly classifying knock signals.

The noise cut circuit 113 mainly includes a transistor T2.
Ig noise is cut by the function of . Transistor T2 is turned on and off by output S2 of monostable circuit 128. The monostable circuit 128 is triggered by the fall of the pace signal of the power transistor shown in FIG. 2(1), and generates a pulse for one mask period. Figure 2 (
2 is the output S2 of this monostable circuit 128, and the time width is 1
1 is the mask section width. The transistor T2 is turned on only in the t7 period when the output side S2 of this monostable circuit 128 becomes P1''.As a result, in this 1. period, the knock sensor output side is short-circuited to ground, and there is no input to the operational amplifier OPI. , produces a masking effect that masks Ig noise. ``As the load impedance of the sensor 100, the resistor R3, 4゜R5, and the capacitors CI and C2 can be considered, but the resistor 5 is a high resistance, For example, by setting I MΩ, the resistors R4, R5, and the capacitor C2 can be ignored as loads. Therefore, the load on the sensor 1oo can be considered to be only the capacitor CI and the resistor 3.

Therefore, in addition to the role of DC removal, the AC coupling circuit 112 can also serve as a resonant circuit means for emphasizing and extracting the knock signal from the knock sensor output. This will be explained below.

As mentioned above, the knock r' is detected from the knock sensor 100.
311 # The impedance as a load when looking at the installed ml is determined by making the resistor R5 a high resistance.

Since the resistor R4, the capacitor 5, and the capacitor c2 can be ignored as loads, only the AC coupling circuit 112 is required.

Therefore, a resonant circuit is also formed in the knock sensor 1o.

and an AC coupling circuit 112. That is,
A circuit consisting of an inductance L, a resistor FL1, a capacitor CI, and a resistor R3 becomes a resonant circuit.

The resonant circuit has a response of 3 to 3 for the knock signal (about 7KH2)
It can function as a bypass filter that dampens engine vibrations of 5KIIZ (approximately 5dB). The gain G at this time is as follows.

(2) Figure 4 shows the characteristics of the resonant circuit thus obtained. Figure 4 (a) shows the capacity of capacitor C1) [is 100P]
This is the time when F is thicker and shows the characteristics of the bypass filter. Figure 4 0 shows the capacitance of capacitor CI as 100OOPF.
This is the characteristic when the size is .

Characterizes a low-pass filter. Figure 4■ shows the characteristic tear when the capacitor Ct is set to the size of '247 opF'.
This shows the characteristics of a bandpass filter. The actually formed courinary melody has the characteristics of a bad pass filter similar to that shown in Fig. 4 [F]. In addition, the 1st channel of capacitor C1 (Y470
When PF and resistance R3 are set to 20Ω.

As a result, the resonance frequency is 7KIIZ, and it exhibits the characteristics of a bandpass filter.

113I'F 114 is a resistor R7, a resistor 9. It consists of RIO, capacitors C3, C4, C22, and operational amplifier 01)2. The non-inverting terminal of the operational amplifier OP2 is connected to the output RV (3V reference voltage) of the operational amplifier OP3 of the power supply unit 120, which functions as a buffer. Therefore, BPF
The output 114 is DC biased to the reference voltage. This BPF 114 has a filter function to increase the level difference between the knock signal and the non-knock signal.

This facilitates identification by level.

An output S1 from the operational amplifier OP2 of the BPF 114 is input to a resistor R16 of an AGC circuit 160, which will be described later.

Next, the operation of the knock sensor failure detection circuit 102A will be explained. The sensor/iodine detection circuit 109 consists of a comparison mco2. The sensor open detection circuit 108 includes a resistor R5.
5. It consists of a comparator COI. Furthermore, C of the comparator CO2
It is assumed that the point potential is set to about 0.2 (V) and the point b potential of the comparator CO1 is set to about 2.8 (V).

Now, if the knock sensor 100 is short-circuited, the potential at point a becomes approximately 0 (v). Therefore, the potential at point a becomes lower than the potential at point C, and the output of comparator CO2 becomes approximately 0 (V). When the knock sensor 100 is in normal operation, the potential at point A is set to be greater than the potential at point C due to the DC bias, so the output of comparator CO2 is greater than the potential at point A (i.e., ocV) (!l). : Iu'X taste). On the other hand, when the knock sensor 100 is open, the voltage at point a becomes approximately 3.2 (V), and the voltage at point b becomes approximately 3.2 (V).
Higher than the point voltage. Therefore, the output of comparator CO1 becomes 0 (V). That is, when the knock sensor is disconnected or disconnected, the voltage at the 1g point becomes 0 (V). These results become a signal S3 from the transistor T80, and the output to the integrator 125 becomes 0 (v). Thus, comparator C
Noyote and open faults can be detected by O2°COl.

Output S1 of BPF 114 is input to AGC circuit 160. The output of the AGC circuit 160 is divided into two systems via the mask circuit 150. The first system is a system that includes an amplifier 151 that amplifies the knock signal and inputs it to one input terminal of the comparison circuit 118. The second system is a BGL circuit 119 consisting of a half-wave rectifier circuit 116, a maximum clamp circuit ti6h, a 4B circuit 117, and an amplifier 117A.

The output of the podium chanter 117A is input to the other input terminal of the comparator circuit 118. The output of the 41G unit 117A is negatively fed back to the AGC circuit 160 via the amplifier 161.

The output is ±5 (mV) ~ 600 (r!
IV). In other words, the sensor output will fluctuate over a range of 120 times. If this output is simply amplified (for example, 100 times), the output will be 0.5 (V) to 60 (V).
becomes. However, in a car, the maximum voltage is the battery voltage (approximately 12 (V)), and a value of 60 (V) is impossible.

Therefore, in the past, either a small gain was used to avoid saturation, or a process was performed with the expectation that saturation would occur. The former suffers from the drawback of poor sensitivity to minute inputs, and the latter suffers from poor sensitivity to large amplitude inputs. In the 1A configuration of this embodiment, the AGC circuit 16
0 is provided, and furthermore, this AGC circuit 160 is BPF
It is characterized by being provided on the output side of 114. With this 14-channel configuration, the level difference between the knock signal and the non-knock signal increases in the nPF 114, and it is necessary to manually input the AGC circuit 160 to 1 of this increased level difference.
It is possible to obtain an output with a good IN ratio.

As is clear from the input/output characteristics shown in FIG.
It is possible to control the output vo to be substantially constant except for the weak content of (Vn).

A specific example of the circuit from the AGC circuit 160 to the comparison circuit 118 is shown in FIG. The AGC circuit 160 has a resistance ratio of 16. R
It consists of 17, R, 19, R18, 92°R102, and operational amplifier OP3 F'ET T17. The ten-wave current circuit 116 includes a capacitor C6, a resistor R20, and a capacitor 21.
, R22, diode DI, D2. Consists of operational amplifier OP4.

Clamp circuit 116AFi, resistance ratio 25. R26°R2
7, R29, operational amplifier OP6, comparator CO5, capacitor C8, and diode DIO.

71y divider 117Fi resistor R29: 7te 7fc 9
It consists of The amplifier 117A has a resistance ratio of 30.几31゜1尤32. 33. R34, R35, obeamphu.

It consists of OF2. The amplifier 161 is an operational amplifier OP8
, resistance R93, R37, R, 38, R, 36°R40,
It consists of R39 and capacitor c1o.

The operation of the circuit shown in FIG. 7 will be explained. BPF114
The filtered output s1 is input to the AGC circuit 160(7) OF2 via the resistor R16. An FET (T17) whose gain is controlled via an amplifier 161 is provided at the minus end of OF2. As a result, the gain of the AGC circuit 160 is the OF of the amplifier 117A.
It is changed according to the output of 2. The output of the AGC circuit 160 is masked at a predetermined timing by the mask circuit i50, and is input to the half-wave circulation circuit 116 via C6,1(,20.The output of the AGC circuit 160 is masked at a predetermined timing by the mask circuit i50, and is input to the half-wave circulation circuit 116 via C6,1(,20). Wave rectification is done, maximum clamp circuit 116A
Enter. In this clamp circuit 116A, the comparator C
Due to the action of O5, a predetermined maximum value (corresponds to the voltage applied to the positive terminal via R27, for example, 5V)
be clamped to. The output of this clamp circuit 116A has a resistance ratio of 29. The signal is integrated and smoothed by an integrating circuit 117 formed with a capacitor C9, and further amplified by an amplifier 117A and output. Next, it will be explained in detail.

The input voltage of the AGC circuit 160 (that is, the output voltage of the IIPF 114) is VI, the output voltage is vo, and the FET (T1
Let the output impedance of 7) be ZF.

The amplification factor of the operational amplifier OP3 is a value determined by ZF and the resistance ratio 19 since the first resistor R19 is set to a high resistance. That is, the gain G is.

becomes. The important point in equation (3') is that the impedance Zy
changes depending on the gate-source voltage Vas of the FET (T15). For example, when Vas decreases from 0 (V) to -2 (V), ZF increases and the gain G decreases due to the output feedback of the amplifier 161. If it gets bigger, the opposite is true. As a result, the BGL output voltage output from the amplifier 117A of the AGC circuit 160 is
It is constant regardless of fluctuations in the input voltage to the circuit 160, and its S/N ratio is close to constant.

AGC works well in the range of VGS2±200 (mV).

Similarly, the resistance ratio 19 is for protection in the event of a disconnection failure of the FET (T15), and is set to a high resistance. Furthermore, ZF is set to a value of about 2.OMEGA. from 200.OMEGA.

The mask circuit 150 includes a transistor T151 and a resistor R, 97.
It consists of Transistor Tl1j is rendered conductive by signal S6 applied to its base. Due to the conduction of the transistor T15, the output of the AGC circuit 160 drops to the ground potential.
Masked.

The resistor 95 of the amplifier 151 performs 360 (mV) DC correction on the output of the operational amplifier OP9.

Furthermore, the resistor R35 and the capacitor CIO have a ripple prevention capability for BGL. Furthermore, through the overall diagram:
The comparator (CO) is Hitachi's HA17901.

Resistors R, 40, R39, R, 38 of the amplifier 161 are BG
When there is no L input, the ■ terminal of operational amplifier OP8 is connected to the output terminal (3V) of operational amplifier OP3, so the input to that terminal becomes 3 (Vl), and at this time, the output v2 of operational amplifier OP8 is shown in Figure 8. It has a DC correction function that sets the voltage to 4 (V).Furthermore, the operational amplifier OP8
is an inverting amplifier, and as the input V1 increases, the output V2 decreases. This output v2 is the gate input of the FET (T15) of the AGC circuit 160, so when BGL increases, v8 increases, V2 goes down, and the gain G also goes down, causing the AGC to move by 1. Here, AGC circuit 16
0 starts working from V223(V), 4(v)~3(
V), IZF remains constant.

The operational amplifier OP6' to the maximum clamp circuit 116 is a buffer provided at the end of impedance conversion, and when the non-inverting terminal voltage v1 attempts to exceed 5 (g), the comparator CO'5 becomes conductive, and the diode DIO1 resistor A current is discharged across R28, and as a result, the terminal voltage of the non-inverting terminal of the operational amplifier OP6 is clamped to a maximum of 5 CV. As a result, the inverted output voltage (background voltage level) of the comparator CO5 does not change depending on the intensity of knocking, and there is no possibility that BGL increases during knocking, making it difficult to detect knocking.

The comparison circuit 118 includes comparators 118A, 118B, resistors 43.几59. It consists of a diode D3.

The comparator CO4 of the comparison element 118A receives a signal including the knock signal which is the output of the amplifier 151, and the resistor R, 41 and the filter 42.
A comparison is made with the divided reference voltage.

The reference pressure is set to 4 (V), for example. on the other hand.

Comparator CO3 performs one comparison between the I3GL output and the output of amplifier 151. As a result, the comparison circuit 118 outputs an output S6 of the transmitter 151 that is higher than the reference voltage and higher than the BGL output.

Note that the diode D3 has the role of transmitting this output.

The resistor R59 serves as a discharging resistor between the resistor R59 and the capacitor C13 of the retard amount setting circuit 126, which will be described later.

The output S6 of the comparison circuit 118 is sent to the primary stage mask circuit 153.
Enter.

A specific circuit configuration from the mask circuit 153 to the integration circuit 125 is shown in FIG.

The lead angle signal setting circuit 127 is a fixed lead angle setting circuit 127A.
and a variable advance angle setting circuit 127B.

The fixed advance angle setting circuit 127A has resistors R56 and R68゜7
0. R981 99. It consists of R100 and transistor Tll. The variable advance angle setting circuit 127B includes resistors R65, R,
69, consisting of a transistor T8.

The advance angle output signal of the fixed advance angle setting circuit is determined by the power supply voltage for advance angle at the time of starting. Variable lead angle setting circuit 127B
The advance angle output signal is determined by the switch S9 of the monostable circuit 128. The output S9 is a signal with a pulse width that is proportional to the number of revolutions, and therefore the advance angle output signal of the circuit 127B becomes an advance angle signal that is proportional to the number of revolutions. The lead angle output signal of the lead angle signal setting circuit 127 is input to the adder 140.

The output of the comparator 118 may be directly applied to the integration circuit 125, but it is applied to the integration circuit 12 via the retard signal output circuit 126.
5 may be applied. Furthermore, by providing a mask circuit 153 between the retard signal setting circuit 126 and the comparator 118, a more accurate retard signal can be formed.

