SU1439628A1 - Multiplying device - Google Patents

Abstract

The invention relates to computing and can be used in analog computers. The purpose of the invention is the accuracy of the device operation. The multiplying device contains a multiplier 1, correction blocks 2 and 3, each of which has a switch 4, an adder 5, blocks of scaling factors 6 and 7 and a frame 8, and a control block 9. The control block contains. There are memory elements, keys, a differential amplifier, a pulse generator, a pulse distributor, an input resistor. The achievement of this goal is ensured by the introduction of a control unit, two switches and two quadrants into the device, as well as by new components. 2 Il. e (L Faye. T

Description

11A

The invention relates to computing and can be used in analog computers.

The aim of the invention is to increase the accuracy of the device,

Figure 1 shows the scheme of the multiplying device; 2 is a control block diagram.

The multiplying device (Fig. 1) contains a multiplication unit 1, correction units 2 and 3, each of which includes a switch 4, an adder 5, blocks 6 and 7 of scale factors, a quadrant 8 and a control unit 9.

Block 9 control (Fig.2). contains the first, second and third elements 10-12 of memory, respectively, the first, second, third and fourth keys 13-16, respectively, a differential amplifier 17, a pulse generator 18, a distributor 19 pulses, an input resistor 20. -

The operation of the multiplying device is based on the following.

The output signal of a real multiplying device can be represented with a sufficiently high accuracy by the expression z a + bx + cy + dx-y + ex2 + f for gx y + hxy +

+ px2u2, (1)

where X and y are the values of multiplied

analog signals; a, b, c, d, e,, p are coefficients.

In practice, it is sufficient to simply exclude errors due to 1–3 members and the difference in the scale factor d from 1. Cross-links between inputs allow you to exclude 5 and 6 members, but errors due to 7–9 members are not excluded. This, in most cases, does not allow to obtain the error of the multiservice devices less than 0.2-0.4%.

We set Z (x + lx) (y + yu); lh, +; x + 1,

“To n + p it + iSiK;

 + y oi.2K + Uck where Z is the output signal of the multiplier; X, Y - values of input signals;

Lh, dy - errors reduced to inputs X and y;

(four)

(five)

(2) (3)

(6)

15

20

 . „

25

 ,

thirty

amendment to the Criminal Code, compensating for errors of X and y;

(i)

KI - errors of displacement, scaling and non-linearity; , 2;

jf, - - amendments to compensate for the corresponding errors.

Let us set y, 0,, then from (2) with

1k .., (V)

we have

. .(eight)

Let the input signal X take values and, and t corresponds to (7), then change y, we set. ,(9)

where Z and Zppi values, and,, from (9), taking into account (2) - (6) we have

".. V v (10) I

Taking into account that the expression (2) is symmetric for the variables X and Y, it is possible, like (7) and (10), to write down the conditions under which the influence of the errors at and y is excluded.

21 lvb,

(eleven)

(12)

40

 ГК t г - Гг1 - 4у.

The condition under which Z does not depend on the values of - and. Can be found taking into account (7), (10) and (11), (12) as follows.

Set, Y and, change one of the values of f, or both at once, set

,(13)

whence, taking into account (2) - (6)

X-P4K + X- .. 1) The values,, Du in (7),

(10) - (14) are much less than the values of d,},, therefore, they can be neglected. Associated (1) and (2), we find

g, (i-f 2); (15)

();

P hcg The device for implementing the described algorithm is as follows.

in a way.

In the first cycle, at the output of the switches 4, blocks 2 and 3 of the correction, the signals are stopped and 55 The control unit 9 changes the offset at the fourth input of the adder 5 of the block 2 until the output signal of the multiplication unit 1

45

 - 1439628

m is carried out

 Ik

2. 9.

Value d. blocking block

the tact of the signal switches 4 blocks and in the second half of the switches 4

according to soo u

on the basis of me

Signals act accordingly and,

The control unit 9 changes the offset olj at the fourth input of the sum {atom 5 of block 3 until the output signal Z equals zero, and the offset value (11) is fulfilled in block 9,

In the third cycle, the switches 4 of block 2 and 3 are set by the signal from block 9 to a position where signals act on the inputs of multiplier 1, and

Control block 9 changes the values of the transfer ratios of blocks 6 until the moment when the relation (13) is fulfilled.

Z, Z

m

Where

z is the third reference signal.

This introduces an amendment to the value of the scale factor-1 multiplication.

The value of the amendment to the scale factor of block 1 multiplication is stored in block 9.

