JPS5884780A - Line printer control system - Google Patents

Abstract

PURPOSE:To enable to print by types at a high speed, by a method wherein a predicting circuit is provided in a feedback loop for automatic control of a synchronous motor, and the rotation of the motor is controlled in accordance with the prediction. CONSTITUTION:The feedback loop having an automatic control function is provided to eliminate irregularities in the rotation of the motor for letting-off a paper or for a printing operation, and the rotational angle of the motor is detected to detect variations in the operation of the motor. Namely, a rotating plate 7 fixed to a rotary shaft 1 set in the XY direction is provided, and the periphery of the plate 7 is so set that the distance from the center of rotation to the periphery is changed in the form of a saw tooth wave as the rotational angle is changed. In this case, said distance is changed with a period equal to an integral multiple as the rotational angle is changed from 0 deg. to 360 deg.. On one side of the plate 7, a light source 5 is provided which is distributed along a straight line extending in a radial direction from the center of the rotary shaft 1, and on the other side, a light receiver 6 is provided for receiving the partial light which is not shielded by the plate 7.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a line printer control system, and more particularly to a line printer control system using a prediction circuit for automatic control of a synchronous motor.

Conventionally, line printers have been used as output devices in communication system or information processing system FCs.

This is made faster by parallelizing all of the low-speed printer functions that are integrated with the keyboard, and requires an electric motor to supply power to the mechanical parts.

These motors are stabilized by operating them as servo motors, but it is possible to maintain optimal operation for a given operating state by using a lug, lead, filter, etc. Electric motors are subject to many changes in load and are required to withstand external fluctuations. Conventionally, electric motors are required to have the ability to adapt to such fluctuations. Therefore, either the applicable conditions are limited or an extra large-scale structure is required.

Conventional line printers are unreliable and have limited speeds because their synchronous motors cannot adapt to fluctuations.

An object of the present invention is to provide a line printer control system that can completely synchronize the hammer with the rotation of the type drum and print desired type at high speed, in order to overcome the conventional problems. It is in.

In order to achieve all of the above objects, the line printer control system of the present invention fully drives the type drum by a synchronous motor synchronized with the operating clock of the logic circuit that drives the type hammer, and
(The feature is that the prediction circuit plays a full role in the feedback loop of automatic motor control, and the rotation of the synchronous motor is controlled by prediction.

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is an original diagram of the printing mechanism of a line printer used once a month in accordance with the present invention.

On the type drum 10, several tens to hundreds of cross characters are mounted on the surface, and a print hammer 102 and the type are provided at each printing position of one line. While the drum 104 is rotating, the moment when the type just faces the hammer] 402, the hammer 102 impacts the entire surface of the printing silk 101 through the ink ribbon 103, thereby printing.

Conventionally, in order to print all the desired type in synchronization with the rotation of the hammer 102 and the type drum 104-, F(m is the total installation number 1 of the synchronous color code generation unit, and the position of the type is sent to the control unit ′t11). It is necessary to generate a signal 2. M of the auxiliary signal generator
Mechanically, the M structure constitutes an encoder that encodes the rotation axis, and one building has a plastic disk consisting of 9 to 10 wind segments, which generates the code.

This wind segment consists of 6-7 segments for generating the printed code and parity for checking.
It consists of segments, shift code segments, etc.

In the present invention, the rotation of the type drum 104 is synchronized with the hammer 102 by being connected to the rotating shaft of an electric motor, which will be described later.

Next, the electric motor connected to the type drum 10 will be described.

FIG. 2 is a configuration diagram of an automatic control section of an electric motor showing all embodiments of the present invention.

In office communication equipment, impact fl is used for scanning various materials, feeding paper, and for printing operations.
Motors are required to transmit driving force to the A structure and the rotating mechanism.

In order to completely eliminate uneven rotation of the motor, a feedback loop with an automatic control function as shown in FIG. 2 is provided. Then, in order to detect all the fluctuations in the motor operation, the rotation angle is detected. That is, as shown in FIG.
A rotary plate 7 fixed to a rotation axis l coinciding with the Y direction is provided, and the entire periphery thereof is structured so that the distance from the center of rotation to the periphery changes in a sawtooth wave shape depending on the rotation angle. Then, the shape is such that the distance changes at an integer multiple period while changing the rotation angle from 0 to 360 degrees.

On one side of the rotary plate 7 is a light source 5 distributed linearly in the radial direction from the center of the rotation axis 1, and on the other side is a light receiver 6 that receives partial light not shielded by the rotary plate 7.
Or each may be provided.

The rotary plate 7 is rotated as an XYtr axis by the rotation of the rotor 2 of the electric motor by the full rotation axis 1, but the photoelectric conversion systems 5 and 6 are fixed in a stationary state, and the photodetector 6
The output is converted into a digital signal by an AD converter 8. By inputting the output of this AD converter 8, that is, the sawtooth wave, to the sample value data system 9, the sample value data system 9 compares all of this input wave with the sawtooth reference wave created by calculation. The rotor 2 of the electric motor obtains rotational force through interaction with the armature coil 3 of the motor, and the rotational force is controlled by an auxiliary coil having an effect superimposed on the effect of the coil 3.

This control force is created by the output of the sample value data system 9. That is, in the sample value data system 9, a control signal for the auxiliary coil is created by converting the processing result into an analog value using a DA converter. Further, the sample value data system 9 operates by receiving all clocks from the voltage controlled oscillator 11, but the control voltage of the voltage controlled oscillator 11 is required when the electric motor is used as a synchronous motor.

For example, in the case of a facsimile receiver, the bit timing clock 13 is extracted from the signal from the transmitting side,
As a result of comparison with the bit clock from the voltage controlled oscillator 11 by the comparator 10, a control error voltage 12 is fed back to the voltage controlled oscillator 11.

Next, the processing in the sample value data system shown in FIG. 2 will be theoretically explained.

In general, fluctuations applied to equipment include load fluctuations, which are caused by multiplying the signal components flowing throughout the equipment or the momentum of objects, and physical shocks, ηi, air bacteria, etc. There are things that are added in an additive form like noise. In the conventional automatic control system, if the tracking characteristic for the lfi person is improved, the latter's total color tube will increase, so unless such a trade-off relationship is completely excluded, the feedback loop for automatic control will not be effective. The design is often poor.

Therefore, first, after clarifying the above-mentioned trade-off relationship, a method for improving the tracking characteristic to multiplicative fluctuations will be explained.

It is assumed that automatic control is performed using a sample value data system with a sampling period T.

Now, the control error is expressed as x (kT). Here, k is a sampling number. Also, σ(kT) total multiplicative variation,
If θ(kT) is the output signal of the fully automatic control feedback loop, the automatic control loop shown in FIG. 1 will be obtained, and the following equation will hold true.

x (kT) =a (kT)−〇(kT)
-, , (1) When performing n-fold overlap processing to increase the overall speed, there is a delay of n samplings.The input signal at this time must be shown as follows.

x ((k-N)T'), (N=n+1) The above signal is passed through a filter to maintain all four dynamic characteristics and a prediction filter to compensate for delays due to overlap processing with those filters. It shall be. Since θ(kT) is obtained as a response due to the transmission of the above signal, this can be expressed in the time domain as follows.

・x((m-n)T)...(2) By taking two transformations on both sides, the following equation is obtained.

#(z)=00(z)・GN(z)-z"-x(z)
-, (3) Here, oo(nt) is a function expressed in the time domain, and G N (n'') is a function when the prediction filter is a.

In order to obtain the closed loop transfer function in such an automatic control circuit, it is sufficient to take the Z transformation of equation (1) and fully substitute it into equation (5) above. That is, x(z) + Z
” Go (Z) GN (Z) X (+1.) = a
(z) Here, the question is how to measure x(kT). This takes different values depending on each platform, and is given by the following equation.

G(z) = Z-N-GM(z) ・Go(z)
...(4) Equation (4) above is a known closed-loop transfer function, and the cause of instability is the term z-H included in G (z) and the term z-H included in G o (z). This is the amount of delay. No. 2-''', 1.2 The number of asymptotes of the root locus drawn in the ordinary way increases to N, which means that the number of root loci running in the unstable region increases, and again G. (z) includes 1 The delay amount is f
Move to the unstable footstep area near JJ1.

Now, in order to provide a method for compensating for the delay due to overlap processing, it is assumed that GN(Z)'t is a transfer function of the entire predicted processing that is an integral multiple of the sampling period T.

In the above equation (2), all prediction models in case 11-1 are considered. Predicted value 'tr' (kT) one sample ahead
If we first find this approximate value, we get the following.

x (c) =2 x (kr)-x ((k-1)
T) Since the prediction error '1(kT) can also be predicted in the same way, the fully corrected value of I(kT) is as follows.

X (kT) - I (1 machine) - (kT) Here, a (kT) is the predicted value of the prediction error - (1 machine) one sampling ahead, and is given by the following equation.

= (kT) =2- (kT)-((k-1)?)
Here, the prediction error #(kT) is given by the following equation.

-(k T) ='; ((k-1)?)-x (kr
) By completely eliminating 5 (k″r) and x (1 piece) from the above four equations, the following equation is derived.

x (kT) −2(2x (kT) −x ((k−
1) T))-(2x((k-1) T)-1[((k
-2) T)) By skipping this operation by the number of predicted samples, a predicted value having all the steps equal to the number of skips can be obtained. That is, the prediction error one sampling ahead is as follows.

7, <kT)−2r2x(kT) −, ((k−1)
T))-(2x((k-1) T)-engine((k-21)
? ))...(4-2) Moreover, G M (z) is obtained from the transfer function G of the predictor when the number of skips is 1, (z) by the following equation.

GN(z)=G,(zN)...
・(5) And since G1(z) is obtained by performing the above-mentioned prediction Ilυ of the skip number 1'ftz transformation, it is as follows.

In the same way, GM(Z) <: becomes as follows.

G1) In the above equation (7), z as seen in the above equation (4)
Since there is no N term, the main effect of delay seems to be eliminated. However, since the order of the denominator of the first term on the right-hand side of equation (7) above is N, the following approximation formula shows that this order is almost -N
It becomes clear that the effect is the same as that of That is, by transforming into , it can be seen that in the range of Z duck 1, it is almost equal to 1. However, this term puts a limit W on compensation based on prediction.

In order to investigate the followability to multiplicative fluctuations, it is necessary to investigate the characteristics of the inverse transformation of x(z) using the following equation.

Here, the above integral symbol represents an arbitrary closed IJ c Th integral path contained within the singular point (12) of the integrand. In this equation, consider the case where α(kT) changes sinusoidally with respect to k. α
If we take all the two transformations of (kT), we get the following.

Here, it is the amplitude of the φ-stop string wave vibration.

Therefore, the following equation is derived from the equation (8-1) (a-y,).

...(9) In the above equation (9), if we omit the transient response part and focus only on the steady response, we get the following from the residue theorem.

Here pH1/ (1+G (R'″T))...
- (11) represents the ratio of the residual component to the input fluctuation amplitude, and is called the phase fluctuation suppression coefficient.

Here, G in the above formula (4). In order to find a desirable form of (z), using the approximate expression shown in equation (8) above, ρ is expressed as follows.

ρ, -1/(1+Go(e)) ...(12) From the above equation (12), it can be said that ω in the range of (e""-1)'≦0.01, that is, in this range Regarding multiplicative fluctuations, even if N=7, this means that as long as ρ is small, all residual fluctuations can be kept sufficiently small.

For Go(z), it is possible to narrow the band of the automatic control loop and eliminate noise within the loop.

A typical example is what is called a lag filter, and its form can be expressed as the following equation without expansion.

This is fft! at one sampling time. A filter that subtracts a signal obtained by multiplying the previous output signal by β□ from all input signals, multiplies the result by α1, and outputs the result is made up of n individual antennas.

-F In equation (13), by setting α and β1-1, it can be expressed by the following equation.

At this time, in the range where ρ, G:te"□T-1, becomes smaller according to F'i (ej□T t ) H.

This makes it possible to create a margin for reducing residual fluctuations. If Kf is reduced by this margin, deterioration of the normal operation margin against external fluctuations (αδ) and electrical noise can be completely prevented.

For multiplicative fluctuations (r), the effect of multiple tandem lag filters can be considered as mentioned above. changes as the load fluctuates. In the case of steady delays, it can be compensated for by a method similar to delay compensation using overlap processing, but in the case of fluctuations, this compensation is adaptive. (adaptive).

In order to create the adjustment operation equation used for this compensation, the operation equation ((4-
2) Use equation).

xi(kT) = 2 (2I(kT)-x((k-1)
T)-(2';'((k-1) T)-((k-2
1)? ))...(15) To create an adaptive behavior formula, first create a linear combination of different predicted values with a skip number of 1.

xo(kt) = Σ a1!1(kT)1・(16)1
= 1 06) It is possible to perform adaptive adjustment using the above equation (16) to adapt to changes in load and optimize it. In this case, a function is required to evaluate the deviation from the proper state. For this purpose, the control error factor (k T)
It is desirable that the two-way average of the above equation (16) created from is 0, so the method usually used in automatic equalization is to perform the difference in this direction Ka, but in this method, al or 0 There is a possibility that it will become.

Therefore, conversely, the difference is applied to al in the direction in which the root mean square of the control error x(kT) becomes 0. In other words, all consideration is whether or not the following equation can be processed.

Therefore, in VC, it is necessary to clarify the relationship between xo and xc/, )lQl, and this is expressed by a partial ligation state equation of a control loop with IO as input and X as output. In other words, the input to the thread to be controlled is χ. As a result of being compensated by this input, the compensated error at is X. The objects to be controlled include the physical structure of the mechanism, movable parts, and drive sources such as electric motors that supply all of the power to these parts. A plurality of variables are required to fully represent the state of this mechanism, and these are represented by the following vector (→T represents the entire matrix).

Here, variable zi is variable 2. This also includes cases where it is a differential coefficient of . Therefore, even a complicated thread expressed by an n-stage differential equation can be represented by such a vector.

Also, how to handle it like this? To do this, input from outside the thread,
In other words, the input variables added to the system from the sample value data system can also be expressed by the following determinant in the same way as the above equation (18).

In addition, the output variable of the system is also expressed by the following determinant.

This type of representation has traditionally been based on the state space method used as a system model or system simulation to control a process control system by computer, and the state equation of the system is as follows. It is expressed as follows.

In the present invention, this method is applied to the lag filter that provides the transfer function of equation (16) above, and to the processing shown by equation (17) above.

J: is determined by the physical configuration of the system.

These temporal fluctuations become multiplicative fluctuations and are a burden on the operation of the sample value data system. , xo is due to noise appearing at positions other than xo.

Next, in order to clarify the predictive adaptive adjustment operation equation, θ k (k?) /a in the above equation (17)
It is necessary to express a i in a measurable quantity. Said (2
1), according to equation (16), it is expressed as follows.

(19) Here, g(kT) is one element of E' in equation (21). If this quantity can be measured, the optimization process becomes clearer.

External fluctuations and electrical M-sound (J1 will be added to the position written as 0 in equation (19) above. For such fluctuations, it is necessary to ensure a full and sufficient operating margin. .ΔO, since the analysis for the general case where fluctuations are loaded is complicated, the load is assumed to be applied at the point where the control signal I is detected from the control residual.The response characteristics of the control loop to such fluctuation signals In order to evaluate can.

When the loop is opened, if the fluctuation variance at the point where the control residual is measured is σI, then the following equation holds true.

2...(23)σφQ=
gl ° σN where σI is the input fluctuation variance. The fluctuation variance σ2 at the measurement point of the control residue when the loop (20) is closed is given by the following equation.

62-σN'' (1+, 2')...
- (24) The operating margin M can be expressed by the following equation, assuming that the variation variance t s2 that causes a malfunction in the mechanism section is t s2 .

°′,. . , 2. ), 2 Therefore, the operating margin fiy degraded by creating an automatic control loop rtc is as follows.

, 2 y=1n tog (-) = 1n tog (t +
g+ri...(26)σN2 Here, by the open loop transfer function G (z), z′f
When expressed as t, it becomes as follows.

...(27) In the above equation (27), G (z) is P1#'f, (
14) G in Eq. I feel sad about (1). This is mainly due to (8
), it is based on the fact that when N=2 or less, equation (8) is equal to 1. Under these conditions, consider the Gaussian noise response of a unit amount due to a lag filter in one section, *f, and consider the change due to such a noise response characteristic being connected in a 9-stage chain.

First, it is assumed that the following equation holds true from equation (14).

In this case, the relationship between α1 and g12 is as follows.

The power of the entire open loop is given by the following equation.

The control constant per stage is 1/n for α and -, but when this is smaller than 1, it is considered that -≦.

According to the above equation (26), when 2,2=0.04,
Deterioration due to control operations can be suppressed to 0.2 dB or less. If the amount of residual fluctuation at this time is determined by the above equation (12), if it is about n-4, ρ<0.02, and sufficient followability can be obtained. The integral of the above equation (27) is
Since it is determined by the range of z-1, the above approximation also holds true for N)2, and sufficient followability can be obtained.

Noise entering the control loop degrades operating margins.

One way to prevent this is to use a full-circle low-pass filter instead of a lag filter.

In this case, the delay caused by the low-pass filter will be significant, so it is necessary to extend the delay compensation using the equation (L6)+(22) to the range of this low-pass filter. If the delay is compensated for by this compensation, in order to obtain the amount of deterioration operation margin, it is obtained by integrating over the passband after performing the following variable rk transformation in the equation (27).

z = ejωT In this case, by assuming that G is constant and taking it larger than 1, equation (27) can be written as follows,
It is important to select all values of G such that the residual fluctuation rate, Equation (11), is small.

g2L=, TΔf...(61)
3 and 4 are logical configuration diagrams of the sample value data system shown in FIG. 2.

In FIGS. 3 and 4, register ((9) 16
22 are registers whose contents are updated every sampling period T. Of these, the register 19 (23) generates a sawtooth reference wave at its output, and the adder 40 compares this output with the data 15 converted by the A and D converters δ in FIG. Ru. Note that data 15 is the fourth
It passes through all the calculators 49 and 52 in the figure and becomes data 14, which is input to the addition i% A, O, and a signal proportional to the phase difference between the two waves is obtained as an output.

FIG. 5 is an explanatory diagram of the sawtooth wave comparison operation in the adder of FIG. 3.

Now, as shown in FIG. 5(a), if there is a phase shift between the reference sawtooth wave 19 and the sawtooth wave 14' manually applied to the A and D converters, the phase shift will be as shown in FIG. 5(b). so that the period t
0, an amplitude difference d0 proportional to the phase difference is output, and during period t, a maximum amplitude difference d2 is output.

In this way, since the polarity of the double-toothed wave changes periodically, if one of the two waves completely changes its polarity, the partial value will change greatly until the polarity is changed, and it will no longer be proportional to the phase difference. .

Therefore, in order to deal with this, adders 42 and 44
' is provided, and the difference between the value of the double toothed wave and the value of the previous sample point is calculated for each sample point. These difference zero phrases are stored in the IWl number table (ROM) via registers 17 and 18.
24, 25 tr access and read out Tr-. By outputting 1, it is determined that the polarity of the double toothed wave has changed. Output from function table 24-
lc, 1 is subtracted from the contents of register 29, and register 29 is changed by 1 inputted from the sweet function table 25.
1 is added to the contents of .

By adding the amplitude of the reference wave times the content of the Kf register 29 to the comparison output of the double-toothed wave produced by the adder 40, any changes in the above-mentioned calibration values can be corrected. Therefore, the output of the adder 43 is a signal d0 proportional to the phase difference between the two toothed waves.

In order to control the 20 rotations of the rotor of the motor shown in Figure 2,
Field current in auxiliary winding 4! . Add (k T).

This becomes the additive component of the above equation (19).

Furthermore, x (ct) is defined as
It is compensated in the sample value data system 9.

If the human power for this compensation process is (kT), then X(kT)
is the component of Y in the above formula (20). This means that the output signal @ of the adder 43 in FIG. 3 is processed by the low-pass communication filter 30? By adding compensation processing to this X (kT) in the circuit 31, the output becomes X.

(kT) Obtain f. As mentioned above, the output signal of the adder 4-3 is created by comparing the waves in the sawtooth 1, and after all the processing to remove the influence of noise by the low-pass overpass filter δ0, ) l (16) and (
Processing of equation 22) Full-stage circuit: Performed in 'S1.

In this case, E(
kT) is determined by the influence of the current flowing in the auxiliary winding on the rotational speed, and may be considered as a constant shell. be.

In the present invention, the rotation of the type drum 10+ of the printer is coupled to the rotating shaft 1 of the electric motor 2 shown in FIG.

Since the electric motor 2 in FIG. 2 is synchronized with the sawtooth wave obtained from the output of the sample value data system register 19 shown in FIG. 5, the electric motor 2 shown in FIG. At which phase of the rotation of the type drum 104 each hammer 102 at each printing position of one line should be added can all be determined from the output value of the register 19 in FIG.

The clock of the logic circuit constituting the hammer drive circuit of FIG. 6 is connected to the clock input terminal 60 of FIG. 6 by the sawtooth wave 1 shown in FIG. Obtained by inputting. That is, by inputting each sample value of the sawtooth wave as a digital value to the terminal 60 and accessing the ROM 55 as its full address, the read barrel 16-bit pattern data is set in the register 56. Ru. Clock φ0. which is output by the combination of characteristic bits of that register. φ, respectively, are sent to the shift register SH and the hammer corresponding register OR, and each AND gate 'r: Ij#J<.

Type drum 1040 revolutions 1. The clock φ in Figure 6 and the sawtooth shape in Figure 7 (C)? It is synchronized with IIr at t7, and printing of one line is completed in each cycle.

ROM Q 7 shown in FIG.
7) This is a special memory for stocking that converts the current price into a code for printing on the drum, and its output is compared with the five OR values of the registers corresponding to each hammer. Each register b9 stores the received type code, and the output of the ROM 57 and the value of the register 59 are compared by the comparison circuit 54 to the H-
[The hammer is compared and all matching timings are taken],
At 04, all 61 print pulses are sent out. In addition, register 5q
The printing codes stored in
It is inputted as a primary input, and further inputted to register 5 by clock φ2.
Moved by 0 inches. The clock φ is a clock having a cycle that is an integral multiple of, or equal to, the clock that generates the sawtooth wave.

As explained below, according to the present invention, the type drum I is driven by a synchronous motor synchronized with the operating clock of the logic circuit that drives the type hammer, and the prediction is performed during the feedback loop of automatic control of the synchronous motor. Since all the rotations of the synchronous motor (28) are controlled based on prediction in all stages of the circuit, the hammer can be completely synchronized with the rotation of the type drum and all desired type can be printed at high precision.

[Brief explanation of the drawing]

FIG. 1 is a principle diagram of a printing mechanism of a line printer used in the present invention, FIG. 2 is a block diagram of an automatic wholesale control unit of a synchronized machine according to an embodiment of the present invention, and FIG. Fig. 4 is a diagram showing the structure of the sample value data system shown in Fig. 2, Fig. 5 is a comparison operation waveform diagram of the sawtooth wave in the adder shown in Fig. 3, and Fig. 6 shows all the embodiments of the present invention. FIG. 7, which is a logic block diagram of the hammer drive circuit, is a waveform diagram of the clock and sawtooth lJj waves in FIG. 6. 1: Rotating shaft, 2: Motor rotor, 3: Armature coil, 4
: Auxiliary coil, 5.6: Photoelectric conversion system, 7: Rotating plate, 8:
AD converter, q: sample value data system, 10: feedback output, 11: VOO155, 57! ROM, 56
:Register, 60! Clock input terminal, 61: Print pulse, 02: Receiving terminal 1, 54: Comparison circuit, 58. F)Q
;register. Figure 1

Claims (1)

[Claims] The type drum is fully driven by a synchronous motor synchronized with the operating clock of the logic circuit that drives the type hammer, and a prediction circuit is provided in the feedback loop for automatic control of the synchronous motor, and the output of the prediction circuit is used to control the synchronous motor. A line printer control system characterized by full rotation control.