The mask circuit fNr153 is connected to resistors R157, a58 . It consists of a transistor T7. The retard signal output circuit 126 smoothes the output pulse of the comparator 118 that has escaped the mask, divides it into 1,000λ, and performs the corresponding retard control.

The retard signal output circuit 126 includes a capacitor C13 and a resistor R.
63, 几64. R66, 几67, transistor T9. T
Consists of IO. The transistor T7 is turned on by the signal S2, and the output S6 of the comparator 118 at this time flows to ground via the transistor T7 and is masked. When transistor T7 is off, signal S6 is applied to capacitor C13.
are vertically stacked and drive the transistor TlO via the resistor R63. Transistor TIO is also driven by signal S3 via resistor 64. Signal S2 is the output of monostable circuit 128. Signal S3 is the abnormality detection circuit 102
This is the output of A. The source voltage applied to the transistor T9 is supplied from a power supply circuit (described later). When starting the engine, the battery voltage drops below a predetermined minimum allowable voltage. The same applies when the battery capacity becomes low.

This abnormal voltage drop results in a high voltage.

The voltage is low during normal voltage. When the voltage is high, transistor T9 is turned on, and the transistor is connected to resistor R6.
3. Continues to turn off regardless of the signal sent through the box 64. On the other hand, when the voltage is low, transistor T9 is turned off.

As a result, the transistor T10 has a resistance R63゜64
The device is turned on and off depending on the voltage value across it.

The integration circuit 125 consists of an integrator 140, a maximum voltage clamp circuit 141, and a minimum voltage clamp circuit 142. The integrator 140 includes an operational amplifier 0P15 and a capacitor C14.
, C15, and 200 resistors. The maximum voltage clamp circuit 141 includes an operational amplifier 0P16, resistors R, 71, and R74.
, R75, 76°It, 77, 78, diode D7, and transistor T20. Minimum voltage clamp l@M142U, operational amplifier 0P17. Resistance 60.
R, 61, and a diode D6.

Next, the circuit operation of the charge division circuit 125 will be explained.

Now, due to the knock signal which is the output of the retard signal setting circuit 126 which inputs the output of the comparator 118, the transistor T
IO is turned on in synchronization with the knock signal. Therefore, as shown in Figure 2 (7J), the knock signal pulse width t (approximately 4
0 to 70μsea), transistor T10 conducts, and current 11 flows from operational amplifier 0P15 to capacitor C1.
4, flows to ground via C15, resistor R67, and transistor TIO. Also, the operational amplifier 0P15 at this time
The output voltage is 3 (V).

Therefore, the voltage increase rate (voltage rise/1 pulse) ΔV1 per pulse of the operational amplifier 0P15 at this time is as follows.

However, the capacitance C is the 111 column capacitance value of the capacitors C14 and C15. As is clear from equation (5), the output voltage of the operational amplifier 0P15 increases in proportion to the number of knocking pulses.

On the other hand, in every cycle, the inverted output S9 of the monostable circuit 128
is applied from transistor T6 to the base of transistor T8, turning off transistor T8 for a fixed mask time t. Therefore, during this time, the current r + Ka = = v,
Color = anti-f (98, R100, capacitor CI4.c
It flows to operation up 0P15 via t5.

The Zener voltage of Zener diode ZD4 is 6M. Also, the ○ terminal of the operational amplifier 0P45 is 13 volts. Therefore, every time one pulse is input from the monostable circuit 128 to the operational amplifier 0P15, the operational amplifier 0P1
The output voltage of No. 5 becomes Tm according to the voltage fall rate (falling voltage value/cycle) Δv2 below.

Therefore, this voltage station reduction rate ΔV2 is set to about 1150 of the voltage increase rate ΔV, taking into consideration the power performance of the engine torque and horsepower. The output of the integrator has its maximum value clamped by the clamp voltage of the maximum clamp circuit 141 and its minimum value clamped by the clamp voltage of the minimum clamp circuit 142.

The integration circuit 125 has a nickname because it has a specific advance angle characteristic (advanced angle value) when the engine starts, by turning on the transistor Tll due to the voltage. The integral circuit 125 issues a command regarding this advance angle characteristic, and the retard circuit 132 performs actual advance (retard) control. This retard circuit 13
2 is, for example, the following document (IJ, S, patent ap
plication, Ser.

JPa 80202 + by Noboru 3ug
iura, filedOCtOber 1, 1979
and assigned to theassign
ee of this application'''I
gnition timing control system
tem for internal combustion
n engine') is used.

Here, the operation of the retard circuit 132 will be explained.

In general, ignition timing characteristics are relative and dis)
It is determined according to a certain operating mode determined by the IJ computer and the ignition system being used. I am still giving the maximum retardation characteristic when knocking, so that this characteristic will be used when knocking. Figure 9 shows the lead angle and retard angle characteristics,
It shows the minimum retard angle (that is, the minimum clamp voltage, knob voltage) stability in the actual operating mode. For example, at low speed.

Below 200r[)m, control is performed to achieve the maximum advance characteristic determined by the ignition timing characteristic. The reason for adopting such characteristics is to ensure starting at startup. That is,
When starting, if the ignition timing is delayed, the engine will generate reverse rotation torque and the load on the starter will be extremely heavy. As a result,
The starter's drive current becomes abnormally high, making it impossible for the starter to turn the engine, resulting in a so-called starting failure.

In order to eliminate such startup failures, for example, 200
Below rpm, the maximum advance angle characteristic is determined by the ignition timing characteristic.

FIG. 10 shows the characteristics of the retard circuit 132 that should achieve the above characteristics. As shown in the figure, the output of the integrating circuit 125, that is, the output voltage of the % integrator 140 has a retard characteristic so as to have a one-stroke angle slope characteristic. For this reason,
The angle advances by a constant angle every cycle. That is, the ignition timing is retarded due to the number of knocking pulses and advanced by a constant angle for a short period of time.

Next, the integration circuit 1 that controls the retard circuit 132
The operation of No. 25 will be described, especially the start-up countermeasure that involves setting the start waiting angle. This start-up countermeasure is related to the power supply circuit, so
First, the power supply circuit shown in FIG. 11 will be explained.

This power supply circuit consists of a start detection power supply device 50 and an actual power supply device 51. Power supply device 50, resistor 86.几8
7. R88, R89, Zener diode ZD3, co/
It consists of a capacitor C19 and a diode D9. Power supply device 51
are resistors 90 and 91°, C20 and C21, and Zener diode ZD4. Consists of ZD5. The battery power source is connected to the BW terminal, and voltages higher than a predetermined voltage (6.2V) are cut off by the Zener diode ZD5, and 6.2V of B2 is output. Voltages A and D are voltages that reflect starting detection. That is, the battery voltage decreases during startup. When the amount of decrease exceeds the reference value, the transistor T
14 is off, voltage A,! :T;) is the same sister.

The same operation occurs when the power capacity of the battery decreases.

If the battery power supply voltage is normal, transistor uT1
4 is on, the D voltage is approximately at ground potential, and the A voltage is equivalent to the drop voltage due to the resistor 86. Resistor R8
6 is set to a relatively high resistance (22 Ω). This D
Base of voltage transistor T9, transistor Tl
It is applied to the base of 1 to set a predetermined advance angle characteristic at the time of starting.

Further, T is a terminal connected to an engine coolant temperature sensor switch. In order to turn off the transistor T14 when the temperature detected by the engine coolant temperature sensor is below a certain value (for example, 70r), a resistor R221,
A diode D12 and a Zener diode ZD4 are provided.

Integrating circuit 12 that controls next retard circuit 132
Let's talk about the operation in step 5, especially the start-up countermeasure that involves turning the engine on at start-up. The Zener diode ZD3 has a Zener voltage of about 6 cV), and when the power supply voltage (V+) is low, that is,
When starting the engine with the starter on, there is a redundant R88,R
The midpoint voltage of 89 will no longer turn on the soener diode ZD3. In this case, the transistor T14 is turned off and the transistor T9°TIO is turned on. At this time, transistor TIO is turned off. Further, when the transistor Tll is turned on, a current flows from the power supply through the resistor R, 70 in the same direction as the current L2, and the output of the operational amplifier 0P15 decreases to the same voltage as the point voltage and is clamped. The voltage at this point is the minimum clamp voltage 1.5 (V
). This clamp output sets the maximum retardation characteristic at the time of starting as shown by the dotted line in FIG. As a result, the retard circuit 132 is controlled and set to the maximum retard characteristic.

Next, the operation of the integrating circuit 125 when the sensor 100 fails will be explained. The outputs from the sensor short detector 108 and the sensor open detector 109 shown in FIG. 3 turn off the transistor T7 shown in FIG. 8, and turn on the l/ran transistors TIO and T20.

When the transistor TIO is turned on, a current of 1 continues to flow through the capacitors C14 and C15 in the same way as in the retard operation described above, so the output voltage of the operational amplifier 0P15 is clamped to the voltage of 6V (maximum voltage), which is the same as the h-point voltage. be done. Furthermore, as a result of transistor T20 being turned on, the voltage at point h is
It is controlled to a fail-safe voltage of 5V, which is lower than the normal voltage. This provides appropriate retardation characteristics even in abnormal situations. Note that the fail-safe clamp voltage when the transistor T20 is omitted is 6V as shown in FIG.

FIG. 12 shows an operating waveform diagram when an abnormal voltage is superimposed on point a in FIG. 3. This abnormality detection is performed by the detection circuits 108 and 109.
This is done by operating in combination. Figure 12 (
1) is the base signal of the power transistor, and if the knock sensor 100 becomes an abnormal signal for some reason, the first
As shown in Figure 2 (2), a voltage higher than the E voltage is continuously generated at point b', and therefore the output of the comparator is as shown in Figure 12 (3).
continues to fall to 0 (V), so the integrator circuit 12
After operation as shown in FIG. 12 (4), the output of No. 5 is clamped to a fail-safe clamp voltage (5.4 V in the figure). Therefore, it is possible to adequately deal with abnormal voltage.

FIG. 13 shows a timing signal generation circuit that generates various timing signals. The monostable circuit 128 includes a resistor 44
．． R45, 几46. R47, R, 48°R50, 几51
．． 52. R53, R54% capacitor C1l, CI2
, transistor 'r3. T4゜T5. T6, diode D4. Consists of D5.

The operation of monostable circuit 128 will be explained. A base signal of a power transistor shown in FIG. 2 α) is applied to the terminal P. This base signal #I turns on the transistor T3 and turns off the transistor T4. Transistor T4
is turned off, the capacitor C12 has [E source B→resistance R
48→R50→C12→A path to the base of transistor T5 is formed. On the other hand, when the base signal is L, transistor T3Fi is turned off, transistor T4 is turned on, and tg
B → Resistor R, 51 → Capacitor C12 → Resistor R50 → D
5→transistor T4→R, 49→ground paths are formed. These two paths are a charging/discharging circuit for the capacitor CI2, and a pulse S2 synchronized with the spark timing having a time width t1 as shown in FIG. 2(2) is emitted at the collector end of the transistor T5. Well, since the transistor T6 has a negative phase relationship with the transistor T5, the signal S2
A pulse S9 having the opposite phase is output. This signal S2 is applied to the base of the transistor T2 in the ignition noise cut circuit [113] to become an ignition noise cut signal, and is applied to the base of the transistor T7 in the mask circuit M153 to play the role of cutting ignition noise. Signal 9 is applied to transistor T8 of advance angle signal output circuit 127, and is used for advance angle control. Further, the signal S2 is transmitted to the rotation detection circuit 156.
0 input signal, which serves as an input signal to the fixed time generation circuit 157.

Retard control of the conventional knock control device configured as described above will be described next.

First, ignition is performed by an ON/OFF signal of a power transistor 134 which provides an ignition signal, as shown in FIG. 14(1). Therefore, an ignition mask signal as shown in FIG. 14(2) is required. This ignition causes the engine/horn to operate, and the knock sensor 100 detects knocking due to the vibration of the engine.

The output from which knock sensor 100 is amplified and
4 Comparator CO as the tapping signal shown in Figure (3) a.
3 (g) input terminal, while comparator C
A signal as shown in b in FIG. 14(3) outputted from the BGL detection circuit 119 is input to the (to) input terminal of O3 and compared. The output signal of this comparator CO3 charges a capacitor C13 via a diode D3 and is integrated. This integrated voltage (Fig. 14 (April) turns on/off the transistor TIO depending on whether it is higher or lower than the Vag threshold level of the transistor 'I'' i o (Fig. 14 (5)). As shown in FIG. 6, control is performed such that the retard value increases in proportion to the on-time of the transistor TIO.

In this way, in the conventional knock control side] device.

The operation signal of the transistor TI'0, which determines the amount of retard, is obtained by masking the signal from the knock sensor 100 and full-wave amplifying the signal by the operational amplifier OP4 and the operational amplifier OP9. It is obtained by comparing this full-wave amplified signal with the background level BGL.

However, the knock control device, which was conventionally installed inside the vehicle, is now required to be installed inside the engine room in order to save space in the vehicle interior and to shorten the wiring of the Ensif (7) knock sensor.

When installed inside the engine room, the temperature changes greatly, and the influence of temperature has a large effect on the operation itself.

For this reason, the conventional knock control device has many elements for determining L'3GL and elements for amplifying the knock signal, and has the disadvantage that it is greatly influenced by temperature.

An object of the present invention is to provide a knock control device that can reduce the influence of temperature changes.

The present invention reduces the number of elements affected by temperature changes by using a half-wave rectified and amplified signal of the masked knock 1s to serve the ON-0FF signal of the transistor that determines the amount of retard. The idea is to reduce the effects of temperature changes.

Examples of the present invention will be described below.

FIG. 15 shows an embodiment of the invention. This embodiment differs from the conventional example in the circuit shown in FIG. The difference between this embodiment and the conventional example shown in FIG. 6 is the capacitor C7.

Resistance R, 23, 24. R216 and operational amplifier OP9
Remove and replace with resistance ratio 21, resistance ratio 21A, 几21
B is provided, and the signal is directly inputted to the (G) side input terminal of the comparator CO3 from the connection point between these two resistors B21A and R21B. That is, BPF114
The filtered output S1 is input to OF2 of the AGC circuit 160 via the resistor R,16. An FET (Tl 7 ) whose gain is controlled via an amplifier 161 is provided at the negative end of OF2.

As a result, the gain of the AGC circuit 160 is the same as that of the amplifier 117A.
It is changed according to the output of OF2.

The output of the AGC circuit 160 is masked at a predetermined timing by a mask circuit 150, and is input to the half-wave rectifier 116 via the C6° circuit 20. Da, iode Di,
Due to the action of D2, half-wave rectification of only the positive direction component is performed,
Input to maximum clamp circuit 116A. In this clamp circuit 4116A, a predetermined maximum value (corresponding to the voltage applied to the positive terminal via R27) is determined by the action of the comparator C05.

For example, it is clamped to 5V). The output of this clamp circuit 116A is integrated and smoothed by an integrating circuit 117 formed of a resistor R29 and a capacitor C9, and further amplified by an amplifier 117A and output. Next, it will be explained in detail.

The input voltage of the AGC circuit 160 (i.e., the output voltage of BI'F 114) is Vl, the output voltage is Vo, and F'ET(T
17), the amplification factor of the operational amplifier OP3 is determined by ZF and the resistance ratio 19 since the value of the resistance ratio 19 is set to a high resistance.

That is, the gain G is.

v 0 , R19G--
〒 (1+ -) ・・・・・・・・・(3″
) Vl ZF. The important point in equation (3”) is the impedance Zp
A: Gate-north voltage V of FET (T 15 )
It changes depending on cs. For example, Vcs is 0 (
When the value decreases from v) to -2(v), ZF becomes large and gain GFi becomes small due to the output feedback of the amplifier 161. If it gets bigger, the opposite is true. As a result, A.G.
The BGL output voltage output from the amplifier 117A of the c circuit 160 remains constant regardless of fluctuations in the input voltage to the AGC circuit 160, and its SN ratio becomes nearly constant.

AGC works well in the range of Vas=±200 (mV).

Note that the resistor R19 is for protection in the event of a disconnection failure of the FET (Tli), and is set to have a high resistance. Furthermore, ZP is set to a value of approximately 200Ω to 2Ω.

The mask circuit 150 consists of a transistor T15 and a resistance ratio of 97. The transistor T15 conducts four signals by the signal S4 or S2 applied to its base.

Due to conduction of transistor T15, the output of AGC circuit 160 drops to ground potential and is masked. Furthermore, the resistance ratio of 35 and the capacitor CIO have a ripple prevention function for BGL. Furthermore, through the overall diagram, operational amplifier ° (OP)
is Hh 17902 manufactured by Hitachi S, and the comparator (CO) is HA'17901 manufactured by Hitachi.

Resistance R40, R, 39, 11, 38 H of amplifier 161
(When there is no 3L power, the ■ terminal of operational amplifier OP8 is connected to the output terminal (3V) of operational amplifier OP3,
The terminal input becomes 3 (v), and at this time the operational amplifier OP
The output V2 of 8 has a DC correction function to set it to 4(v) as shown in FIG. Furthermore, operational amplifier O
P8 is an inverting amplifier, and as the input Vl increases, the field V2 decreases. This output v2 is A
Since the gate of FET (T15) of the GC circuit 160 is long, when BGL increases, VI increases and V2
The downlink gain G also goes down, and AGC works. Here, the AGC circuit 160 has a voltage of V 2-3 (V) (4V) when it starts working.
~3 (V) The interval u h Z p is constant.

The operational amplifier OP6 of the maximum clamp circuit 116A is a buffer provided for impedance conversion, and when the non-inverting terminal voltage V1 becomes 5 (v) or more, the comparator CO5 becomes conductive and the voltage is passed through the diode DIO1 resistor 28. The current discharges and as a result the terminal voltage at the non-inverting terminal of the operational amplifier OP6 is clamped to a maximum of 5 (v).

As a result, the inverted output voltage (background voltage level) of the comparator CO5 does not change depending on the intensity of non-king, and there is no possibility that BGL increases during knocking, making it difficult to detect knocking.

However, the positive half-wave signal of the knock signal output from the AGC circuit 160 is input to the f'r-) side input terminal of the comparator CO3 from the connection point between the resistance ratio 21A and R2111.

The comparison circuit 118 includes comparators 118A and 118B.

Resistance ratio 43. It consists of a filter 59 and a diode D3.

Comparator CO4 of comparison element 118A compares the signal including the knock signal which is the output of amplifier 151 and the reference voltage divided by resistors R41 and R42.

The reference voltage is set to 4 (b), for example. On the other hand, the comparator CO3 compares the BGL output and the amplifier 1510 output. As a result, the comparator circuit 118 outputs the output s6 of the amplifier 151, which is thicker than the reference voltage and larger than the sBGL output.

Note that the diode D3 has the role of transmitting this output.

The resistor 859 serves as a discharging resistor between it and a capacitor C13 of a retard amount setting circuit 126, which will be described later.

The output s6 of the comparison circuit 118 is sent to the next stage mask circuit 153.
Enter.

Since it is constructed in this way, an ignition signal is given to the power transistor 34 or ONL by the base signal of the power transistor 34 as shown in FIG. 16(1), and ignition is performed.

Since a large ignition noise is generated during this ignition, this ignition noise is mistakenly determined to be a knock, so it is necessary to mask this ignition noise signal. This kme s Figure 16 (
2) requires an ignition mask signal as shown in FIG. This ignition causes the engine to operate, and knocking is detected by the knock sensor 00 due to the vibrations of the engine. This knock sensor 1
The output from 00 is amplified, and a positive half-wave signal as shown in a in FIG. 16(3) is input to the (g) input terminal of comparator CO3, while the , the first output from the BGL detection circuit 119
A signal as shown in b in FIG. 6 (3) is input and compared. The output signal of this comparator CO3 is the diode D
3 through the capacitor C13, integrated and the first
The immediate signal shown in Fig. 6 (4) is obtained.

The transistor TIO is turned on only when the integrated voltage shown in FIG. 16 (4) is higher than the threshold level of VBI of the transistor T10. !lO signal is input. This on/off operation of the transistor TIO causes the control shown in FIG. 16 (6) in which the retard amount increases in proportion to the on time of the transistor TIO. .

Therefore, according to this embodiment, the number of parts can be reduced and maintenance management can be made easier.

In addition, the temperature coefficient due to the element that changes depending on the temperature, which is conventionally used when comparing the knock signal with respect to the background level (BGL), is the temperature coefficient of the rectifier circuit,
These are the temperature coefficient of the amplifier circuit and the temperature coefficient of the BGL average value circuit, and they all move independently. On the other hand, according to Honsho Shido, when using a half-wave rectified signal and taking the ratio of the signal to the background and land, only the temperature coefficient of the average value circuit becomes the actual amount of change (relative to temperature). ,
The influence of temperature changes can be reduced.

As explained above, according to the present invention, the influence of temperature changes can be reduced.

[Brief explanation of the drawings] Fig. 1 is an overall configuration diagram of a conventional knock control 1lIA device, Fig. 2 (1) to (8) are time charts, Fig. 3, Fig. 6, Fig. 8, and Fig. 11. Figure 13 shows the specific circuit side entrance,
Figures W~(Q, Figure 5, Figure 7, Figure 9, Figure 10 are valleys and
Diagram for explaining crying, Figures 12 (1) to (4), Figure 14 (1)
〜(6J i-1: Main waveform diagram, FIG. 15 is a circuit diagram showing an embodiment of the present invention, FIG. 16 (υ~(6) is a waveform diagram showing an embodiment of the present invention. ° ° ″ 1 pull l soft // [K1-1zJ Fig. 5 Fig. 12 Fig. 13

Claims (1)

[Claims] 1. A knock sensor, a first means for removing ignition noise during ignition from the output detected by the knock sensor, and a predetermined signal obtained by half-wave rectifying the knock signal from which the ignition noise has been removed. a second means for comparing the signal level with the first signal level and outputting a pulse signal; a third means for integrating the output signal from the second means; and an integral signal output from the third means. a fourth means for outputting a rectangular wave by comparing the signal level with a second signal level; a fifth means for outputting an ignition timing retard amount proportional to the output pulse width from the fourth means; and a sixth means for fully retarding the ignition timing based on the output from the means.