US3555256A

US3555256A - Automatic electronic counter apparatus

Info

Publication number: US3555256A
Application number: US747963A
Authority: US
Inventors: Norman L Westlake
Original assignee: General Radio Co
Current assignee: General Radio Co
Priority date: 1968-07-26
Filing date: 1968-07-26
Publication date: 1971-01-12
Anticipated expiration: 1988-01-12

Abstract

An improved automatic electronic digital counter is disclosed in which frequency-range switching is automatically effected and a quotient of the count of a clock pulse-fed counting register and a signal-period counting register is continually indicated.

Description

United States Patent APPARATUS 5 Claims, 3 Drawing Figs.

US. Cl .1 235/156, 235/164 Int. CI G061" 7/38, G06f 7/39 Field ofSearch 235/152,

[56] References Cited UNITED STATES PATENTS 2,932,450 4/1960 Knight et a1. 235/159 3,229,079 1/1966 Zink 235/164 3,249,745 5/1966 Burkhart. 235/160 3,385,960 5/1968 Buu 235/159 OTHER REFERENCES Divide Circuit, SURE BULLETIN, Vol. 2, No. 4, Dec. 1959 Primary Examiner-Malcolm A. Morrison Asristanl Examiner-James F. Gottman Attorney-Rines and Rines ABSTRACT: An improved automatic electronic digital counter is disclosed in which frequency-range switching is automatically effected and a quotient of the count of a clock pulse-fed counting register and a signal-period counting register is continually indicated.

CLOCK s GATE 4 M 23 i' 4 FREGISTER I L O C I (DIVISOR) NUMERICAL INDICATORS PULSES F*- E m F-REGISTE n I SAMPLE R j I I l (QUOTIENT) I c NTROL I I 1 i I 112 I l s gram 1 l S 20 l l i 5 (-1 R co ruw ARITHMETIC M ION u 1 H6 MONITOR UNIT PROGRAM ma l GATE ; SIGNAL L I WM ilGNAL SIGNAL i N PROCE s SOR (DIVIDEND) J. Tai, IBM TECHNICAL DISCLO- A PATENTEUJAN12|97I SHEET 3 OF 3 mw owmzv INVENTQR NORMANLWESTLAKE QUE ATTORNEYS AUTOMATIC ELECTRONIC COUNTER APPARATUS The present invention relates to automatic electronic counter apparatus in which frequency-range switching is automatically effected.

In copending application of Richard W. Frank, Robert G. Fulks and James K. Skilling, Ser. No. 730.430, filed May 20, 1968, for Automatically Adaptive Counter Apparatus, automatic counter apparatus of the character described is disclosed wherein a pair of counting registers is provided, one for accumulating clock pulse counts and the other for accumulating signal periods. Programming means controls the commencement and duration of the period measurement in accordance with an indication from monitoring means as to the degree of filling of the registers. While the systems therein disclosed are quite useful, they have certain limitations that in some applications are not desired. Such limitations may include either relatively slow computation of the reciprocal of the period measurement, or nonoverlapping switching ranges and loss of resolution at the low decade frequency range.

An object of the present invention, accordingly, is to provide novel automatic counter apparatus of the type described in said copending application, but void of the above-mentioned limitations. In summary, these ends are attained by a combined measurement computation system in which counting registers for accumulating the number of signal periods and the number of clock pulses are provided (the ratio being the frequency); and divisor, dividend and quotient registers are employed, with the counts in the divisor registers being subtracted iteratively from those in the dividend register to compute the quotient, the quotient register counting the number of subtractions to indicate the calculated frequency.

Other and further objects are hereinafter pointed out and delineated in the appended claims.

The invention will now be described with reference to the accompanying drawings, FIG. 1 of which is a block diagram of a preferred apparatus embodying the invention;

FIG. 2 is a similar diagram of a suitable monitor circuit for use in the system of FIG. 1; and

FIG. 3 is a logic and block diagram of a suitable arithmetic unit for performing the functions required of such a unit in the embodiment of FIG. 1.

Referring to FIG. 1, a source of clock pulses 1 is shown applying the same at 1' to a gate 2', enabled by an appropriate input at 11 from a programmer 11 comprising a flip-flop 113, start and stop controlling

gates

112, 116, and a sample rate control 111. A signal processor 3 similarly applies pulses corresponding to signal periods or cycles at 3" to asimilar gate 2", enabled by an input at 11" from the programmer 11. The gates 2' and 2" feed respective counting registers (T-register-divisor) and 7 (N-register-dividend) by

conductor lines

4 and 4, respectively. The

registers

5 and 7 have their accumulated counts fed at 5' and 7' to a monitor circuit 9, assuming the form of a switching gate control logic circuit, later described, that, via conductor 9 controls the programmer gate 116. Programmer gate 112 is controlled by the sample rate control 111. Gates 112 and 116 are further connected at 3 to the signal processor 3 which preferably produces one pulse for each cycle of the signal, with such pulses having a substantially uniform time relation to the zero crossings of the signal.

An arithmetic means 20 is shown connected at 20' to re gister 5 and at 20 to register 7. Under the control of a conventional computer programmer 24, the arithmetic means 20 subtracts the contents of register 5 from register 7 (NT), adds 5 to 7 (N+T) or multiplies the contents of? by 10 (ION) and returns the result to register 7 via 20", replacing the original contents, with carries from a lower order being fed from the N-register 7 via conductor 20", as later explained. Whenever a subtraction (N-T) results in a positive remainder, an output is'produced at 22 and applied to a frequency (or quotient) F- register 26 that, in turn, enables frequency display or other utilization in any desired manner, such as the numerical indicators 23.

The problem underlying the invention resides in the determination of the preferred realization of an automatic-ranging digital frequency meter which operates on the principal of measuring the time of a multiple of the period of the input signal and calculating the frequency from the measured data. The objectives include the minimum amount of apparatus and the best realization of the potentials of the system to provide the greatest accuracy in the shortest measurement time. Although the problem may be subdivided into a measurement problem and a computation problem, the objective of minimum apparatus requires that some of the circuits be used for both measurement and computation.

In the measurement part of the program, the two

counting registers

5 and 7 are used; one countingthe number of cycles of the signal (N), and the other counting the number of time units (T), with frequency F determined by the relation F=N/T.

In the computation portion, three registers are required: first, the divisor register 5 in which the divisor is stored; second, the dividend register 7 in which the dividend is placed at the start of computation and from which the divisor is subtracted iteratively to compute the quotient; and thirdly, the quotient register 26 which counts the number of subtractions. The number in the quotient register 26 at the end of the computation is, of course, the calculated frequency. In accordance with a preferred form of the invention, the divisor register is made the same as the time-unit T-register 5 and the dividend register is the same as the N register 7, as shown in FIG. 1.

These two

registers

5 and 7 are designed to function as counting registers during the measurement part of the pro gram. During the computation portion, one or both registers are required to function as shift registers in accordance with the design of the arithmetic means 20. These registers may be either binary or decimal in accordance with the design of the arithmetic means 20 and the monitor circuit 9.

The arithmetic means 20 may be designed to receive all of the digits or bits from both registers simultaneously, which is the parallel form, or it may be designed to receive a single digit or bit at a time from each register beginning with the least significant pair, which is the serial form. In general, the serial system usually results in fewer components, while the parallel system gives higher speed of computation. Since adequate speed can presently be achieved in a serial system built with economical components, the serial system is to be preferred except where application requires the ultimate in computational speed. When the serial form is used in the arithmetic means 20, then the

registers

5 and 7 are required to shift toward the least significant digit or bit.

In making period measurements, the gating logic is arranged so that N is measured exactly and T is measured to within plus or minus one time unit. The resolution, therefore, depends upon the number of time units measured. The measurement program is preferably arranged to constrain N to be any value which makes T lie in a predetermined range; with optimum resolutions attained under conditions: 100 percent of register capacity) T2( 100 percent l/F).

The required resolution is equal to that of the result which is to be calculated. Thus, if a six digit result is desired, then about 10 time units should be accumulated. lf decade counting is used in

registers

5 and 7, then the T-register 5 will have the same number of decades as the F-register 26. When 5 and 7 are binary registers, the number of bits should be such as to give the same equivalent capacity. A six-digit result, for example, requires a 20 bit T-register 5.

Requiring the lower bound of T to be continuously variable, makes the monitor circuit 9 unnecessarily complicated. A preferred compromise for decade registers is to monitor only the least significant digit of the N-register 7 and the two most significant digits of the T-register 5 setting. 100 percent T percent for N 8, and to program wider ranges of T for N=l to 8. This provides overlapping ranges and without unduly complicated logic. Preferred numbers together with frequency ranges for a T-register capacity of 100 ms. are shown in the table below;

Since overlapping ranges are provided, the gating logic need not be fast. Frequency shifts smaller than the range overlap will not cause erroneous readings. Since the T-register 7 is always at 90 percent capacity (except at low frequencies), the full resolution capability of the system is very nearly achieved.

For binary registers, the monitor 9 can be further optimized by taking advantage of the fact that the capacities of the registers are not precisely powers of ten; that is to say, the minimum number of bits gives a capacity larger than that required for resolution. A six-digit result requires a T-register 5 of 20 bits or a capacity of 1,048,575. The nature of the computation, however, requires that the N-register 7 be 10 times as large, which requires 24 bits or 16,777,215. By adding one extra bit to the T-register 5, we can, therefore, increase its useful capacity to 1,677,721, or binary 1 10,01 l,001,lOO,l 10,011,001. A preferred monitor circuit 9 for binary registers will monitor the six most significant bits of the T-register 5 and the two least significant bits of the N-register 7. Preferred numbers together with frequency ranges for a 21 bit T-register 5 are shown in the table below:

T (binary) 2 T, msec. F range, Hz.

In this realization, the resolution will be in excess of that required for all frequencies requiring more than the three periods for measurement.

The principle of the division. program is to subtract the divisor from the dividend iteratively until the remainder is smaller than the divisor, N-FT T. (Systems which perform the operation N+FT Tor Fl+FT T are equivalent'and need not be considered separately). Subtracting is begun at the highest possible order (as is done in hand calculations). When the remainder is less than the divisor, it is multiplied by 10 together with shifting the order of the decade in which the quotient is accumulated.

At the start of computation, the divisor or clock pulse count is stored in the T-register and the dividend or number of periods in the N-register 7. The steps in computation may, for example, be as follows. The F-register decades 26 are reset to zero. The arithmetic means 20 subtracts T from N. if the remainder is positive, this is repeated until a negative remainder results. The number of subtractions which result in a positive remainder is counted by the first decade of the F-register 26 and is the most significant digit of the quotient. When a negative remainder occurs, the computer programmer 24 switches to addition, and the arithmetic means 20, via 24', adds T to N once. The arithmetic means 20then multiplies the remainder to the N-register'7 by (in a decade register this is done by shifting once to the left), and shifts the order of the F- register decade in which the quotient is accumulated. 1f the first digit is zero, the shifting of the F-register 26 will be inhibited until a nonzero digit has been computed. The said subtraction, addition and multiplication steps are then repeated until the desired number of digits has been calculated. lfroundoff is desired, one more digit is calculated, but this time the subtractions are counted by a scale of live and any carry allowed to propagate through the register.

Thus, to summarize, when the monitor 9 indicates that the count in the clock pulse or T-register (divisor) 5 has reached a predetermined number less than the complete full count thereof, the programmer 11 effects termination of the measurement at a next appropriate output from the signal processor 3. The gate control logic circuit of the monitor 9, moreover, having received the most significant digits or bits of the T-register 5 and the least significant digits or bits of the N- register 7 and having determined when the former count has reached the said predetermined number required by the value of the latter count, applies a stop signal at 9 to the programmer 11 which is caused to terminate the outputs l1'-1l" to the gates 2'-2", respectively, upon the feeding of the next signal via 3 from the signal processor. The division of the count of register 5 into that of register 7, as indicated by quotient register 26, provides a measure of the frequencies corresponding to the signal periods.

It now remains to describe the types of circuits that may be employed in the system of FIG. 1.

The gate control logic monitor 9 for switching control of the programmer 11 may assume the form, for example, of FIG. 2. FIG. 2, appropriate to decimal counting in

registers

5 and 7, The circuit illustrated is a configuration for decimal registers (1-2-4-2 decade) and consists basically of an inverter 90, NOR gates 91 through 96, an OR gate 97,

NAND gates

98 and 99 and flip-flop 100. lnputs 7174 are from the N-register 7 representing the least significant digit thereof, such as B, C, D (at 74, 73 and 72) from three flip-flops of a register decade, and 15 (at 71), the inverted output of the third flip-flop.

Inputs 5159 are from the T-register 5 with the most significant digit represented by F, D, (j, E, B, K and A at 5157, respectively, and the second most significant digit 19, D, applied at 58 and 59. Initially the B C D inputs are at 0 level and the inputs 5 are at 1 level, as later discussed.

Input 24] is a reset signal from the computer programmer 24 which sets the output of flip-flop to the 1 level before the start of a new measurement, and is simultaneously applied to

registers

5 and 7, setting their respective outputs at

conductors

52, 55, 57, 59, 72, 73 and 74 to the 1 level, and their outputs at 51, 53, 56, 58 and 71, to the 0 level. 1 level inputs at 51 and 56 cause the output of gate 92 to be at the 0 level; while 1 level inputs at 53, 54, 56 and 58 cause the output of gate 96 to be at 0 level, which, together with a 0 level input at 52, produces a 1 level output from the gate 95. The 1 level output from gate 95, in turn, produces a 0 level from gate 93. 0 level inputs at

conductors

55, 57, and 59, on the other hand, result in a 0 level output from gate 97; which, applied to gate 98, produces a 1 level output. This 1 level output from gate 98 and the 1 level input at 53 cause a 0 level output from gate 94. 0 levels from

gates

92, 93 and 94, however, cause a 1 level output from gate 91, which thereupon causes a 0 output from the inverter 90, serving as the stop signal at 9. Thus, the stop signal at 9 is initially at the 0 level. In order to change to the 1 level, one ofthe inputs to gate 91 must change to the 1 level.

Upon commencement of the measurement, the inputs at 51- 59 and 71-74 will change in accordance with the number of counts accumulated in

registers

5 and 7, respectively. If less than two counts have been accumulated in the N-register 7, as for a one or two period measurement, the stop signal at 9' will change from 0 to 1 level when the two most significant digits of the T-register 5 are four and eight. At this count, inputs at 53 and 58 will have changed from the 1 level to the 0 level, and the input at 59 will have changed from the 0 to the 1 level. The 1 level at input 59 causes a 1 level output from gate 97, which, together with the 1 level output from flip-flop 100, produces a 0 level output from gate 98. Since the inputs of gate 94 are now all at the level, a 1 level output will be produced and will be applied to gate 91, causing the stop signal at 9' to change to the 1 level.

If two to seven counts have been accumulated in the N-register 7, as for a three to eight period measurement, the stop signal at 9' will change from the 0 to the I level when the two most significant digits of the T-register 5 are seven and eight. A two to seven count in the N-register 7, moreover, will cause one or both of the

inputs

73 and 74 to have changed from the 0 to the 1 level, producing a 0 level at the output of gate 94 for any T-register count. At a T-register count of 78, inputs at 53, 54, 56 and 58 will have changed from the 1 level to the 0 level, and inputs at 55, 57 and 59 will have changed from the 0 to the 1 level. 0 inputs at 53, 54. 56 and 58 cause a 1 level output from gate 96 which, in turn, produces a 0 level output from gate 95. The 1 level input at 71 and the 1 level output from flip-flop 100 produces a 0 level from gate 99. Together with the 0 level from gate 95, this causes a 1 level output from gate 93 causing the stop signal 9 to change to the 1 level.

If eight or more counts have been accumulated in the N-register 7, as for a nine or more periods measurement, the stop signal at 9' will change from the 0 to the 1 level when the most significant digit of the T-register 5 is nine. When eight or nine counts have been accumulated in the N-register 7, however, input 71 will have changed from the 1 level to the 0 level, producing a 1 level output from gate 99, which produces a 0 level output from gate 93. Input 72 will have changed from the 0 to the 1 level, which produces a 0 level from the output of gate 94. When the 10th count is recorded by the Nregister 7, the input at 72 will return to the 0 level, which correspondingly sets the output of flip-flop 100 to the 0 level. This produces 1 level outputs from

gates

98 and 99 which cause 0 level outputs from

gates

94 and 93, respectively, for this and all higher counts in the N-register 7. When the most significant digit of the T-register 5 reaches nine, inputs 51 and 56 will have changed from the 1 to the 0 level and inputs 52 and 57 will have changed from the 0 to the 1 level. 0 level inputs at 51 and 56 cause a 1 level output from gate 92 to be applied to gate 91, causing the stop signal at 9' to change to the 1 level.

The arithmetic matrix unit 20 may, as an example, be of the type shown in FIG. 3 to perform the required before-mentioned functions. Conductors 20 apply to the unit 20 pulse train inputs from the divisor or T-register 5 (which may be termed ADDl, T, ADD2, 2, ADD4, l, ADD2 and 2 Apulses). These are applied through respective gates 60 through 63 to the addition gate 64; whence, for each cycle, they are added into the corresponding decade in the N-register 7 along conductor 20".

A carry from a lower order, as before discussed, is fed from the N-register 7 via conductor 20" and inverter 80 to flip-flop FF, to which the 0 level set is applied at 81 and the carry reset, at 82; the carry complement being available at conductor 83. The output from the 1 section of the flip-flop FF is shown applied at 84 to a carry gate 65 to which a carry gate pulse is applied along a conductor 24' from the programmer 24 (FIG. 1). A further gate 66 is energized by an AC level controlling count gate pulse fed via conductor 85 and the output of 80, applied at 86, and also feeds the adder 64. Also applied to adder gate 64 is the output of gate 65, as differentiated at 67. When the carry gate 65is turned on, therefore, the output thereof is differentiated at 67 and produces a pulse to add one more at the adder 64 to the next higher decade. If, during the time the carry gate 65-is open because, for example, ofa carry from further down the line, a lower order carry at 20" (from the N-register 7) will complement the flip-flop FF and send a pulse through the adder gate 64 in the same manner abovedescribed. The basic system of addition herein involved is further described, for example, in Arithmetic Operations In Digital Computers, by R. K. Richards, Van Nostrand, 1955, commencing with page 234.

second down to 0.6 Hz. All six digits are always used; a measurement at the counters low frequency limit is actually presented with microhertz resolution. A 10 MHz. clock was employed. A register capacity of l67 ms. was provided, which corresponds to a frequency of 5.99 Hz. The lower limits of measurement time were programmed to produce the following results:

Minimum measurement Number of Periods: time, msec. 1 78. 6 2 91. 8 3 98. 3 4 and higher 101. 6

It should be borne in mind that the steps referred to in the above program do not represent range changes in the usual sense; that is, there is no interruption of the measurement, but only a decision as to when to terminate. The range (i.e., decimal-point position and measured units) is determined entirely in the subsequent computation.

In order to measure lower frequencies, the measurement time must be increased, as by the lowering of the clock frequency.

In summary of the previously described operation, the re gisters 5 and 7 function either as counting registers or as shift registers by application of proper programming signals. They count signal pulses and clock pulses during the measurement; and during computation, they become, respectively, the dividend and divisor registers in a serial computer. The method of computation thus involves subtracting the divisor from the dividend and testing for a positive remainder. If the remainder is positive, a pulse is applied to the first decade of the quotient register 26 and another subtraction is performed. When the remainder is negative, the divisor is added to restore a positive remainder, completing the calculation of the first digit. The remainder is then multiplied by 10 and the process repeated to calculate the second digit. Usually the dividend is initially smaller than the divisor (except where 'the signal frequency exceeds the clock frequency), so that the first digit is usually zero. The quotient register 26 is designed to ignore such nonsignificant zeros, however, and will wait until a nonzero digit is recorded in the first decade before transferring its input to the second decade. The steps required to perform this normalization are counted to determine decimal point position and units. Computation is continued until seven digits have been computed, the last of which is not displayed but is used to generate a roundoff in the quotient register 26. The time required for computation depends upon the frequency, varying from about us. for 10.0000 MHz. to about 675 ,u.s. for 9.99999 Hz. in the above-mentioned instrument.

Because of this short computation time, the result is displayed at 23 directly from the quotient register 26 without the need for buffer storage.

Further modifications will, also, of course, suggest themselves to those skilled in this art; and all such are considered to fall within the spirit and scope of the invention as defined in the appended claims.

I claim:

1. Counter apparatus having, in combination, a signal processor for producing impulses the signal periods of which are to be measured, a source of clock pulses, gating means connected with said source and said signal processor, respectively, and each having output circuit means, a pair of counting registers connected with said output circuit means, respectively, one to accumulate counts of said clock pulses and a second to count signal periods during a measurement time. programming means connected with the gating means for controlling the commencement and duration of the measurement time, monitoring means connected with both counting registers for determining when the count in the said one register has reached any one of several predetermined numbers less than the complete full count thereof and dependent upon the concurrent count in said second register means for connecting the monitoring means with the programming means to cause the programming means to effect the termination of the said measurement time in response to the said determining, and means connected to said registers for dividing the count of the said one register into that of the said second register to provide a measure of the frequency corresponding to the said signal periods.

2. Apparatus as claim in claim 1 and in which the said counting registers are decimal and the said monitoring means comprises gate control logic circuit means having means for receiving the most significant digit count of the said one register and the least significant digit count of the said second register and in which the programming means has means to produce a stop signal for application to the gating means upon the advent of the next pulse from the said signal processor after the said determining.

3. Counter apparatus having, in combination, a signal processor which produces one pulse for each cycle of the signal, the periods of which are to be measured, said pulses having a uniform time relation to the zero crossing of said signal, a source of clock pulses, gating means connected with said source and said signal processor, respectively, and each having output circuit means, a pair of counting registers connected with said output circuit means, respectively, one to accumulate counts of said clock pulses and a second to count signal periods during a measurement time, programming means connected with the gating means for controlling the commencement and duration of the measurement time, monitoring means connected with both counting registers for determining when the count in said one register has reached any one of several predetermined numbers less than the complete full count thereof, and dependent upon the concurrent count in said second register, means for connecting the monitoring means with the programming means to cause the programming means to effect the termination of the said measurement time in response to the said determining, and means connected to said registers for dividing the count of the said one register into that of the said second register to provide a measure of the frequency corresponding to the said signal periods.

4. Apparatus as claim in claim 3 and in which the said counting registers are decimal and the said monitoring means comprises gate control logic circuit means having means for receiving the two most significant digits of the said one register and the last significant digit of the said second register and in which the said programming means has means to produce a stop signal for application to the gating means upon the advent of the next pulse from the said ,signal processor after the said determining.

5 Apparatus as claimed in claim 3 and in which the said counting registers are binary and the said monitoring means comprises gate control logic circuit means having means for receiving the second through the sixth most significant bits of the said one register and the two least significant bits of the said second register and in which the said programming means has means to produce a stop signal for application to the gating means upon the advent of the next pulse from a said signal processor after the said determining.

Claims

1. Counter apparatus having, in combination, a signal processor for producing impulses the signal periods of which are to be measured, a source of clock pulses, gating means connected with said source and said signal processor, respectively, and each having output circuit means, a pair of counting registers connected with said output circuit means, respectively, one to accumulate counts of said clock pulses and a second to count signal periods during a measurement time, programming means connected with the gating means for controlling the commencement and duration of the measurement time, monitoring means connected with both counting registers for determining when the count in the said one register has reached any one of several predetermined numbers less than the complete full count thereof and dependent upon the concurrent count in said second register, means for connecting the monitoring means with the programming means to cause the programming means to effect the termination of the said measurement time in response to the said determining, and means connected to said registers for dividing the count of the said one register into that of the said second register to provide a measure of the frequency corresponding to the said signal periods.

4. Apparatus as claim in claim 3 and in which the said counting registers are decimal and the said monitoring means comprises gate control logic circuit means having means for receiving the two most significant digits of the said one register and the last significant digit of the said second register and in which the said programming means has means to produce a stop signal for application to the gating means upon the advent of the next pulse from the said signal processor after the said determining.

5. Apparatus as claimed in claim 3 and in which the said counting registers are binary and the said monitoring means comprises gate control logic circuit means having means for receiving the second through the sixth most significant bits of the said one register and the two least significant bits of the said second register and in which the said programming means has means to produce a stop signal for application to the gating means upon the advent of the next pulse from a said signal processor after the said determining.