In the fourth cycle, the outputs of the switches 4 of blocks 2 and 3 are set to signals x 2 and y, y.,, The value of X periodically takes the values X or-X j,

The control block 9, the d5y of which in the particular case can be performed by the operator, changes the transfer coefficient of block 7 until the average for the period value Z is equal to zero. In this case, relation (10) is satisfied, i.e. the correction Aqx is entered into the adder 5 of block 2.

Similarly, in the fifth cycle, the correction y and Y is introduced into the adder 5 of block 3, while the signals at the outputs of the switches 4 x ,, - x are respectively set and the mask y periodically takes the values + y. or The achieved values of the transfer coefficients of the blocks 7 are fixed until the next correction.

- -

,

ten

20

25

.

In the sixth cycle, the switches 4 connect the signals X and Y to the inputs of the adders 5 and the output of the block 1 takes into account the corrections introduced.

The correction process is periodically repeated, so the change in the parameters of the multiplication unit 1 in time or on temperature is automatically compensated.

 The time spent on correction is much less (100–1000 times) compared to the interval between corrections, g The influence of the coefficients y and by an order of magnitude less than the influence, or f,., Therefore, the correction of errors and y is sufficient once manually, for example, before use.

Only the first, second, third, and sixth bars automatically repeat.

The introduction of automatic correction allows reducing errors from time and aging, makes it possible to allocate and compensate errors from third order nonlinearities.

One of the most practical applications of the device is multichannel measuring systems. At the same time, the time spent on correction may correspond to the time spent on multiplying the signals along one of the channels.

The principle of operation of the control unit (Fig. 2) is as follows. The keys 13-16 and the switches 4 of the blocks 2-3 are controlled by signals from the distributor 19 pulses, clocked by the generator 18. In the first cycle, the key 13 is opened, and the remaining keys of the block 9 are closed. The bias signal ai from the output of the memory element 10 operating in the recording mode is fed to the quarter input of the adder 5 of the block 2, the output signal Z of the block 1 is fed to the first input of the amplifier 17. Since the blocks 5,1,1,1,13 and 10 form If the following sys- tem is negative feedback, which processes the signal at the second input of the amplifier 17, then in the steady state we will receive a signal. After the end of the first cycle, the key 13 is closed and the 55th memory element 10 goes into a storage mode, in which the signal of

thirty

35

40

45

50

element output does not change.

10 memory practically

Similarly, the signal of the current output element 11 is processed, while all the keys are closed for the duration of the test, except for key 14. After the end of the cycle, key 14 is closed and memory element 11 goes into storage mode. The testing of the control signal for blocks 6 is performed as follows. The blocks 6, 6, 17, 17, and 12 form a following system, which repels the deviation of the signal Z from the multiplier output from the reference value Z, which is fed to the second input of the differential amplifier 17 through the key 16. All keys except keys 15 and 16, while closed. In the steady state, the signal at the output of the memory element 12 has a value at which the transfer coefficient of the multiplying device is equal to the nominal value. After the end of the clock cycle, the value of the control signal is stored in the memory element 12.

The requirements for the accuracy of setting correction signals are not great, given that small signals are compensated for errors of the order of 0.2-1.0%. The loop gain of the servo system may be in the order of 10-100, which allows block 9 to be implemented quite simply.

Claims (1)

Formula from the range  A multiplying device containing a multiplication unit, the output of which is the output of the device, and the first and second correction blocks, each of which contains two blocks of scale factors and an adder, the first and second inputs of which are connected to the outputs of the corresponding blocks of scale factors, and the output is connected to one From the inputs of the multiplication unit, characterized in that, in order to increase accuracy, a control unit is inserted into it, made in the form of the first, second and third memory elements, the first, second, third and fourth The key of the differential amplifier, between the first input of which and the zero-potential bus, includes an input resistor, a pulse generator and a pulse distributor, whose input is connected to the output of the pulse generator, and the first, second and third outputs are connected to the control inputs of the corresponding keys, which are switched on by the medad the output of the differential amplifier and the inputs of the corresponding memory elements; the fourth key of the control unit is connected between the first input of the differential amplifier and the reference input A new signal and its control input are connected to the third output of the pulse distributor, and the first and second correction blocks additionally contain a quad and a switch, four information inputs each of which are connected respectively to the zero potential bus, one of the device inputs and two the inputs of the reference signals, the control inputs of the switches are connected respectively to the fourth and fifth outputs of the pulse distributor of the control unit, and the outputs to the third inputs correspond to their summation The fourth inputs of the first and second correction blocks are connected to the outputs of the third memory element of the control unit, the fourth inputs of the adders of the first and second correction blocks are connected to the outputs of the first and second memory elements of the control unit, the second differential input. the amplifier of which is connected to the output of the device multiplying unit; quadrants in the correction blocks are connected between the output of the switch and the information input of the second block of scale factors. five 0 five "3 Fig .: