1970-12.pdf

HEWLETT-PACKARD JOURNAL

DECEMBER 1970

Computing-Counter Measurement Systems

Automated measurements and data processing don't necessarily require

a computer. Systems based on the HP computing counter and a new programmer

have computer capabilities but lower-than-computer costs.

By David Martin

outputs for time synchronization, delay generation, and

integration with the outside world to make a practical,

working, measurement system. Most programmer capa

bilities are plug-in options. Thus the system configuration

can be tailored to fit exactly the requirements of the appli

cation. Capabilities not needed are not paid for, but can

be added easily in the field if requirements change. The

programmer is described in detail in the article on page 7.

IF THE HISTORY OF THE ART AND SCIENCE OF MEASURE

MENT could be viewed as a sequence of ages, the current

one would have to be called the Age of Automation. Sys

tems are everywhere, performing with varying degrees of

automation a variety of functions that used to be drudgery

for human beings.

The most automated systems have digital computers as

system controllers. In these systems, automated measure

ments are only a beginning. The computer can be and

usually is programmed to perform arithmetic operations

to reduce the raw measured data to the most advantage

ous form for its ultimate use.

Computing-counter systems are a new type of low-cost

computerized instrumentation system. These systems have

no computer as such, but instead are built around the

HP 5360A Computing Counter1, an instrument which is

part computer and part digital measuring instrument. Like

computerized systems, computing-counter systems have

the ability to make measurements automatically and per

form arithmetic operations on the measured data, all un

der program control. True, these systems don't have the

'horsepower' of a computerized system. They are, how

ever, simple to operate, they have precision measurement

capability, and because they contain no computer, their

cost in a given application may be much less than a com

puterized system designed for the same application.

Fig. 1 is the basic block diagram of a computing-coun

ter system. The arithmetic unit of the computing counter

provides the mathematical functions add, subtract, multi

ply, and divide. The instrument part of the computing

counter is called the measurement section.

Programmability is provided by a new instrument,

Model 5376A Programmer, which also acts an an inter

face between the computing counter and other instru

ments or peripheral devices. The programmer has

â€¢The Model counter can also be programmed by its keyboard, Model 5375A (see

Ref. 2). For systems use, however, Model 5376A Programmer offers several advan

tages: it is rack mountable, it retains its program when the power goes off, it has

a larger program memory, and it has interface facilities.

Cover: Crystal plating is a

typical process-control ap

plication for the 5360 A Com

puting Counter and its new

Programmer, Model 5376A

(on top of the computing

counter). See page 5 for a

more complete description.

Other applications for the

computing counter and programmer are in data

reduction and statistical analysis.

In this Issue:

Computing-Counter Measurement

Systems, by David Martin page 2

Programmer Is Key to Computing-

Counter Systems, by Eric M. Ing-

man ...

page 7

Measuring Noise and Level On

International Telephone Systems,

by Jim Plumb and Jacques Holt-

nger

page

PRINTED IN U.S.A.

O HEWL

programmer as an interface device, greatly extends the

measurement capability of the system. Fitted with its op

tional digital input/output facilities, the programmer can

control many kinds of HP digital instruments and receive

data from them. The digital I/O might be used, for ex

ample, to add a digital voltmeter to the system. Digital

output is also useful for recording results using a digital

printer or other peripheral device.

Another programmer option, a two-channel analog

output, can be used to record data on X-Y plotters. Ana

log output can also serve as a feedback control signal in

process-control applications.

Programs

Programs are written in the machine language of the

computing-counter system. This language is well docu

mented and easy to learn. Programs are implemented by

diode arrays or punched card, both of which act as read

only memories, retaining their programs when power is

turned off or lost.

The maximum program memory is 200 steps long â€”

short by computer standards. However, the machine

language is quite efficient so the memory capacity is

sufficient for most applications for which this type of sys

tem is suited.

Fig. 2 is a photograph of a typical system.

Computing-Counter Systems

There are three principal categories of applications for

computing-counter systems:

â€¢ data reduction

â€¢ statistical analysis

â€¢ process control.

Here are several examples.

Fig. 1. Computing-counter systems, like computerized

systems, can make measurements and arithmetically

manipulate the data resulting from the measurements, all

under program control. Applications are in data reduc

tion, statistical analysis, process control.

Measurement Capabilities

By itself, the computing counter measures frequencies

between 0.01 Hz and 320 MHz directly, with 11-digit

precision. It can also measure the time between two

events to a resolution of 100 picoseconds, about the

time it takes light to travel one inch. Its precision is of an

order matched only by the most complex of systems,

and is the result of making its arithmetic capability an

inherent part of its measurement functions.

With an appropriate signal conditioning unit to convert

the physical quantity being measured to a form compati

ble with the computing counter's measurement section,

other kinds of measurements can also be made. Tempera

ture, for example, can be measured with microdegree

resolution. Phase, inductance, resistance, and other quan

tities can be measured by indirect techniques, using the

arithmetic capability of the system.

Addition of other instruments to the system, using the

Fig. 2. A new modular systems

programmer, Model 5376A, is

the cornerstone of computing-

counter systems. It is pro

grammed by plug-in diodes or

by punched card. Details are

in the article on page 7.

where T is the averaging time for each frequency sample

It. Fig. 4 illustrates the kind of results this measurement

gives â€” a complete characterization of the frequency

instabilities of the source in the time domain.

Equally simple to implement is a program to measure

FM deviation. Average carrier frequency and peak devi

ation above and below the average can be measured.

The flow diagram of this program is shown in Fig. 5. It

consumes 79 steps of the 200 available.

Data Reduction

A good example of the data-reduction capability of

computing-counter systems is the area of nonlinear sys

tems analysis. Most transducers are to some extent non

linear. By determining the transducer transfer function

y=f(x), measuring the transducer output x, and pro

gramming the computing-counter system to solve for f(x),

a direct readout of the transducer input can be obtained.

For a specific example, take the measurement of boron

concentration. Boron concentration gives a measure of

the rate of reaction of an atomic reactor. A transducer is

available which produces an output frequency dependent

upon the boron concentration. Fig. 3 shows its transfer

function. Programmed to solve the equation of this func

tion, the computing-counter system becomes a 'boron-

meter,' displaying boron concentration in real time. The

transfer function need not even be known. Given sample

coordinates, a computer program is available that can

calculate the transfer function that best fits the coordi

nates over any given range. Logarithmic, exponential,

square-law and polynomial functions up to eighth order

can be handled by this program.3

10Â»iS 100 MS 1ms 10ms 100ms Is

Averaging Time T

Fig. 4. Short-term stability of an oscillator. The comput

ing-counter system can be programmed to make the fre

quency measurements and perform the computations

needed to obtain a direct readout of o&t/,,

The program consists of making a series of mea

surements fi of the frequency-modulated signal and

computing

J_ N

lav == Â»T â€” . It'

1=1

the average carrier frequency. The number of measure

ments N is determined by the setting of one of the

front-panel thumbwheel switch assemblies of the pro

grammer. Typically, N would be set at 1000. The

program also compares each frequency f i with maximum

(fmax) and minimum (fmln) values stored in registers in

the programmer. These registers are updated if f i exceeds

the previously stored value. After N measurements the

quantities stored in these registers represent the maxi

mum and minimum frequency excursions of the signal.

The program automatically displays the average fre

quency and the peak deviation above and below the aver

age. It typically takes, including the measurements, less

than ten seconds.

Another example in the area of statistical analysis is

the measurement of integral nonlinearity. This quantity

gives a meaningful measure of the amount of nonlinearity

in analog devices such as voltage-controlled and FM

Transducer Output Frequency f (Hz)

Fig. 3. Transducer transfer function relating boron con

centration to transducer output frequency. Computing-

counter system can solve this equation and display boron

concentration directly, like a 'boronmeter.'

Statistical Analysis

Standard deviation is a frequently used statistical quan

tity. Programming the computing-counter system to

determine the standard deviation of a number of meas

urements is a simple matter. Probably more meaningful

where frequency measurements are concerned, however,

is the Allan variance.4 This is a specially modified form

of standard deviation which provides a direct measure of

the fractional frequency deviation or short-term stability

of a frequency source. The Allan variance is defined as

SET Â¡,2= O

MEASURE f

SET U, = fâ„¢ = f.

DIGITAL

PROGRAMMER Digital (HP 3480A)

Voltage

Data

HP 5360A COMPUTING

COUNTER

Fig. 7. Computing-counter system to measure integral

nonlinearity.

measurement XÂ¡ and the equation of the reference line,

the nominal frequency Ynomi is computed. Then AYÂ¡ =

YI â€” Ynoml is calculated. The maximum AYmax and

minimum AYmin of all the AYÂ¡ are retained. From these

values, the integral nonlinearity is computed and dis

played directly in percent.

The system can also be programmed, of course, to

determine the reference line. Least-squares fit and aver

age slope are two generally used criteria. A disadvantage

of both these criteria is that all the AYi have to be mea

sured twice, once to determine the equation of the

reference line, and second to measure the integral non-

linearity with respect to this reference. This can be

avoided by specifying the reference as that line passing

through the origin and intersecting the curve at the speci

fied full scale value YL.

Fig. 5. Flow diagram of program lor computing-counter

system to arrive at average frequency and peak devia

tions of a frequency modulated signal.

oscillators, voltage-to-frequency converters, transducers,

and so on. For a voltage-controlled oscillator, the inte

gral nonlinearity is illustrated by Fig. 6.

The equipment setup to make such a measurement is

shown in Fig. 7. The computing-counter system gener

ates a voltage stimulus, which here is a staircase, under

program control. At each voltage step the voltage XÂ¡

and the frequency YI are measured. From the voltage

Process Control

Its ability to output digital and analog data under

program control makes the computing-counter system

suitable for process-control applications. These outputs

would normally be used as the feedback or control signals

in such applications.

A typical application where the analog output is used

is in crystal plating. In manufacturing crystal oscillators

and filters the final adjustment in frequency is made by

depositing gold onto the crystal. The heater which con

trols the gold deposition is in turn controlled by the

analog output voltage. A typical setup is shown in Fig. 8.

The crystal output frequency is continually monitored

by the computing-counter system. The control voltage

generated is proportional to the difference between the

actual frequency fc and the desired final frequency f0. In

practice, to ensure that fc does not overshoot f0, the out

put voltage would be proportional to the logarithm of

(fc â€” f0) as illustrated by Fig. 9.

Fig. 6. Integral nonlinearity of a voltage-controlled

oscillator.

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: