Вы находитесь на странице: 1из 22

THE

GENERAL RADIO

Exp e rime nte r


VOLUME44

NUMBERS

7, 8, 9
1970

JULY-SEPTEMBER

www.americanradiohistory.com

THE

GENERAL RADIO

VOLUME

.
W.

Experimenter

44

NUMBERS

7-9

JULY/SEPTEMBER

1970
Complexity

Signal Analysis with D i g i tal Time-Series Ana ly zers .

in the

mathematical treat

ment of circuits and systems has been re

duced in the past ten years. Small computers

make possible our present-day electronic ana

The H on o rabl e Society .

13

mathematical analytical techniques to design

Deviations from Accuracy

13

time and dollars. As part of simulation de

Not a Cinderella Instrument!

14

New Shoes for an Old Workhorse

15

Measurement of Transfer Function and Impedance .

lyzers. programmed to apply all sorts of


and test projects that will save engineering

vices. the analyLersalso can perform real-time

functions in imitation of contro l by humans.

thereby helping to reduce hazards to human


o pe rato rs .

The majority of design and test engineers


are relatively too unskilled in the assembly

Reports from The F ield

17

and utiliLation of available instrumentation

Out of the Past

18

motely. the complex analytical test sets pre

P rogrammable Decade Resistor

20

mechanics of assembly but in the understand

GR

1790

More Versatility and Capacity

Recent Technical Articles by G R Personnel

to create some t hi ng which resembles . even re

sently sold. This Jack of skill is not in the

21

ing of the techniques involved.

21

who b ra vely entered No rbert Wiener's class

There are, undoubted ly. nume r o u s readers

room at M .I .T. to study the secrets of effec


tive communication and of communication
analysis. I f. like th i s Editor. you departed

The

General

Radio

Experimenter

is mailed

without

charge t o

from the seminar with a modicum of new

engineers, scientists, tec hn ician s , educators. and others interested i n

knowledge and a plentitude of frust ration.

the instrumen ts and techniques o f electrical and electronics meas

you w i ll welcome another chance to tackle a

Editor. General Radio

phase of s i gna l analysis. helped by the less

urements. Add ress all

1970

correspondence to

Experime11ter, General R adio Co. . Concord, Mass.

01742.

General R a d i o C o mpany. C o ncord, Mass., USA 01742

d i fficult presentation on the next page. After


that. perhaps, you will be ready for the age of
the Fas

Fourier Transform and associated

techniques and for t he instruments that in

corporate them .

The Cover illustrates several of many successful applications of signal analysis in real tune
The ability to match action with reaction , wh i l e events are taking place, 1s an excellent

t1ons, as contrasted with hindsight de ci sions forced upon us by delayed data . The end

cost -reduction measure . It also permits knowledgeable experimental control dunng ac

result 1s an 1ntcl l1gent, economical approach to research, development, test, and produc
tion problems_

www.americanradiohistory.com

C.

E:-. W l 1 it e
(ditor

Signal Analysis with Digital Ti me- Series Analyzers


Although time-

ries analysis techniques are not .1ew, they have not been used extensively because of the

lack of suitable equipment, and therefore are not familiar to many engineers . The literature describing
these techniques generally has been written for an academic audience and has been often clouded rn
abstract mathematics rather than being presented in user-oriented terms. This is an unnecessary obstacle
to place before the potential user, because there is no more need to be a mathematician to use a t1meseries
analyzer than there 1s to use a spectrum analyzer. In this article, the basic principl es will be presented rn
non-rigorous physical terms as much as possible and will be illustrated with examples from several typical
applications.

Analyzing electrical signals is a fundamental problem for

Consider first some examples in the field of structural

engineers and scientists in all fields, whether they are working

mechanics. where the basic quantities to be analy7cd are

in the research and development laboratory, on the produc

mechanical vibrations in aircraft, automobiles, buildings, etc.

tion line, or in the field. Whatever the physical , biological, or

These vibrations may be caused by many external forces such

chemical system being studied, the basic phenomena can usu

as wind, engine combu tion and rotation, road rougl111es .

ally be converted into electrical signal

by su.itab.le tran -

earth tremors, and impact. The designer would like to deter

and analyzed to provide fundamental information

mine such things as the s urce and transmission paths of the

ducer

about the system producing the signals.


nlil recen tly , only a few basic analysis procedures were
used, because of the limitations of the technology available
for building test equipment. The basic ignal-analysi proce

vibrations and the e xpec ted

variou

tresses and displa cements al

points, by analy7ing the signals from points on the

structure. For instance, during the testing of an automobile.

an ob ec t i onal vibration may be found to exist in the passen

dure, other than direct observation of the signals on an o cil

ger compartment. Time-series analysis techniques 1Nill deter

loscope, has been power-spectrum analysis, and will probably

mine if the vibration is coming from the engine, the road, or

w ind

continue lo be so. However, there are many other basic analy

from the

sis procedures that would provide much more valuable infor

vide the transmission path for the vibration can be identified

mation about the system under test, but they have not been

and suitably

available with the performance required for most practical

vibrations. The analysis system can even be used to simulate

In addition, the structural members thal pro

modi fied to filter out or to isolate the unwanted

problems. Even spectrum-analy is equipment, although serv

the structure and the external forces to test the new design

ing adequately for a wide variety of problems, has severe

before it i

limitations due to the analog-circuit technology available.

from the engine can also be analyzed to dctcr i ne its condi

In the last decade, the availability of general-purpose com


puters allowed the advanced

ignal-analysi

techniqu s that

had been pioneered by Wiener, Lee, et al, to be developed to

committed to production. Vibrations coming

tion. Failures may be predicted. or at lea t interpreted, from

such a "signatu re analysis."

Another example is the determination of the flutter char

practical computational routines and were applied successful

acteri tics of aircraft in flight or in wind-tunnel tests. The

ly lo many data-analysis problems. They do not, however.

resonant frequencies and damping of the various vibration

i r

help the experimenter or analyst who requires re ults rapidly

m des of the a rc aft arc mcasu red as a function of airspeed,

so that he can interact with his system. The cost of computer

lo determ ne if there are any conditions that will produce

analysis for this type of problem is al o very high, especially

excessive or unstable vibrali ns .

when the los of engineering time while awaiting resulls is


taken into account.
The revolution in digital-proces ing technology that is
now taking place has brought advanced signal-analy is tech
niques into the laboratory. This advance enables economic
production of instrumentation systems which contain all of
the computational power

required for

processing. The TD I 9 23 Time-Serie

fast, accurate, si rnal

Analyzers, soon to be

introduced by Time Data and GR, are the most advanced and
complete line of this type of signal-processing equipment
availab le . Besides their computational ability, they contain
signal conditioning and display capability, with flexible, sim

ple controls that allow them to be u ed a

easily as conven

tional laboratory instruments.

analysis

by

of high or unexpected loadings.


The field of bio

m e dicine has

oin t areas

already provided many im

portant applications but il has only now begun lo make u e of

i
u ces

sophisticated analysis techn q ucs. The ner ous system or the

human body naturally p od

lcctric s gna s who. e charuc

teristics are indicative of the condition of various parts of the

body. The most common e amp e is the elcclrocardio raph


(EKG) signal , which provides information about the condi
tion of the heart. l::lcctroenccphalograms (EEG's), or brain
wave tracings, are analyzed to

tudy brain damage and the

effects of various, timuli and drugs on the brain. The tech


ignals are very primitive compared lo those

Every process in nature gives rise to "ignals'' that arc


to

Jn the third exampk, rumbling vibrations, caused by auto

mobiles driving across a bridge. arc analyLcd to p in p

niques now being used by clinical physicians to analyze th..:sc

WHERE AND WHY THEY'RE USED


amenable

timc-scrie

analysis techniques.

wh i ch

can be

use d. The widespread u e of time-ser es analy is techniques in


medicine should increase as they become better understood.

. They

Another ap plicat on is gcophy ical exploration, an e cho

can be grouped into categories based on the type of process

ranging application that is in the same category as radar and

Therefore, the list of potential applications is cndle

that is desired. Some

ing that is used and on the informatio

exam ples from sev eral categories are illustrated on the front
cover and described below.

sonar. A v i b rat on signal is transmitted and echos

nrc

n:ccivcd

from the rellccling geo og ical strata. Compari on of he

re

ceived and transmitted signals determines the time delay

JULY /SEPTEMBER 1970

www.americanradiohistory.com

and,

bclween them

th erefore , determines the location of the

reflcding s u rfaces. Poor


an a l y is techniq ue . ,

and

ignals c a n be enhanced th rough


f the underground terrain

a profil

obtained.

C. L. Heizman is a graduate of City C oll ege

of New York (8

1955) and Columbia

University (MS - 1 956). He was involved in

TIME-SERIES ANALYSIS
The prece d i n g pr blems ca n a l l be olvc d \ ith various

ic corn pu tati o n a l rou tines t hat con


stitu tc the to I of time-. erics analysis. By a time scric or
ign a l, we im p l y mean a succession of data value resu l t i n g
from or sim ul at ing a phy sical p rocess. The data can be

corn b ina t ion

of the ba

research and product development of inte


grated circuits, high-speed computers, and

data-analysis

instruments prior to joining

Time/Data Corporation (a GR Subsi iary) in

1 968. As Applications Manager at T ID. M r


Heizman devotes much time t o real-timeana

lysis. He is a m ember of 1 E

E.

amount of rainfall, earthquake tremor , elec tr ic voltages

i me i us ual ly
t he independent vari a ble, but n ot always. T h e comm n thing
about all these records is t h a t the s ucces ive data values in
each of the tim e serie. arc rela te d in som e way to the ther
value in t h a t t ime series, a n d perl1aps to value in omc o ther
time series. T h e e re l a tions h i ps may be detc rmi n i . t i r t a t is
tical. T i me- erie
t u d i es a n a lyle these r la t i n hi ps so t hat
fr m a brai n , or a n y other physical process.

the phy sical process can be better understood.

can then be
physic ally.

mo deled , or

imulated, either m at h ematically or


sis procedure. required

The basic signal-a n a l


ins tru m e n t

can

be

g roupe d

in a

practical

into t hree mea urement cate

gories. as shO\\ n in the table below. The proce

must gcncrully be

his proce s

fast enough

s i ng fu nct i on

o that data a re pr cc sed in

rc<.1! t ime , that is, as fast as t he y are bein' acquiretl. Real- t i me


processing al o is ncces. a ry becau. e the
s.im p l e s i. usually much l a rger than t he
a n y practical ma hi ne .

In a

n umb e r of i n p u t d a t a
m e mo ry capacity of

ty p ic a l mea sureme n t , t he i n p u t

sampled for I 0

samples /seco n d . The total number of i n put


sign a l

may be

m i nut es

a. tron omic al hu t , after bei ng pr


analysis p rocedure , t h e d a t a may
JOO numbers.

a m ple b ec mc

.it a rate of

ccssed

by

the

I 0 0 000

a p propr i a t e

h a ve bi:cn reduced t o o n ly

1. Correlation analysis
2. Spectral analysis
3. Filtering

ome fur t her

pro c essin g

b e perfor m e d o n this result to p u t i t i n to t h e form mos t

the information being sou g ht . Other pro


thi by perfo rming basic a r i t hme t ic
coordinate t ransform ation , . moothing opera

suitable for show ing

ces.i n g f u n c t io ns a l l o w

tions, and time / frequency transformations.

r he most i m port a nt time-series analy is tool

quare d d i fference term


mc<isure
the

to o b ta in a

f the similanty. T h i

i ngl e num ber tha t i a

nu mber, w h e n normalilcd by

number of i n d e pen dent meas u re men ts, is the

quarc d i fference. It can then


pl acemen t s of

be

me a n -

cal c u la te d for variou s dis

ne sign al with respect to the other. By si m p l e


a m e information

alge bra, y o u can show t he

by calcu l a t ing

the corre l a t ion, or covari a n ce. f u nct i o n which is the s u m


f the ord i n a t e-by-ord i n a t e
f r111s. * The

re s u l t
ue

show. that they

of 0.7 millisecond

m u l tipl i c a t i

for the t wo r a n d o m

F i gure I

n o f t h e two wavc
ignals of

most si m i l a r w h e n t h ere is a dis p l a ceme n t

between them. If

t h ese signal

represe n t

t h e v i brat i n level a t t wo po i n t s o n a stru c tu re, this time is the

pro paga t ion t i me for vihra t io n s between the t wo poi n t s . The


ign of the dis p l acemen t i n d icates the direction of pro paga

t ion.

M ea uring t ime d e l a y s in this wa y leads to m<iny useful

a p plica lions.
lead of calc u l at i n g t he cross-corre l a t io n function of

a sign a l w ith

Figure

be t w ee n sign a ls . S pe c tr a l

are those

provide the a m e m e asu re o f si m i lari ty ,

mea n -. q u a red di fference

between the ig n a ls,


diffe ren l f rms. Although pc tral
more fa m ilia r to most people tha n

of si mi l a rity and therefore will

in

I, sh ow that

the upper signal of F i g u re

F igu re

2. for t h e t wo

ucccs ivc a m p les of

arc more correlated, or arc more

F i gu re 3 , o f a inc-wave
the sine w a ve is n t rea d

de pende n t up n i:ach other. t ha n those of t h e lower ign a l .


nothcr exa m p le i
signal

b uried

s h o w n , in

in noise. Fven though

ily obser able in t he origi n a l signal , the u u t o-correla t ion func


tion shows its p rese nce clearly. 1 h1. abili t y to det ect periodic

com ponen t s in a sig n a l i

an impor t a n t applica t ion of the

a u t o-correl a tion fun c t i o n .

The measure m e nt of simil arity is often ca icr to in terpret

analysis, c rr c l a t ion a n a l ysis, a n d f i l t e ring prov i d e this infor

tha t give a measure of the i milari t y

correl ation. the l a t t e r offers more insight in t o the concept of


measure mcn t

ord in a t e,

Spectral Analysis

MEASUREMENT OF SIMILARITY

but present the re u l t s in

ordinate by

that arc porp rtional only

The au to-corre l ation fu n c t ion

res u l t desire d . Frequentl y , however,

analysis and filt e ri ng arc

quantitic

to the magni t u de of the difference, and then to sum all the

random sign a l s of

The p roduct of t he pr cessi ng operation is often the fin al

the

each term to give

succc si ve s a m p le of a signal are related.

C. Measurement of Statistical Distributions


1. Probability density functions (amplitude distribution)

based u pon

. quar

the two waveforms of hgu re I

one from the oth e r ,

it elf to give the auto-correl a t ion f u n ctio n . Th is sh ws how

B. Measurement of Waveforms
1. E nse m bl e averaging

mation. I n fact, t h e

way to compare

to subtract

two diffcrcn t s i gn als a!\ a hove, you ca n co rre l a te

A. Measurement of Similarity

operations,

n a t ural

is

In

Basic Signal-Analysis Procedures

must

Correlation Analysis

be d iscu s ed fir t .

\hen the opera tion is done as a pcctral cal c ulati on.

*The

It

is a l so

mean-square difference is
En

1"

( n

Ln

(L.::
ll

Xn2

Yn rl2
+

L11

Yn2-r

::!

Ln

(x.,

Yn-r)2)

All the information about the similarity of the signals i s contained in


the third term, which is a !1iaximum "hen the signals are most imilar.
.
This cc r m 1s chc correlation or covariance function.

GENERAL RAD10

www.americanradiohistory.com

Experimenter

0
0

TIME - MILLISECONDS

Figure

rLAG-0.I ms/POINT

100

100

0
0 0.7

TIME- MILLISECONDS

13ms

1. C ross-cor relation f unction of two waveforms.

r LAG - 0. 1 ms/POINT

13 ms

SINE-WAVE SPECTRUM

><

(/')

><

>
><
a:

.t

Figure

>
><
a:

13ms

r LAG - 0.1 ms/POINT

Figure

2. Auto-correlation functions.

m o re com mon to see the m easurements in this form because


in the past spectral analysis w as easier to implement with
electronic instrument than correlation a nalysis.
Spectral analysi i based upon the theorem that any repet
itive signal can be considered to be the um of inu oidal
components w h ose frequencies a re integral multiples of the
ba ic repetition frequency . Fourier-ttansform a nalysis, famil
iar to most engineers and scientists, involve the calculation
of the a m plitudes and phase a ngles of the e component. In
principle, the c alculation is really a cro correlation of the
JULY/SEP TEMBER

FREQUENCY- HERTZ

5k

3. Analysis of sine wave buried in noise si gnal .

signal with a sinusoid of each of the possible h a r m onic fn:


quencies, re pectively . The result is always another .inusoid
of the amc frequen cy, w h ose a m plitude and phase arc pro
portional to the corre ponding c o mpon e n t in the signal. In
practice, the ourier-transform calculation is done in a much
faster, m o re direct way by ef ficient computational methods
that have been developed in recent year_
Auto-spectral analy sis involves th e calculation of the
squared magnitude of the Fourier spectrum and is the quan
tity produced by most spectral analyzers. Because it is pro
portional to the power of a ignal, il i com m only called
Power pectral Density ( PSD). It gives exactly the same in for
mation as the auto-correlation function: in fact, it is the Four
ier transform of the auto-correlation function, and it can be
calculated in that way .
A spectral measurement that is not commonly availa ble
from analog spectrum a n alyzers, but wh ich is extremely use-

1970

www.americanradiohistory.com

TIME DOMAIN

FREQUENCY DOMAIN

TIME SERIES

FOURIER

TIME

SERIES

FOURIER TRANSFORM

y (!)

Figure

CROSS

4. Time/frequency-domain diagram.

COMPLEX
SPECTRUM

TRANSFORM

Xt1wl

COM Pi.EX
SPECTRUM

CONJUGATE MULTIPLICATION

CORRELATION

CROSS
CORREL ATIQ
FUNCTI ON
Ra: Id

FOURIER TRA SFORM

CROSS
SPECTRAL..

i ble if the fil t e r is to be ea ily v a r i a ble . With

ful, is the cross-spec t ra l func tion. This func t ion is norma lly

cally impo

calcul a t e d b y the mul tipl i c a ti on of the Fourier t ransform of

digital implement ation, the f i l ter design become

two s ignals.

merely the s pecifi ca t ion in eithe r the time or frequency do

ike the au to-spectral funct ion , it con ta ins exact

ly the same informa t ion about the

irnilarity ofthesignal as

the cros -correla tion function, and i t can be calcul a t ed from

the c ro ss-corrcl a lion by the Fourier t r:in form.

main of the waveform lo b.: detected - and the f i l t e r char


a c teristic can be changed in mic roseconds .

MEAS UREMEN T OF WAVEFORMS

The relation be t ween co rrela t io n and spec tral functions is

Ense m ble av eragi ng* is ve ry useful for deter mi ning t he

i llust rate d by the ti me- and frequency -doma in map of Figure

4. It shows that the corre!a tion func t i o n can be calcula te d in


two way s : d i rectl y in the time domain, o r indir ectl y in the
f requency domain via Fourie r t ransforms.

The fast al go

rithms developed to calcula t e Fourier tran forms have en


bled

the cli.:ulations to be done m ore quickly via the f re

quency domain than in the time domain and thu

have nrnde

the Fourier t ransfo r m the key opera t i o n in modern t i me


scr ie

a na lyz ers.

The a u t o- spe c t ral function for the ine-wave signal buried

in noise i. also sh wn in Fi g u re 3, along with the a u to-correla


tion function. The sine-wave signal is clear!

discernible in

ei the r function.
A filte r can be considered a s a device that continuously

compares an input signal with a

lored refere nc e signal and

a ma xi mum output w hen the[\ o are mo t i mi l a r.

The s tored reference signal i

the i mpulse response of the

fille r, i.e., the response of the fille r to a ve ry n a r row pulse

applied at the input. The c rite ri on for the measurement o f


imilarity is the same a

for corrcl alion and srectral an alys is :

in fact, t he ca!cu l ati ons arc carried ou t in e x actly th e same

way. Figure S sho w

shape of a sig nal thnl is obcured by random noise when t he


sig nal is repe titive or when its time of occurrence is k nown.
his la t ter condition exist when a s ystem i

being

timulalcd

in a controlled manner.

An example is the clectroenceph:ilograph (EEG) signal

pr o du ce d by the fla hing of a light in a per on's e y e. The


res ponses to this sti mul a t ion (evoked re pon cs) arc added or

averaged t o gether . The

ignal-to-noise ra t i o is in cre uscd be

cau c the ignal component s, bi.:ing in rhase

i t h each o ther,

will add linearly while the noise components, hcing random ,

will p a rti alJy ancel each other and add at a rak pro po rtio nal
to the square root of t he number of average .

F i ltering

pro duce

trivial

a fi lter that was designed to de tect th

presence of a " chi rr " or sw ept- frequ ency

inc-wave

En emble averaging is a lso commonly applie d lo th e mca

suremen l of co rrelati on and spectral functi o n to improve t he

s t a tist ical accuracy of the measurement. Consider t he auto

spcclral measure m ent of a short segment ( I 000 samples in

t his case) of a fil t ered random noi e signal, as sh ow n i n Fig ure

6.

The au t o-spect ra l func t ion it sci f is a lso a random fu nction .

vc rag ing the auto-s pec tra l measurements of successive seg

ment s of the ignal reduces the statistical variation s .

MEAS UREMENT OF STATISTICAL DIS TRIB UTIONS

The a mplit ud e histogram is o ften the fi rst meJsurc mcnt

ignal

buried in noise. This type of signal is used often a t he t rans

made in the an aly sis of random datu. You dekrmine i t by

on a r, and geophysical echo-ranging

div iding the a mpli t ud L range into many equally r a ced levels

the signal is rrc ent and has the me shape as the auto-cor

the signal i at each level. The histograms of the two random

relation function of the signal.

signal

mit ted signal in rad;.ir,

ystems. T he outpul of the filter increases significantly when

The impl emen tat ion of filters to "maleh " a p res c r i bed

signal is quite di fficult with analog components, and p racti-

and by count ing lhi.: number of t i mes the mc:isur d vulue of

*A

of F i gun: I arc sh O \\ n in Figure 7.

tatistical average

dom process.

GENERAL RA010

www.americanradiohistory.com

Besides

showing the

cvalualcc.J from thl.! probability tlc11,ity

ofn ran

Experimenter

T LAG

-2.Ss

0
-

5 ms/ POI NT

+2.5s

"
"
en

0
0

TIME- SECONDS

Figure

FREQUENCY- HERTZ

5k

6. Averaged auto soectrum of band-limited noise.

Figure 5. Signal detection by filtering.

extremes of signal amplitudes to be expected, the amplitude


distribution

give information about the linearity of the rela

tionship between the

ignals. If the e signals re pre ent the

input and output of a system under te t, the fact that both


ignals have the same distribution, Gau sian in thi case, indi
cates that the system is probably linear.
From amplitude histograms, the mean value, rms value,
and higher-order m oments can be calculated.

SUMMARY
discussed, and some of their applications have been given.
he f"ourier tran form i

Figure 7. Amplitude histograms of random noise.

and filtering. 1any other calculations

often arc required, such as coherence function, transfer func

cessing. It is now practical. by application of modern digital


tcc.:hnology, to build instruments that can ace om plish all

tion * , cepstrum, etc. but these ar extensions of the funda

these functions with the

mental procedures de cribed and are obtained by further pro-

problems.

See

page

8.

JULY/SEPTEMBER

+2

AMPLITUDE - VO LTS

ecn to be a fundamental calculation

required in digital time-series analyLers for doing correlation


and spectral analysi

-2

The basic tools required for analyzing signals have been

1970

performance required for real-world

C. L. 1 l eizman

www.americanradiohistory.com

Measurement of Transfer Function and Impedance


The concept of transfer function is widely employed in the analysis and synthesis of linear systems or
networks. Conventional instrumentation is well suited to the problem when a simple noiseless single
input, single output system is involved, particularly if the rate at which measurements must be made is
low. If not, modern F FT (Fast Fourier Transform) techniques provide an efficient means for performing
these measurements at lightning speed. This paper derives the best (least mean-squared) transfer-function
and impedance estimator, which is shown to involve the ratio of the input-output cross-spectral to the
input auto-spectral density.

(3)

General

The transfer function of a linear network i a popular and


useful descriptor of a system or network. It i a dimen ionles
quantity that relate the input and output by specifying the
gain (or attenuation) and pha c hift at all frequencies. Thu ,
the lran fer function of the system shown in Figure I i
II(J) / 0 (J)

t_j;Jj]
(f) L!PCfl

( I)

Y(f)

i the gain of the sy tern for a frequency of jHz a mea. ured


by the ratio of the output function Y(f) and the input driving
function X(J),

Xtfl

INPUT

LINEAR SYSTEM

H(I)

.
II(J) c xp[Jl'J(D]

( _)

H(J)=
(J)

wher e

is the relative phase shift introduced by the system, as


measured by the difference in phase angle between the input
and the resultant output.
In term of complex notation, the transf er function is

(J)

=y

while

Y(f)

OUTPUT

YU) exp[jl/IU)]
=

(4 )

(J) e pljO(J)]

H(J) exp[jO(J)] =YU] exp UI i,/J(J)


XU)

O(J)]

(5)

The transfer function often must be mea sured for pur


system analysis or simulation. Simulation is particu
larly important when the system under tudy is inaccessible
or unwieldy for experimentation. Examples of thi type
might include simulation of proces control ystcms or sub
sy tern t hat can only be ob. ervcd in ope ration , imulation of
the respon c of pace vehicles to transient excilati n without
running the risJ... of a ct ua l damage, or analyzing and sim u lat
ing the effective transfer function of a vehicle suspension
sy tem. Knowledge of the tranfer function would permit a
convenicn t analog to be contructcd.
poses of

Test equipment ca pa ble of providing the necessary meaurcments has been d e v e lo p e d over the years and generally
includes a sine-wave generator, voltmeter , and a phase-angle
meter as the basic tools. The amplitude of the response and its
relat ive phase angle constitute the n ec essa ry measurements.
A measur em ent is usually obtained when the frequency of
excitation i varied while lhc input is kept con tant.

of this type arc usually


and are particular!} suitable when the. y tern under
ob crvtion is
Standard measurement technique

adequate

I)

Figure 1. Linear-network derivation.

1oiselc s

2) Has a single

input port
GENERAL RA010

www.americanradiohistory.com

Experimenter

Figure 2. Noisy-network block diagram ;


m u l tiple inputs.

I n o t he r words, if the system is as shown i n Figure 2, the


input L1. and t h e output Y t_:l!._ a rc not sim ply rela ted by the
effec tive t ra n fer fu n c t io n desc ri bing t h e gain and pha e
bet ween the i n p u t and Y output, becau e of t h e presen ce of
n u merous noi e ources, pro pagat i ng toward the o u t p u t port.
We fin d , t herefore :

+
+

fu + n - 1 I In / n- 1 + en

n-2 Hn- 1

H n I n-2

+ On - 1 + 8 n . . .

The equivale n t n o i e vector has i n t roduced u n certainly in


t he a m p l itude and pha e mea u re m e n t s of t he o u t p u t vector
Y1.1._. The mea u re m e n t procedure must min i m ize these ef
fects.
n an alogou p robl e m a rises when t he i m pedance of a
noisy net work is mea ured. I n t h is case, t h e i n p u t vector of
igure 3 a n d equation ( 7 ) . Yt.J!.. , become a vol t age or velo
city vector, 1L1_ a voltage or velocit y noi5e vector, a n d L!P_
the c u rren t o r force vector. The equation relating the various
fac tor i :

(6)

where t he vector form AL!!!.= (j) /Cl.(f) .


Thus, t he o u t p u t Y L:l!_ c a n be t h ought of a s t h e su m of t w o
vectors

L1!. zt..2. + Ll

( 7a )

where Zt..!!_ represe n t s the i m pedance of t he network t o be


measu red .

\ i t h these d i fferen ces i n m i n d , t h e results and discu. sions


that follow for estimat ing the t ra nsfer fu nction will apply
d ire ctly for e t im a t ion of the i m pedance fun c tion.

(7)

, - -- - - - ---- -- ,

I
x if_ I

T h u , the system o f F igure _ can b e replace d b y a n e q u iva


l e n t n e twork a shown i n F igure 3( a ) , pro ided that we a re
only concerned w i t h fi n d i n g the tra n fer fun c t io n bet ween
t he i n p u t terminal w here LP.. is applie d and t h e o u t p u t
terminals where Y L2_ is observed. The vector relationsh i p are
hown in F igure 3( b ) .
I t c a n b e h ow n t h a t a si m i l a r i n pu t-ou t p u t rel a t ionshi p
exists for a n y of t he " n " i n p u t ports. H e nce, i f H d!_ c a n be
measur d , " n " s i milar measure me n t s w il l describe the total
set of network t ra n fer fu nctions. Jn ome cases, only one
t ransfer fu n c t io n i req u i re d , that which applie. between the
-input a n d Y-out pu t terminals. ow, t he signals designated
1 ffi t hrough 111 Lln_ can b e t hought o f a . i n ternal noise
a

! :I

I
I va_

I: 'i i{_j

H f__

L _ __ _ ________ J

__

-7xH
x

N
(b )

sources that may n o t be d i rectly observable. In e i ther ca e ,


t he problem i s essen t ially t h e sam : l easure the t ransfe r
fun ct i o n i n t h e presence of o t her spuriou signals, w h i c h we
will conven iently call noi e .
J U LY /SE PTEM BER 1 970

Figure

3. Noisy-network derivation.
9

www.americanradiohistory.com

E . A. Sloane received

Performing the indicated operations in equation ( 9), we


obtain

his B S E E degree

from Northeastern U n i versity i n

1 95 1 .

He

was a staff member at M I T Lincoln Labora


tory ,

radar systems group. from

1 954; he joined

1 95 1

to

Ampex Corp. as manager first

=I

for product plan n i n g , then for adva nced tech


nology plan n i ng,
Vice

1 955-1 965.

President and

Presen t l y , he is

Technical

Ti me/Data Corpora t i o n , a

l= I

GR subsidiary,

0 2

;..
-

least-Mean-Square Estimation

88

Because the spurious signals are treated as a single equiva


lent noise vector, we can e x pre it in term of the transfer
function inpu t and output as follows:

I= I

L: I
l= l
L

where
and

LI
/= l

( 8)

Y1

e x p ( j 1 )

/\
or e x p { 2j 8 ) =
1 e x p (j </>1 ) H e x p ( j l1 )

12

l= l
L

L:

(9)

/\

t1

- X 1 Y1 e x p ( -j (!1 + </>1 - 1/1 1)]

X1 Y1 e x p [ -H</>1 - l/ll) ]

=O

( l 2a)

X1Y1 e x p [ + j ( </>1 - iJ! 1)]

X1 exp ( -j <f>1) Y1 exp ( +jiJ! 1)

l= l

( J 2 b)

X1 e x p (+j <f>l ) Y1 e x p ( -j l/1 1)

The essential e timator for the phase angle of t he transfer


function involves the um of appropriate forms of the prod
uct of t he input vectors, X 1 e x p ( -j </>1) , and the output vec
tors, Y1ex p(+j l/J1).

I n a similar manner, by differentiating with respect to the


gain fu nction, H, we obtain the following least-mean-squared
estimator :

L X 1 e x p(j</>1) Y1 e x p(-j 1/11) + e x p (-j O )


I

2
u mber of independent observ ations_

(1 1)

I= l

H =

L:

L:

/ ex p ( H1) is lhe zt h noise vector


Y1 e x p( j iJ! 1) i the zt h output vector
X 1 ex p(j </>1 ) is the /l h input vector.

e xp ( j O )

l= I

e x p( j l/1 1) -

( 1 0)

ti
X1 Y1 e x p [ J(U + 1 - l/1 1) ]

L:

e x p ( 2j 8 ) =

The horizontal bar indicates ensemble ( or statistical) average.

X1 Y1 e x p ( j (8 + </>1 - 1/1 1) ]

The mean-squared value of this error after L * measu remen ts will be:

so that

e x p( j ) = Y e x p ( j l/J ) - X e x p( j <f>) H e x p (j8 )

- H

Because we desire to minimize t error by choosing an


appropriate e timator * * for the gai n , H , an d the phase angle,
/\
8 , we proceed by differentiating with respect to the pha e
angle and set ting the derivative to zero. Thus,

ber of Tau Beta Pi and Eta Kappa N u .

tJ = Y t_:P - X Ll_

Y ,2 + H 2

- H X1 Y1 e x p ( -j( 8 + </>1 - 1/1 1) ]

D i rector of

which he hel ped found. Mr. Sloane is a mem

or

L:

L X 1 e x p( -j</>1) Y1 e x p(+jij/1)
I

( 1 3)

L: x /

I= l

A n approxi mation to some value function.

10

GENERAL RA010

www.americanradiohistory.com

Experimente r

fl

ote that the estimate for the gain fun xtion , H, depends on
knowledge of the phase function, e x p ( j 8 ), as well as of prod
ucts of the input and output vectors.
quation ( 1 3 ) can be restated as follows

ex p ( 2j )

1 e x p ( j 1 ) Y1 e x p (-j l/.11) +

so that substitution of equation ( l 2b ) yields

L X1 e x p(-j,) Y1 ex p( +j l/1 1)
fi e x p ( j )

( 1 5)

2.: x /

------

Because Equation ( 1 5 ) may be e x pressed a

H e x p ( j ll ) =

For t h e sake o f completeness, we should m e n t ion that


these vector function may be e x pressed in terms of rectilin
ear coordinates re prese nting i n-phase and ou t-of-phase ( co
and quad ) components by mean of the following ide n ti t ies

2.: x ?

( 1 6)

fl

21)

so that

H (j f) = ----

and

"
S x y (f)

( 22 )

2: x1 e x p(-j1) Y1 e x p(+j l/1 1 )

( 1 7)

=L

L..J

!= I

( 23)

l= I

hile the denomin ator is t he A uto-S pectra l Density estima te


S x x of t h e input signal, or

Sx x =

( 20)

Z(f) e x p ( j (3 (f) ] = Z(f) cos (3(f) j Z (f) in fj (f)


= Z ( jf)

the numerator can be recognized as the Cros - pect ral Den


sity estimate S Y of the input and output, or

and

( 1 4)

L X1 e x p(-j1) Y 1 e x p(+j l/11)


I

"
Sxy =

1)

e x p [ j(3(f) ] = cos(3(f) j sinfj ( f)

Equation ( 1 5 ) represen ts the least-mean-squared estima te of


the system transfer fun c t ion.

"'

L: x/

"'

L X 1 e x p( -j</>1) Y1 e x p( +j

L
x2
L L..J I '
I= l

( 1 8)

so that the t ransfer-fu n c t ion estimator becomes


,,
"
H ex p ( j O ) =

/\

Sx yU)

----

( I 9)

/\

I
L..J I X 1 ( j f) Iz
L
!= I

( 24 )

Figure 4 is a photograph of a transfer fu nction o f an elec


t romechanical net work (vibrat ion e x ci t e r o r shaker system )
c tima tcd in t his manner on a Time/ Data M ode l 1 00 Rapid
Fourier A nalyzer. The u pper trace show the in-ph ase a n d the
lower ou t-of-phase or quadra t u re components. The horizon
tal ax is is frequency.
Rel iabi l ity of Esti mate

This is simply the ratio of the cros


spectrum of the i n p u t function.
J U LY / S E PTE M B E R

spectru m t o the auto

Finally, the error in esti ma ti on can readily be determined


from equation ( 1 5 ) by substi t u ting equation ( 8 ), so that

19 70

11

www.americanradiohistory.com

L X1 e x p -j 1 ) [ H cx p( +j8 ) X 1 e x p( +j 1 ) +
l

, c x p( +j i,'; , ) 1

A cx p (j ) = -

(25)

or
/\

/\

exp(j8 )

H expU8 ) +

I:

L: x 1 N1 e x p ( j( 1J,i 1 - 1) ]

"

Sn n (f) = L

This tends to ap proach zero a s the n u m ber o f observa lion


increase indefi nitely, provided t hat the input an d noi e ig
nals are uncorrelated , because t he nu merator w ould represen t
the s u m of a large nu mber of vectors having ra ndom phase
angles ( ljl1 - 1) , while the denominator would be a real posi
t ive de fin ite q u an ti ty t hat increases linearly with the nu mber
of observation .
The mean-squared error,
shown to be
-

;:(,

in mcasuremen t can be

'"
L...J
l= I

/ (f)

( 29 )

( 30 )

where

"

S x y (f) = L

"
L...J

X1 Y 1 exp f j ( 1/11 - 1 ) ]

'

(3 I )

(32)

'" , 2
S" y y (f) = LI L.J
Y

where S nn(f) i s the equivalen t out put-n oise spectral density.

The mean-squared error is inversely proportional to L , the


num bcr of indepe n d e n t observations en tering i n to the esti
mate.

and, by equations ( 9 ) and ( 1 6 ) , becomes

( 28 )

r/' (f) = -L

( 26 )

This error involves k nowledge of the spectru m o f noise ;


therefore, we must est imate t he error by estimating the
noi e spectru m . The n oise spectrum is approx imately

(27)

I: x /

1) ]

------

Therefore, the error i n estimation , 1/ , i n t raduced by t hi s pro


cedure, is
'

1 ex p( j( 1J,i 1

(33)

The mean-squared error, i n terms o f m easurable quan


tities, becomes

(34)

w
en
<l
I
a.
I

<l
(!)

or

w
a:
=>
I<l
a:
a
<l

or

=>
a

Figure

(35 )

[I

- !'({)] ,

(36)

5k

F R EQ U EN C Y - H ERT Z

4. Transfer function of vibratio n -e xci ter system.

w h ere

f(f)

12

!sxy(f) l2
SxxU ) S_ y(f)
GENERAL RA010

www.americanradiohistory.com

(37)

Experimenter

From t h is it is appare n t t h a t
o <

r (f)

< 1.

(38)

T h e fun c t ion, r(f) , i k nown as the Coherence F u n c tion. I t i


.tero when the ou t p u t spectru m is due e n t i rely to noise ( S y y
S n n ). and u n i t y when t he system is noise-free ( S n n O). T h i
i s a popular measure for t h e reliabil i t y of a t ran fer-fu nction
estimate.

In su mmary , we have found t h a t the estimator t h a t m i n i


mized the e ffects of spu rious sign al or noi e sources in vol ed
measuremen t s of the magnitude a n d phase of t h e input and
o u t p u t ignals for each fre quen cy of in t e re t. The e measu re
ments coul d , in p ri nciple, be obtained by mean of voltme ters
and a phase meter. H o wever, if the n u m ber of frequency
poi n ts t h a t mu l be used to de c ribe the t ransfer or i mped
ance function i large, t he n more e fficien t estima tes can be
made by means of digital-signal proce sors using fast Fourier
t ransform te h n ique . The T/ D 1 odels l 9 2 3 A T ime-Series
A n a ly zer, 1 9 2 3 8 R eal-Time Fast Fourier-Transform nalyz
er, or I 9 23 C Fast Fourier- ransform A nalyzer, Figure S ,
directly perform t he operat ions i n d icated i n quation ( I 9),
utilizing t he Time/Data 90 Rapid Fourier Proces or. The
I 9 2 3 A , for example, wil l perform a t ran sfer-fu n c tion a naly is
for 1 28 equall
paced frequencies in less t han 1 0 milli
second per complete e limate.
- E . A. Joane

F igure

5 . T/D 1 923-C

Rea l - n me Analyzer.

F or Further I nformation
Bendat,

J. $ . , and Piersol. A . G . . Me<1surement <1nd An<1/ysis


1 966.

o f R<1n

dom Data. John Wiley <1nd Sons,

Lee, Y. W . . St<1tistical Communications Theory, John Wiley and Sons.

1 966.

Sloane, E. A., "An I n t roduction to Ti me-Series Anal ysis," Mono


graphs I, II, and Ill, Ti me/Data Corporati on .

Th e Hono rable Society

C. E. White

A . P. G. Peterson

On J une 1 6 , 1 970, t he a tional Con


ference of Standards La boratorie p rc
sen ted i t s first Awards for O u tstanding
Service lo t h ree member of the organi
z a t i o n . O n e of the recipi e n t was

Charles E .
h i te of General R a d io.
W h i le present i ng t he award plaque,
C S Chairman J . L. H ayes made t he
following citation :
" M r. Charles E. W h i te has un c l fishly
devoted hi time and energie lo formu
l a t i ng, editi n g, and sustaining the opera
t ion of t he CSL
e wsletter for t he
pa t eight a n d one-half yea r . M uch of
t he credit for the growth of t h i public
ity and i n fo rma tion media is reflected
u pon M r. White, who served as chair
man of t he
SL ewsl c t ter Commit
tee and Editor of t he ews/etter. '

Deviations from
Accuracy

JULY/SEPT E MBER

The
udio E ngineeri n g Soci e t y ha
an nou n ced t he impending aw ard of Fel
low membe r h i p to Dr.
. P. G. Peter
son of General R a dio. Pre en l a t ion will
take place at the annual
wards Ban
quet i n ew York on 14 October 1 970.
Dr. Peterson i well k n o w n in the
field o f acou t ics. I le received t he J o h n
H . Pot t s 1 e m orial
ward from the
ES, in 1 96 , for out l a n d ing achicve
men t in t he field of audio engineering.
He is, also, a Fellow of the I nsti t u te of
lectrical an d Elect ronic E n gineer and
of the Acoustical Society of A m erica.

We are a w a re of t w o small errors t ha t crept i n t o the


M arch-J une, 1 9 70 is uc. The fir t , on page 5 , hould h ave
tran l a ted t h e te t pressure o f 1 000 microns t o I m m l- lg. T he
second, on page I , shoul d h ave specified t ha t switching t i me
of the G R frequency syn t he izer had been red uced to _00 s
( not 200 ms). orry !
13

1 970

www.americanradiohistory.com

i m i l a r to t h e p rece d i n g e x am p l e . W h e n
t h e regu l a t o r is u e d t o cont rol h e av i l y
l o a d e d c a p a c i t ive-i n pu t i n s t ru m en t s, or

i n d u c t ive-i n pu t power s u p p l i e s , o r J U S t
p l a i n o rd i n a r y m e c h a n i c a l y t e rn s ( a l l
o f w h i c h res p o n d t o t he a v rage v a l u es

of t he s u p p l y vol t age ), distort e d

utput

fro m t he regu l a to r c a n q u i c k l y l ea d t o

i n st ru m e n t o p e r a t i o n i n fe r i o r t o r a t e d
perfo r m a n c e.
The a d ve n t o f d i gi t a l i n s t ru m e n t a

t ion has c r e a t e d a n a w a rc n e

, a m o ng

i n s t r u m e n t use r. , of t he devas t a t i n g e f
fe c t s of

N O T A C I N D ER E L LA I N S T R U M EN T !
The fa te o f or d in ary , p ra c t i c a l , well
kn

wn,

a n d w i d e l y-used i n s t ru m e n t a

t ion a p pa re n t l y i

relega t i o n t o d r u dg

ery work a n d to n e a r-obsc u r i t y w h e n i t


comes t o p u b l i c i z i n g n e w m o dels. T h e

a n a u t o m a t ic vo l t a ge regu l a t o r u c h a

t he G R 1 5 9 2 m a i n t a i ns t h e i n s t ru me n

t a t ion su p p l y v o l t a ge a t t h e d e . i r e d low
o p e ra t i n g p o i n t .

a p p a re n t l a ck o f n e w-de ign i n novat ion

not

o t e that all v I t age regu l a t o r

are

a p a b le o f d o i n g t h i . C'o n s t a n t

c o m p o u n d e d by the m a n n e r i n which

vol t age t ran f rmers a n d reactor- t y pe

t h e ge n e r a l p u bl i c d i s rega rds o r d i n a ry

regu l a t ors operate sol e l y to m a i n t a i n

but

t he i n s t ru m e n t a t i o n s u p p l y vol t age a t
t h e n o r m a l vol t age e s t a b l i s h e d b y t h e
power c o m p a ny 's d i s t r i b u t ion t ra ns fo r
R 1 5 9 1 is <i n e l e c t r o
mer. The

c sen t i al

in t r u m e n t s .

cided to c h a l l e n ge t h i

W e 've

d e

a tt i t u d e by w r i t

i n g a b o u t t h e n e w GR

1 5 9 2 V a ri a e

au t o m a t ic vol t age regu l a t o rs w i t h t he

a t t i t u d e of " H o w c a n you do w i t h o u t
it?"
I f y o u a re a t y p i ca l rea der, y

con t ro l over, or access t o ,

u h ave

ne or m o re

mechan ic a l regu l a t or w hose o u t p u t i s


c o n troll e d b y a servo-d ri ve n V a ri ac
a dj u t a ble a u t o t ra nsform e r of p roven
c a p a bi l i t y .

T h e re i s a more i m po r t a n t p o i n t t o

r a c k s of te t e q u i p m e n t . Con c e i v a b l y ,
e a c h i n s t ru m en t i s w e l l e nginee re d a n d
h a.

i t s o w n regu l a t e d

power s u p p l y .

Consc q u c n t l y , y ou fore ee n o n e e d for


a

eco n d , or m aster, regul ator. Con s i d e r

for o n e m om e n t , h o w ever, t h a t a l l y o u r
i n s t r u m e n t s may operate sa t i. fa c to r i l y
w i t h a n ac su p ply volt age o f l 0 5 v o l t s .

I OS

M o d e r n vol t age regu l a tors u su a l l y con


t rol a n y vol t age a bovc t h
vol t

by

mea n s o f a

'

nominal

em i-con d u c t o r

p o w e r-d issi p a t i n g c i rc u i t

to d ro p t h e

v ol t a ge c l ose t o t h e n o m i n a l v a l u e .
I n the case c i t e d a bove, u n der n or

b e con si d e re d ; t h e
i n t rod uce

G R l 5 9 2 does n o t

d i t o rt i o n

into

t h e i ns t ru

ment s u p p l y vol t age - a factor i gn ore d


too often by c u s t o m e rs of regu l a t o rs. I n

a p re v i o u s Experimen ter

a rt ic l e , 1

we

men t ioned t ha t t he G R e l e c t ro m e c h a n
-

i c a l regu l a t o r could t ra c k t h e ave rage

and peak v a l ues of s u p p l y v o l t age wh i l e

actua l l y d e t e c t ing t he rms \'a l u e . T h i

fea t u re h a s sign i fi c a n c e i n a n u m be r o f

si t u a t ions. Ta k e , for e x a m p l e

c a p a c i

t ive - i n p u t d e power s u p p l

w i t h a l igh t

load. S uc h a u n i t respon d

to peak s u p

ply

vol t ages.

A 3% d is t o r ted o u t p u t

m a l c i rcu rn s t a n ces a fai r a mo u n t of h e a t


i. d is s i p a t e d w i t h i n t h e r a c k o f i n s t ru
m e n t . E x pa n d t h e n u m be r o f rack a n d

as

y ou deve l o p t h

n e e d for a n a i r-c on d i

ou t pu t , even t h ough t h e reg u l a t o r h e l d

t io n e d e n v i ro n m e n t to k e e p t h e m e a n

t o a spec i fi e d 0 . 1 % l i m i t o f dev i a t io n

t i me-bet ween-fa i l u re ra t c o f t h e i n s t ru

from a n o m i n a l l y rated s u p p l y volt age .

from t he regu l a t o r cou l d cause a s m u c h

3%

c h a n ge i n t h e d e power-s u p pl y

me n t s from i n c re a i n g rap i d l y . It wou l d

[ f t he regul at o r w e re in con t ro l o f a

a p pear t o b e good l og i c , t h e re fo re , t o

t herm a l dev i c e , w h i c h re. p o n d s t o t he

re d u ce t h e s u p p l y vol t age t o 1 0 5 vol t

rms v a l u e of t h ' s u p p l y vol t age, d i s t ort

n o m i n a l , a n d to m a i n t a i n con t i n u ou

ed re g u l a t o r ou t pu t wou l d

ope ra t io n

t ion o f t he t he r m a l u n i t i n a ma n ner

at

t h a t leve l .

S u c h a s t ep

wou l d re du ce t h e a m ou n t of h e a t ge n
erated and gre a t l y i n cre ase t he useful
l i fe of the i n s t ru m en t s . I n s t a l l a t ion of

p i k e s or s h a r p peaks in t h e s u p

ply vo l t age. False t riggeri n g of d igi t a l

a ffe c t

opera

I C h i t o u r a s , C . . "Considerations In
Choice of a Li n e- V o l t age R e g u l a t o r .
Experimen ter, Oc t o be r 1 9 6 7 .

"GR

The

14

ci rcu i t . i s c o m m o n place w he n s u p p l y
vol t ages a rc u s e d i n c o m m on w i t h d is
t o r t ion-p rod u c i n g i ns t ru me n t a t i o n . Use
of the

G R I 5 9 2 a a bu ffc ri n g o l t age

s u p p l y u n i t to a b l ock of d i g i t al i n st ru

m e n t s h e l ps to red u c e fa l c d ig i t al ou t

p u t s.

T h e G R 1 5 9 2 c<1n eve n be considere d

a a tool t o h e l p m i tiga t e " b ro w n-o u t s " ,


o w i d e l y pre d i c t e d b y m e t r o po l i t an
power co m p an i e s d u ri n g peak-power
Joad ea son s. I f y o u r l ocal p o w e r com
p a n y i s forced to l o w e r sup ply v o l t ages

clrast i c a l l y , use of t he regu l <i l r w i l l


assure con t i n u e d <1 n cl sa t isfa c t ory o p e r
a t ion of i n s t ru m e n t a t ion .
F o r reade r w i t h p ro h l e ms of u p ply
for i l l u m i n a t ion devices. p l a t i n g b a t h s ,
or s i m i l a r a p p l i c a t i o n s d ra s t i c a l l y a f
fec t e d

by

l i n e-vol t age v a ri a t ion

the
10

k V A . L ig h t e r load d e m a n ds u n d o u bt

edly cou l d be m e t by t h e 2-k V A m od e l .


C o m p l e t e s p ec i fi c a t i o n s for t h e
models l i s ted b e l o w a re a v a i l a b k i n G R
C a t al og U . A p a m p h l e t descri b i n g utr

ren t G R m odel of v o l t age regu l a t ors is

ava i l a b l e to reader.. A d d ress y o u r re


quest t o :

l d i t or. ( ,' R Fxperi111c11 tcr


300 Baker

Concord,

v e n ue
1 assa c h usc t t s 0 1 7 4 2

Des i g n responsibil i ty for the G R 1592


G. Ch1 tou ras and W . A.

was s h a r ed by C .

M o n tagu e .

Catalog
Number

Desc r i p t i o n

1 592-9700
1 592 9 7 0 1
1 59 2 9702
1 5929703
1 592 970'1

1 592 Variac
automatic
voltage regulator

1 20 V 1 0% mput
- 0% mput
230/7'10-V
% 1npu1
230/2'10 V 10% input
230/2'10-V 70% input

1 20 V

Price

$525.00
525.00
525.00
525.00
525.00

P r ices net F O B Concord, MA, U S A .


S u b1ect t o q v a n t i t y discount.

GENERAL RA010

www.americanradiohistory.com

G R 1 5 9 2 i s ava i l a b l ' for l o a d u p t

Experimenter

second to conventional cine' and high


speed motion picture cameras when the

I 5 4 1 M u ltiflash Generator is

new GR
use d .

The

imple m e n ting

ystem

use

stroboscope, m u l tifl ash ge nerator, and


still camera to provide frozen action on
a

s i n gle

Polaroi d * *

film

within

15

secon ds of the eve n t . A series of com


plete i mages, each uniquely positioned
o n the film i n time and space sequence,
is available for tudy and action.
A G l a n ce at the Featu res
m ong the n u merous features of the
GR 1 5 4 1 generator are :

GA

1 54 1 Multi flash Generator

Fla h bursts, adjustable in n u m bers

Versatile trigger circuit, designed to

and in tcrvals
accept a variety of i n p u ts in terms of

signals and connec tors

N EW S H O E S F O R A N O LD W O R KH O RS E
For a period of about 3 8 * y ears, the

tion in engine ering and re earch activi

principle of observing high-speed mo

ties, is most useful in

tion

by means of electronic strobo


1
scopcs has been i m p l e me n ted by port

tion of extre mely high-velocity actions

topping the mo

without the use of high-speed cameras.

able G R equipment of various ty pes.

This is the mul ti-fla h or burst-flash

There has been wide acceptance by in

tec h n ique by w hich a single sheet of

of this instru mentation,

fil m is expose d , in con ecutive order, by

dustry '2 ,3 ,4

capable of " topping" fa t motion with

a series of trobe flashes.


The same technique can be expand

out physical contact for quick analysis

as

and of preserving the stopped-motion

ed,

on fi lm for later study. 5

com plex, to provide fo r a pu lse burst

ers, which has not received wide atten-

a burst of individual flashe

A technique, familiar to photograph

ee pag e 1 8 for a partial


original GR strobe article.

repri n t

of t he

test require me n ts become m o re

from a single stroboscope ( Figure 1 a) or

tiple strobes ( F igure

fro m m u l

1 b ) . Tt provides

equivalent-shutter spee ds of a micro-

Flash in tervals that can be calibrated

Adap ta bility

Small , light , an d rugged construc

Highest intensity retained at I 0-s

Burst m od e proVides for initial signal

tion
i n tervals

to activate contact-bounce and noise

rejection circuits.

Stopp d" I n I ts Tracks


M a n y l!pplications i n motion-analy
sis work have been developed d u ring the
years

t h a t the stroboscope h a

been

**Reg i s tered trade mark of the Polaroid Corp.

0 0 0 0 0 0 0 0 0 0
1 2 3 S ll 7' 1 9 IO

0 0 0 0 0
'6 T 8 9 10
0

Comero

FreQuency control

t o existing strobo

scopes

C a m ero

FreQuency control

M u l l i f l o s h generotor

M u l l 1 flosh generoror

a
F i g u re

J U L Y / S E PTE M B E R

1 . F lash-burst tech n i ques . a. Si ngl e strobe u n i t ; b.

19 70

M u l t i s trobe uni ts .

15

www.americanradiohistory.com

FILM

APERTURE PLATE

F i gure

wi t h

U'i.

s t u d ic

2. Gol fball stroking anal ysis - upper , toppi ng ; lower , good lift .
Some c o m m o n uses i n c l u d e

of sp o r t s e q u i p m e n t i n a c t i o n ,

halls ( Figure 2) or t h e m o t i o n of t o p

such as t h e i m p a c t a n d fl ight of gol f


pl i ng

bow li ng

pins.

A t hl e te

and

woul d-be a t h l e t e observe t h e i r form i n

p h o tos t a k e n w i t h s t ro b e ligh t in g. T h e

veloci t y o f proj c c t iles ( F i g u re s 3 a n d 4 )

Figure 3. I n strument arrangement for projecti le-motion study.

You c a n o bserve t h e accele ra t i o n o f

d i ffere n t a re a s o f t ile f i l m t o p rod uce a

of a n o t h e r t c c h n i q uc. The e n d of

record , l i k e t h a t of t he b u l l e t s t ri k i n g
t he st e e l p ri ng i n Figu re 4 . y n c h ro n i

a h i ghly re fl e c t i v e , ro t a t i n g s h a ft by
me a n

the haft is p a i n t e d fl a t b l a c k w i t h a sin

gle

p e ri p h e r a l

w hi t e d o t .

V i e we d

hy

m i cro p h o n e

detects

t he

s trobe d l igh t , t he single d o t gives a p i c

b u l l e t 's sh ock w ave ) a n d i t is p a rt i c u l a r

from w h i ch

t h e b u r t . T h a t i s , t h e fi r t p u l se o f t h e

t u re w i th a se r i es of well-defi n e d dots

the accel e ra t ion can he

and the ac ce l e r a t i on o f m a c h i n e ry u n i t s

compu t e d . A p ra c t i c a l v a ri a tion of t h is

c a n b e d e t e rm i n e d by t h e use of t w o o r

tech n iq ue i s t h u s e o f s c e ra I s t robe

m o re fl as h es.

u n i ts, p l u gged i n to the t ri ggeri n g j a c k s

/\. d e s i r a bl e fea t u re in p u l sed s tro be

z a l i o n of t h i s s y s t e m is q u i t e si m p l e ( a n

i n c x p e n i ve

o f t he GR

1 5 4 1 b u t n ot a dj ac e n t t o

ly

e n h a n ce d

hy t h e c o h e re n t n a t u r o f

burst is p r o d u c e d m i c r o se c on d s a ft e r

t h e i n p u t - t rigger s i gn a l is a pp l i e d .

The ch oice be t w e e n the single-s t robe

sys t e m in F igu re l a and t he m u l t i p lc

l i gh t i n g is t h a t t he m ov i n g o bj e c t u n d e r

each o t h e r . B y s k i p p i n g j a cks, y ou c an

st ro b c sy s t e m i n F i g u re lb w i l l d e p e n d

observ a t i o n

i ncrease t he p o i n t separa t ion a t l o w ve

on

ge t

o u l of i t s o w n w a y

bet ween flashes. If i t does n o t . su cces

sive i m ages w il l ove r l a p a n d "wash ou t "


t h e observed ac t io n . Overl a p i s p er m i s
si ble , however, i f p i c t u res a rc t a k e n for
d a t a or r e c o r d purposes o n l y , i n a s m u c h

l oci t y , t he re b y i n cre asi n g re s o l u t io n o f


t he t e s t d a t a .

You can se p a r a t e consecu t i ve i m;igcs


w i t h a d i fferen t t e c h n i q u e - shadow

p h o t ogr a p h y . T h e com pact a rc i n t he

as t h re e o r f o u r overl a p p i n g i m ages can

G R 1 5 3 1 , 1 5 3 8 , a n d 1 5 3 9 s t robc l a m ps

n o r m a l l y be resolve d . I f c o l o r fil m is

a p p ro x i m a tes a p oi n t source of l i gh t ,

used , w i t h d i fferen t colore d filter over

e a c h st robe l igh t ( F igure l b ) , over


l a p p e d - i m a ge recogn i t i o n is s u bst a n t i a l

ly i n creased .

w h i c h can cast u n u s u al l y s h a r p shndows.


By u

f m u l t i p l e , separa ted strobes in

t he sy s t e m o f F igu re 3, the i mages a l

each s u cce e d i n g i n t e rval o f t i m e fall o n

t h e re q u ire d O ash r a t e d u r i n g t h e

b u rs t a n d t h e req u ired e x p osu re gu i d e

n u m b e r . The fl ash r a t e fo r t h e s i n gk

s t ro bc sy stem is l i m i t e d Lo the m a x i
m u m a l lo wable r a t e fo r e a c h i n t e n s i t y

nmgc se t t i n g . F or e x a m p l e , a G R 1 5 3 1 .

1 5 3 9 , or 1 5 40 t: a n be fl ashe d u p t o 400

t i m e s per . e co n d , or a

h igher flash ra tes re q u i re l o w e r i n t e n


sity

e t t i ngs : conse q u e n t l y , l a rge r l e n s

a p e r t u res arc n:q u i red . resu l t i n g i n re-

GENER AL RA010

16

www.americanradiohistory.com

G R 1 5 3 8 up l o

2 5 0 0 t i m e s per secon d . U n fo r t u n a t e l y ,

Experimente r

F i gure 4 . M u l tiple-flash shadowgraph sequence.

d u ce d d e p t h of fi e l d . Th is d e c rease i n
e x posure c o r res p o n d s t o a p p r o x i m a t e l y

2- 1 / 2

f-s t op s per i n t e n s i t y -range s t e p

be et l

for G R st roboscopes. W h e n m u l t i p le

st robe

;i re u ed, each may

highest i n t e n s i t y

Some Techn i cal Points


The G R 1 5 4 1 M u l t i fl ash G e n e r a t o r

can p rovide fo r b u rst gro u ps o f t w o t o

i ts

ix t e e n

e t l i ng , t h u s recover

be t we e n

i n g as m u c h a s S to 7- 1 / 2 f- t o ps, w i t h
conseq u e n t l y i n crea e d d e p t h o f fi e l d .

E a c h s t ro b e is s i ngle-fl a shed , so t h a t t h e

n a l sources, l' l e c t ri c a l , p h o t oe lec t ri c a l ,

f l a . hes,

flashes

with

s e p a r a tion

con t i n u o u l y

a b l e fro m 1 0 m i crose c o n d

a n d c l c c t r o m c h a n i c a l . T h e t rigger c i r
c u i t is designed to rej e c t noise or sign a l .

t h a t o cc u r a f t e r t h e i n i t ia t i n g s i g n a l .
hish - i n t c rv al c o n t ro l has a b as i c e r ro r

adj u s t

l i m i t a t i o n o f 3%, b u t p ro v i s i o n is m a d e

t o I 00 m i l l i

f o r a c:i l i b ra t io n s i g n a l t o a n e x t e rna l

sec o n d s . Trigger c i rc u i t s in t h e in t ru

eil' c t ro n i c c o u n t e r i f grc <i tcr a c c u racy is

m e n t pro v i d e f r a v a r i e t y o f i n p u t-sig-

re q u i re d .

resu l t i n g m a x i m u m b u rs t freq u e n cy i s

I 0 0 , 0 0 0 p e r ccon d , l i m i t e d o n ly b y t h e
G R I 5 4 1 ge n er a tor.
Using

t robe

l igh t

as

the

shutter

e l i m i n a t es b l u r a n d seve ral d is t o rt i o ns
pre s e n t e v e n in high-speed c a me ras a n d

o ft e n

p r ov i d e s a m ple l i g h t i n g a l c o n

input

power.

ilicra b l y n: d u ce d cos t , w e i gh t , a n d l i ne
F o r e x a m ple , fou r G R

1 5 3 8 s t ro bosc o p e . a n d a 1 5 4 1 ge n e ra

tor

can

be

used

to

p ro d u c e

1 0 ,000

fl ashes-pe r-se cond l ig h t , a b o u t t h e u p


per s p e e d l i m i t fo r a good f u l l-fra me
high-speed c a m e ra .

R E F E R E NCES

1 . Lamson , H . W . . " T he S 1 r obo sco pc . " Gen


eral Radio Experimenter, De c e m b er 1 9 32.
2 . F i tzmorris. M . J . .

Mod rn

cl

al , "New Eyes for

I nd us lry ," General Radio Experi

menter, September

1 960.

3. F i tz m or r i s . M . J , ' T lash-Oelay U n i t S i m
p l i f i es Motion Ana l ysis i n H igh-Speed Ma
c h i n es , " G R Experimenter, August 1 963.

4. H o l tj e , M. C., ' T lash ! A N ew Strobotut

Development of the G R 1 54 1 was b y C . E .


M i l ler an d W . F . R o ger s , w h o a l s o c o l l a bor

a ted on the above materi a l .

Spe<. i f ic <J l i o n s for l he G R 1 54 1 M d t i f l ash


Generator are i nc l uded as a lc<Jr sh ee t 1 n t he
l.md of this issu
Catalog
N u m be r

Desc r i p t i o n

electroni stroboscope, " GR Experimen ter,


April 1 966.

5. M i l ler . C . E . . "Detai led V i ew i n g i n A rn


b i e n 1 B r i g h t n e s s . " G R E x p erimen ter,
September /October 1 969

P r i ce

1 54 1 -9 7 0 1

1 54 1 M u l t i f lash

$675.00

1 54 1 960 1

Generator
Cab l e Assembly

7.50

P rices net F O B C o n c o rd , M A , U S A .
S u bjoct t o q u a n t i t y d i sc o u n t .

R e ports from t h e F ie ld
As a n os t <i lgic t o u c h in t h e l as t i sue

of the 1xperime11rer, we i n c l u d e d a re
p ro d u c t i o n of the o r i gi n a l <i rt i c l e t h a t
a n n ounced
Bri dge

in

the

1 933.

GR
Lt

650

I m pe d a nce

brought rea c t i on.


of

from t h e re a d e rs , all plea a n t . One i n


particular,

E ve re t t

chncr

San

D i e g o , C a l i fo r n i a , was please d to d e
scnbe

t he

manner

in

rn o de rn iz d his o w n b ri dge , p u rchased

l ie

on the surplus m a rk e t t h ree y e a rs a go .

( I 00,

u. e d t he bat tery box to house

s i s t o r i z e d s i g n a l gen e ra t o r

t r <i n
400,

l k , 4 k , a n d ! O k l l e r t z ) a n d a n os<.:i l l o
scope

d is p l a y . The l a t t r fe<J l u re per

m i t s t h e o p e ra t o r l o b al a n c e n u l l p o i n t s

quite simply and v i ually.

w h i c h he h a d

17

J U L Y / S E P TE M B E R 1 970

www.americanradiohistory.com

O U T O F T H E PAST

The G E N E RA L RA D I O

EXPE RIME NTE R


VO L . VI I . No. 7

DECEMBER, 1 93 2

E L E C T R I C A L C O M M U N I CAT I O N S T C N I Q U E
A N D I T S A P P L I C AT I O N S I N A L L I E D F I E L D S
T H E STR OBOSCOPE
t ro bo cope

ITJ\I I E

con . i s t s

m o em e n t i

The fi lm ca n t he n hP proj ec t ed a t n or

\\ h ie h pe r m i L , i n t crm i l t c n l
.
.
.
obs n a t 1on , e i t h e r v 1 u a l

8t ru<' t i e . o r <'VCn ha< " k \\ a n l w i t h r c -

such

a m a nner a

du e

th

The

f,

m o t ion.

l o "' - m o t i o n

pic t u re

is

ex a m p l

inter

fa m i l i a r

of L h e

est ing

p ro fi t a b l e

an d

i n fo r m a t i o n

w hich

m a y be t l er i " ed from a
lei

u re l y

e ve n t

'\

tudy

h ic h

of

nee s

t a k e p l ace at a
h i g h ra t e of I eed . The

saril
t

nni

player

slow

ca n n o t

c h a m 1 ion-

t he

1cs11 l t s

t he

Tn E

decei

know

t o e_ c l u d e I igb t from
quick nes
a

th

of t he hand

es t he eye. B u t tb
trick or t wo,

aided b y ingeniou

and,

rea ona b l e

m ee'.' h a n

' t rob

cop ,

bu t

not a l l

t u re

The

t ro bosc o pe

n u m ber of
The Edgc..r t un

bop

b u t ter,

opera t i n g

at

u p t he a c t i o n i n t o a

m a lJ eleien t s,
1 r obot1<'0 f i._

bort t h a t

t i l l be

r-v e n t i
rat

\ iew

t he

of

br ie f

th

na

eeo nJ

s t roboscope,

l y pe is t l ie ou t s t a n d
i ng e x a m p l C' . O l n i o 1 1 s l y
t y p1'

fo r v i s u a l obsen a t io n s .
mu t

of w h i h t he E d ge r t on

5.

tar

am ra

Thi

of

S l ) le

ap

t lti

a re ea rn ra .
h i g h spee d ,

scope i

is

for

exce p t

t,

fla h

w i t b . The Edgerton strobo

a mera

t he

t he l i g- h t from t h o b

plications a r (' described bere


cope on page

t ha l

t a i n d L y s h u t t i n g off

'1jg he r speeds t h a n the t rick


ster's hand ach ieves. H ence
tb strobo cope, which is nol
n w, and tbe Edgerton s t ro b
troboscopcs and t b ei r

for

a m e resu l t c a n be o b

i m s , i t i s n o l deceiv d by t h e

oscope, w hich is.

exce p t

fi l m

b r i e f f l a hes. 1 t . ee m s

gyrat i ons of m achinery a l far

l a ggard

m o t i on p i c t u re

th

i11-

t i o n o f t he s h u t t e r is

of t he novice, b u t t he cam ra ca n ,

anJ

a re

that

t h a t m a y be a m u i n g:. T he fu n c -

t roke t o ac om

hip

mod a te
ey

uh

\\ i t h

to

speed

or s t op, t h

peed

mal

o r phot ogra p h i c , of a m o...- i n u o bje c t i n

n o t a p pare n t i n a n o n .

de i < 'c

f11 n d a m en t a l l -y of a

u.

viewed.

ed .

if

to

of

P h o t ogra phy
n o n -rC" p e a t e d

tore t h

a n d to rekas

s t ro b o -

\\ e l l a d a p t e d

l c me n

t hem l a t r a t a

t h a t t he e y e a n d m i n d c a n f o l l o w .

w ho

on ider, h ow ev

ten n i s pl a yer

r,

a n i n d efa t i ga b le

r e pea t s h i s st rok ,

ril("velopmenL of Prof. Harold E. Edgrton, MasaC'hu"<"Hfl l n " t i tutt" of Tec-h,.olo

18

www.americanradiohistory.com

"

i d e n l ica l ly

on

Jf Lh

th u a n d t im e s a m 1 1 1 -

u t e i n a d a rk necl room .
b

l ig h t

Th

c1 u a l l o h i

w i l l a p pe a r a
con t i n uou
p ell he

t rok i n g ra t , h e

t hough m o t ion I '

I f t he

i l l u m i n a t ion.
l i gh t l y

d a rk,

fla h

<' I arl

tr

fk he

t i me t he l i gh t

mad ly

reta i n

t he imag

t roke
t he

bc t \\ ee n fla

t he

pla) ing or o t her

ed . l\ l achine can, and


w her \er om p l ica l d ma h i ne are d -

w1

, canno t repea l u n i form

a n y such

des a l

ignecl, b u i l t , or u

d, t he a b i l i t

to

w a t ch t he i r o p ra t i on i n slow mol i n
' i t hou t pho t ogra p h
The

t robo c o p

p rm i t

t op p i ng

t he mot ion of the ma h i ne ( v i 1 1 a l l )

a t any p a r t o f i t

or prod u t

opera t i n g cyclc ' h i l

t hP grom m e t

flow i n t o t he bopp r ' a l

u n d i m i n i hed

I ccd .

r,

p rha ps,

sq ueaking d u l ch , a ,- i br a t i n g
I \\ en a llC\\

land

be

. low

\ \ h id1 " i l l not w a i t long.

l ic

111 t ion

t u d ) \\ i l l shO\\ t he t ro u h l

t he primar
and

th

m o t ion may h

hand a t forma l guard m ou n t .


a

omet in1

th

q u i t o-bru

bing
a

f r t he eye ev n w i t h t h pri
moti n
t opp d. If r phot g
of t h

p l a e a t too high a

ra p b y i s re orted to for a
i n g dow n

com l

lo\

L ran ien l .

l i t t le con idcra t ion o f w h a t i


i n g done h)
cienl to

one.

mean

of L imin g

th

sponse to t he flash con t rol are


t ial,

ot herw i e

of

object

th

en

will

be

v ie \\ e d a t i rregu lar i n terval , a nd


hra t ions

not

fl a

pr

nt

in

t he

Ii m u s t be of e x t rem I

iewe<l w i l l b
Th

uffi

cope i

a prompt and ac u ra t e r -

aceura t e

flash a n d

be

c l u p t he req u i re me n t

sa t i sfac l or
An

t he . l rob

be

t lo b

obj

omplel

A more d ist inct image i o b t a ined al part i a l


comp e d o f m k

t h ro u ghou t the cycle of motion,

huller open

i n t r du e d .

d u ra t ion. O t herw ise a p p recia b l

1-

objecl

hort
mo-

(I)

w h re 11 i an ;n tegral n u m ber, i t w i l l b
dent t ha t

aeh point o f t be obje t w i l l

exactly t h

a me p o 1t 1on 111 1 ls

tbe bod

On

j.

h i' i n

e l e of motion

n d i tion of " pt>rfcc t"

sy n -

" i l l bc arrc ted, so t ha t i t w i l l app ar

lo be stati nary at some particu l a r phase i n i 1 s


th

ti rl. pro ided t h a t t he opening of

b u l l r or t be flash of t he lamp i

l remel

short d ur a t ion. Tf t h i

of ell.

int rval o f ob

of suffi ieu t d ura t ion, the mo' ing

ervation i

object, even w he n v iewed

st roLoscopica l l ) ,

pc rce pt i h l

w i l l app a r blurr d i n o u t l i ne, . i nc<' ('ach poi n t


o f the body exeeu t
motion d uring t h

rnk spok

s.

Further,

it

hown t h a t such a \1 heel will

P spokes

t a tionary w heel having

(3)

PU = 11 1-'
o t her h a n d , the

th

t ime,

w h(' I

having

I'

spokes w ill app a r as a b l a l ionary w heel h a \ ing 11 /' spokes whenever

( I)

hows th al t h en..

11 PR = F

R cference Lo equation (3)

are, the rctica l l ) , an i n f i n i t e n u 1 1 1 Le r of v a l ue'

ccorc li ng l , all apparen t motion of

c cle of m

in

e. g., by a

F f or w h ich a w heel of J.> bpokes w i l l


h e c n as a s t a t i n a ry w h eel o f f .J spokes. The

of R o r o f

a t ca h obs rva tion, rcs11 1 L i11g i n w ' 1 a t ' e ,hall


desi g nate as a

spaced,

equally

w bt>never

u p po c fu rt her that t he ob

R = nF

h a v i ng

appear a

of mo

na hes of l i g h t in u n i t time. Th n, if

part

c a n read i l y be

ob er ed i

cy cl e

I.

a d isc w i t h

synchronism i f t h e body i
i d en t ical

seen

F' brief. u n i form l y t i ml'<l i n tervals,


i l l u mina ted by F u niform i n s t a n t a n e u

a mou n t of

l l1cc

l a rger the v a l 1 1

f P, t l 11' greater w i l l he t ht'

n u m Lt>r of

p a r t i a l sy n c h ro n i s m s " h il'h

occur w i t hi11 a given ra u gt> of v a l 11r of R or


Th s

CaN

st roL bCop

are of i m por t a nce in


LO determine

or

t h e frequcney

speed of cyr l ic mot ions.

We haYc so far a n a lyzed t he fund amen t a l

l a w s of L i t <' st roboscope f o r
act

synchronism,

e i t her

con I i t ions of e x

p11 r t i a l

or

pcrfcl' t .

Consider now t he case " h!'re t he cyclie frc

q uenc

of motion i s s l ig h t l y grl' a l ('r t h an an

i n t egral m u l t iple of t h e frcq urucy of 01 1,r


v u t io11 " h ere .

N = nF + s

(S J

i s m a l l compared to R. Tit is melf1'

t h a t t he mo-. ing object w i l l

i n terval of obsPrv a t ion.

V.

using t hl'

execu lc

Hlii;h t l )

v i den t t ha t t h(' pha e of tlw

more t h a n 1 1 cycles of mot ion d u ring t l 1t i n


l e n a l he t " ccn observ a t ions , o t h a t t hr p h a . .

motion m a y be con t rolled a l " i l l merely b)

at " h id1 it is RC'cn blroboscopica l l ) " i l l con

Tt is furt her
ob erve<l p

C) clt of

i L ion of t he ohj!'ct i n i t
phase of t h e

sh ift i n g t b

1 1chrono11

shul lcr

pc ial ca e o f I erfcc L synchron ism,


whi h L bc frequency of mot ion a n d o f

or light fla h w i t h resp

t to the motion.

The
in
ob

t ra n ien t moveme n t

or v i bra t ion t a ke
peed

or

t ppcd

v i bra t i ng mem be r m a d

con picuou

m a ry

haft , or

model a nd a w a i l i ng p u b

a dia t teri n rr v a l v e spring

e n tire!

not

i t her '"ie" ed t hro 1 1 g h a

jl'cl i s

hroni m.

a b on.

for e a m i n a l ion of ma h i n

will

een a

rad i a l l i n e .

w heel

tion i n u n i t t i me.

or

i f de ir c l .
t nni

mu t be mad

cl t a i l

u l t.

bri l l i an t . 0 t h r -

i one radial l i ne i s

i ng for

su re! , a nd a . i n o-le t roke can h


H u man

om

ex cu t i n g u niform l y R

t rok i ng p l a y r w i l l ,

over a m i n u t

and

upp e t h a t tb

ach

It

and, a

l ig h t m u

Stroboscope A rith m etic

i n g ra t e, h i a r m w i l l h e i l l u m i na l ed a
l i u le fa r t her a long i n l he

th

w1

u nd er

low r t h a n hi

f d e t a i l wiU r

t b

t ion. a n d b l urring

flashed o n h i m a t a con L a n t ra te,

exac t l

cl uri ng i l l um i na

t ion w i l l t a ke p l ac

identicu l , is J.. now 11 us " funda

r v a t iou ar

mental"

) n hroni s m .

I f n is g rea t er t h a n
ser ed nnl

I, the olij cl \\ i l l h t > oh


c y rlr of mo t ion, o

u t every 11 t h

t h a t t he in tegratcd i l l u m i n a t ion
l o t b e Cra t i n a l amou n t
i n a t io n a t f1111dam n t a l
A l t hough

any

" time

arrebl

of

If,

on th

reduc<' t l

the ilium

pCT fcc l

w here

otb r hand,

{:.l)

t i mes per cycle of motion a n d w i l l , accord ingl) .

k point. eq u a l l y

spaced, i n t ; m e , t hroughou t t h e c cl
tion.

uch a

"parti a l "

ond i t i n , w h ich

o f mo

known

as

ynchrooi m, w b i l e appare n t l y ar

res ti n g tbc motion of t he obj ect, is not, i11 gcn-

ral, sa t i factory

for v isual

t rob

C ) l i e freq 1u11t') o f

(SaJ

11 F

C) t>Je, i 1 1 u n i t t i ml' a n d Lo tra \ ..I i 1 1 t lu sa nw


dire L ion as l ht oLj rc t i s act u a l l 1110\ i ng .
Co1nrrscl) ,

if

the

cyclic

copic ob

crvation . For exampl , a rota tin disc h a v -

i,

frc 1 ucncy

blih t l ) ll'ss t h a n a n i n tegral m u l t ipl<' o f t lw


freque ncy of obser v a tion t he ph asl' at " h il'h
thr objec t is sc{'n

lo

t rol.Jo8copically \\ i l l 10 1 1 -

t i11 u a l l )- recede s o t h a t t h e obj e c t \\ ill a p pl'ar


mo' e a t a blow r yc l ic freq u nC) in a d i ne

tion opposite LO t he true motion:


.

(01

11 F- R

The olo" bt roboscopir 11101 ion " h i ch

k i s any i n L ral n u m lwr g r a ter t h a n l ,


h poin t of t b ohjcct w i l l b v i i blc /;
ucces i v el y at

o b t a i n e d in t h ib 11i a n n cr, and " hich !'an

be ob crved

rnnd i t io n

nchronism.

F = J..R

t hen

S) n

L iu motion, i L

bv iousl de iraLlc LO worl.. at th

of fundame n t a l

is

y nchronism.

co1,dit ion

chronism will com p l tel

is

t i n u a l l y a d v a nce. The obj ec t w i l l t l w rrf o re a p

pear to m o v e 11 t a slo"

j us t ed t

'""l ) ,. .
be a d

h e o m e a v e r y s m a l l fract ion of t he

true speed, ma l-es the st rohoscopc <''-L rt' rn d )

v a l uable i n w a tching L IH cy<'lc of mot ion of


machi11ery running a l sprJs loo high t o lw
followed w i t h the u n aided cyr.
The freq uency of s t rolJo copi<' mo t ion,

."i.

ma) be made as slow as d ired. On t h l' ot her


band i f is increased above a certa i n l i m i t t l ie
observed

motion

becomes i n tcr rn i t l.ent

a11d

l ess sa tisfactory for purpo e s o f v i u a l b L U d ) .

- Horatio

W.

Lamson
19

J U LY / S E PTE M B E R 1970

www.americanradiohistory.com

PROGRAM MAB L E
D E C A D E RES I STOR
The G R 1 43 5 Programmable Decade
Resistor was de igned for maximum
customer-use flexibility consistent with
accuracy and cost. The basic instru men t
covers the five-decade span from I 0-D
to 1 00-kD per step, with each decade a
plug-in board. Mechanical and electrical
provision has been made to allow simple
conversion to a six- or seven-decade in
stru ment, hould the need arise. Reed
switches used throughout the instru
ment are of the miniature mercury
wetted type, for low and repeatable
zero resistance as well as bounce-free
operation. The high and low terminals
o f t h e r e s i s t ors are isolated from
groun d ; this permits use where a float
ing resistor is required.

d e c a d e resistor terminals remotely.


Grounding pin 1 8 of the 36-pin rear
panel connector horts the resistor ter
minals, while grounding pin 1 7 open
the resi tor terminals. This is particular
ly useful if discontinuities are obj ec
tionable, which may exi t when resis
tance settings are changed ( the worst
case being the change between "7"
and " 8 " where all four reeds change
state). With proper timing the resistor
terminals could be either opened or
horted during this switching interval,
whichever is more pertinent to the ap
plication.
A few typical applications are illus
trated in the figures, ranging from a
simple programmable amplifier load to

Three distinct modes of operation


are provided for the user's convenience :

a programmable oscillator or time con


stant.

Manual Mode The desired resistance is


set on the front-panel dials, just as one
would set a conventional decade resis
tor. This is useful, for example, when
you are making accuracy check on the
G R 1 43 5 , to determine how many de
cades need be remotely con trolled for a
particular application, or when you are
manually checking proper system oper
ation during initial set-up stages.

Cat:alog
N u mber

1 435-9700
1 435-9701

Remote Mode You can select this


mode by turning all decade dials to the
" R " positio n ; by turning the power

switch to the " R EMOTE" position; or


by applying logic O to pin V of the
rear-panel connector. Resistance is set
by application of negative true 1 -2-4-8
BCD signals at standard DTL or TT L
levels or contact closure to ground, to
each decade via the 36-pin rear-panel
connector.
A feature that deserves special men
tion is the ability to short or open the
"

this article.
Complete specifications for the GR 1 435
are in G R Catalog U ; minor revisions are
shown below.

F requency Characteristics: At h igh-resistance


values, frequency
characteristics depend
mai nl y on capacitances and on the type of
con nections used ( 2- or 3-term inal. grounded
or guarded ). At low resi stance values. they
depend mainly on the inductance. Calcula
tions based on values shown should g ive
approximate series-resistance error.
C)

HIGH

Parameter

$750.00
730.00

Bench Model
Rack Model

_L'T' 10 'l' C4
ye Ic C5T
=

Decade Resistance
R = 1 MO
100 kO

19 pF

C2

76 pF
19 pF

16 pF

cs

247 pF

276 pF

C6

1606 p F

1 606 p F

23 H

23 H

51

46 pF

OSCILLATOR

OR

pF

TIME CONSTANT

E
"'
R,

PROGRAMMABLE VOLTAGE SOURCE

R,

23 pF

C3

.v

1 1 pF

'

C4

PROGRAM MABLE

R,
....

Cl

ll

Prices net F O B Concord, M A , U S A .


Su bject to quantity discount.

PROGRAM MABLE CURRENT SOURCE

E-:-1
1 o LOW

-l
--'o<>-T

-:- GROUND

P i e

1 435 Programmable
Decade Resistor

G R 's Component and Network Test

ing Group, who contri b uted the material for

rc

Description

M anual/Remote Mode Some of the de


cades may be set to the " R " position on
the front-panel dials and be remotely
con trolled, while the remaining dials are
set to a particular value of resistance
and held constant. This has the advan
tage of requiring four less control lines
for every decade which is manually set.

This instrument was designed by Peter


Gray of

PROGRA M M ABLE

VOLTAGE

;s

R,C .tn

D I V I D ER

Vz

14' R,.

"

PROGRAMMABLE I NTEGRATOR
Rp

1435
Ei

-1'
:>----<8>>---14351

PROGRAMMABLE

o
E

o
E -

R f

E1dl

20

Vo

AMPLIFIER

RL

GENERAL RArno

www.americanradiohistory.com

LOAD

Experimenter

..

Underside of

S tandard

1 790-9603

Device Adaptor

Kit showing socket h o l es


and typical wire-wrapped
connections.

Uni versal Device Adaptor.

G R 17 9 0 - M O H E VERSATI L I TY A N D CA PAC ITY


V e r atilily i s t he name for t he u ni
versal device adapt r recen t ly made
available to G R 1 7 9 0 custo mers.
The logic-circui t analyze r ha proven
itself facile enough to h a n d le most log
ic-ci rc u i t test problems presented i t i m
ply by means of suitably adapted in ter
facing.
GR engineers incorpora ted several
im porta n t pri n ciples in the design of the
in te r[ace adaptors. Connections w i t h i n
the adaptor are made simply, by wire
w r a p ping to terminals m o u n ted on
p rin ted boards. A uxiliary con t rol o r
moni toring circui t ry a n d loads are easi
ly connected within the adaptors.
ustomers are given the choice o f a
standard adaptor or of a univer al adap
tor. I n the un iversal adaptors, i n pu t and
o u t p u t connect ions can be determined
by t he test program; t h i permi t s accep
tance of a greater variety of devices as
well as provi ding checks of outpu ls a n d
i n pu ts. Use o f the universal a d a p t o r per
m i t s le ts for shorted inpu ts.
The adaptors are easily i nserted in to
or removed from the analyzer by action
of a single lever. Provision is made o n
o n e stan dard board to m o u n t sockets i n
rows 0. 2 5 0 inch a part a n d spaced a t
O. l 2 5-inch i n tern als. Aduptors are avail
able with 2 4 , 48, 7 2 , and 9 6 i n pu ts and
48, 7
96, 1 20, and 1 44 o u t p u ts i n pre
determ ined com binations.
Complete speci fication detail for
t he GR l 790 uda p t ors are availa ble on
t he tear sheet u t t he back of thi i sue.

As customers for t he standard G R


1 79 0 Logic-C i rcu i t A nalyzer become
familiar w i t h its operation we a n tici
pate t he i r needs will grow lo expand its
appl ication t o m o re complex tests. Or, a
need for expanded memory storage will
be evide n t as test programs length e n . It
i even possible t h a t origin ally l i m i ted
fu nding f o r capital ex pen d i t u res may be
i ncreased as t he savings, made po si ble
by the GR l 790 i n a c t i o n , are brough t
to managemen t 's a t t e n tion. For a n y o f
t h e e reasons, G R is p repare d t o h e l p i t s
custome rs e x pa n d t h e i r tan dard a n a
lyzers by supplying a n d instal ling sever
al ret rofit k i ts.
K i t s are available in two ba i c for
mats - one to ex pan d me mory by 5 0
Desc r i p t i o n

Retrofit K its, for i nsta l lation of options i n


t he field b y G R personnel
K i t 2A K for Add i tional Memory . 60-Hz l i ne
Kit 2B K for Add i t i onal Memory , 50- H z l i ne
K i t 3 K f o r Programmable Leve l s , 50 t o 60 H z l i ne

Dedicated
I n puts
O u t pu ts

1 790-960 1
1 790-9602
1 790-9603
1 790-9604

no soc k e t holes
no socket holes
socket holes
soc ket holes

72

Price

72

96
72
96

1 44

72

1 20
96
72
48

72
1 44

1 30 . 00
1 95.00
1 35.00
1 95.00

U n ,iversal Device Adaptars

1 790-9605
1 790-9606
1 790-9607
1 790-9608

n o socket holes
no socket holes
no sock e t holes
no socket holes

48
24

211
48

72
96

600.00
825.00
1 000.00
1 350.00

P r ices net F O B C o n co r d , M A , U S A .
Subject to q u a n t i ty d i sc o u n t .

" C o m p u t e r A i ds R e d u n d a n t Logic
Search , " G. R . Partridge, Electronic De
sign News (EDN}, 1 5 J u ne 1 9 70 . * *

C a l i bra
"Se mi-Automati DC-D V M
tor, ' ' R. P. A n derson, Measure m e n ts '&
Da. M ay u nc 1 9 70. * *

"Planning I nve t m e n t s i n R e earch a n d


Developmen t," W . D . H il l . Managerial
Plann ing. J uly/August 1 9 70 . *

JULY

Progra m m a bl e
I n puts/Outputs

Standard Oevke Adaptor K i t s

"The H u man Factor i n Precise M ea u re


men t s C . E . W h i te, Measurements and
Data,
arch-Apri l I 970. *

$ 1 0,500.00
1 0 ,600.00
1 3 ,500 .. 00
23,000.00
23, 1 00.00

K i t 2A- 3 K for b o t h opt i ons. 60-Hz l i n e


K i t 2 B - 3 K f o r both options, 50-Hz l i ne

Hecent Tech n i cal Articles by G R Perso n n e l

* R e pri n t s available from E d i t o r


* * Repri n ts n o t avai lable.

Complete peci fication det ails for


G R 1 7 90 options a re available on
t 11c tear sheet at t h e back of t h is issue .
t he

1 790 L ogic-Circu i t Analyzer

_,

,"

ti mes ( Op tion 2) a n d the other t o pro


v i de capabil i t y for t h e a d d i t ion of p ro
gra m m a bl e logic levels a n d p rogra m
mable power supplies ( O ption 3 ). Both
opt ions a re availa ble i n 5 0-I I z a s well as
6 0- H z version , to acco m mod a te over
seas cu l mers. O p t i o n s a re also avail
a ble se parately or combined.
either
option req u i res more p h y sical space.
The k i t s w i.11 be i nstalled by G R dis
t rict office service-depart ment person
nel. Tra i n i ng req u i re d for opera t ion
with Option 2 w ill be provi ded the cus
tomer; no fu rther t ra i n i ng is req u i red
for Opt ion 3 .

Experimenter,

"A oise E x posu re t c t c r . " G . R . Pa r


t ri dge, lo be presen ted 3-6 N ovem ber,
Acoust ical Society of A m erica . * *
" A pprox i m a te Tra nsfer C h a rac tcris t i t:s
of a Condenser M icrophone w i t h D i a
p hragm S t re t ched Over R a ised Poi n t s of
the Back p l a t e , " S. V. Dju ric, t o be pre
sen t e d 3-6 N ove m ber, Acoustical Socie
ty of A merica . * *

General R a d i o .

/ S E P TE M B E R 1 970

21

www.americanradiohistory.com

GEN ERAL RADIO

T TS 0 1 742

' CONCO R D . MASSAC H U S

6 1 7 300

4400

SA L E S A N D S E R V I C E
800 62 1 -8 1 05

A L B U Q U E R QU E
ANCHORAGE

505 265 1 097


907 279-574 1

DETROIT
ERIE

Z E N I T H 2203

ATLANTA
BOLTON

G R E ENSBORO
G R OTON

800 638-0833
03 445 8445

BOSTON
B U R BA N K
CH ICAGO
C LEVE LAND

800 638-0833
6 l 7 7 79-5562
6 1 7 646-0550
7 1 4 540-9830
3 1 2 992 0800
800 62 1 -8 1 05

HARTFORD

203 b58-7496

HOUSTON
H UN TSV I L L E
I N D I A NAPO L I S

COCOA BEACH
D A L LAS
D A Y TO N

800 638-0833
2 1 4 3 7 2240
800 62 1 8 1 05

L O N G I SL A N D
* LOS AN G E L E S

7 1 1 4 4-5 1 1 2
800 638-0833
800 62 1 -8 1 05
2 1 2 964 2722
7 1 4 540-9830

( N Y ) 2 1 2 964 2 1 72
(NJ) 201 941 3 1 40
P H I LAD E L P H I A
2 1 5 646 8030

* N EW Y O R K

tZ

P I T TSBU R G H
ROC H E S T E R
SAN D I EGO
SAN F R A N C I SC O
S E ATT L E
SY RACUSE
WASH I NG TON .
BALTIMOR E

N I TH 2203
3 1 5 454 9323
7 1 4 540-9830
4 1 5 948-8233
206 747-9 1 90
3 1 5 4b4-9323

301 88 1 -5333

Vie operator

I NTER NATIONAL DIVISION


'ARGENTINE
and PARAGUAY
Coasin S A.

HONG KONG and MACAU


Gilman

Werbunon Frankl I ndu s tries

Sydney. Tel. 29. 1 l 1 1


Melbourne, Tel 69 0 1 5 1
Brisbane, Tel. 5 1 .5 1 2 1
Adelaide, T el . 56.733 3

' BR AZ I L
Amr>riex S . A.

Rio de Joneiro, Tet .

Si't Pau l o,

&

C o. Ltd.

e.c.c . .

Hong Kong,

Ltd.

Te'I. 2270 1 1

INDIA

Lucknow, Konpur, New Oe1hi,


Bangalore, Mad ras

Tel. 52-7806

JAPAN
Midoriye E l

ec

tric Co . . L t d ,

T o k v o . T e i . 5 6 1 -8851

General A d i o Canada Umi1ec1


Toronto, Tel. ( 4 1 6)
Mon t rea l , Onawa

252-3395

CH I L E

M-C I n ternational

.
s
(
=:i :;.i 6 l

1 5)

39 7 1 4 55

MALAYSIA
Vanourd Company

Coasln Chilo LtQa


San tiago, Tel 39 6 7 1 3

K u a l a Lu m pur ,

COLOMBIA

T o i . 882 1 3

" M E X ICO

Man ue l T r u 1 l l l o

Venegas e H i t o . Ltd.a.
Bogota 2 , D E., Tel 320679

E lectronic

Fredin, S

M e x ico, 1 0

A.

D F . . Toi 2 0 89 4 8

NEW ZEALAND
W & K

Karachi, Toi. 4 7 2 3 1 5

PERU

I mportacionos v

RepresenteciOnes
Ele tronicas S. A.

272076

ff &

P H I L I P P I N ES

J. Wol

Company

M a k a t l , Rizal, Tel. 89 34 86

S I NGAPORE

Vanguard Company
Si ngapore, Tel. 94695

T A I WA N

KOR E A

'CANADA

Pak Land Corporation

L i m a . Tel

Motwane Prl'll ate L i m i ted


Bombay , Tel 252337, C lcutta.

242- 7990

ISA

P A K I STAN

ECUADOR

AUS T R A L I A
Pty

ASSACH U S E T T S 0 1 742

Sumlnistros Tec nicos Ltda


Guavactu1I, Toi. & 1 2 4 1 9

Tel. 52 3 1 8 5

Buenos A ire-s,

CONCO R D .

Heighten Trading Co., L t d .


Tai pe i , Tel . 5 1 8324

TH A I L A N D

G S im on Radio. Co. Ltd


Ba ngk ok , T e i . 3099 1 -3

URUGUAY
Coesin U rugu.ava S . A .
M ontevideo, T I . 9 79 78

V E N E Z U E LA

Coasm C A.
Caracas, Tel 72 96 37

McLean L 1 m l t d

I 58 7 039
Well1ng1on Tel 555 869

Auckland, T

" G E N E R A L R A D I O COMPANY (OVE RSEAS)

P . O . Box 1 24, C H-8034, ZUrich, Switzerland, Tel. 4 7 70 20

AUST R I A

l n g Peter Marche t t i

D1pl

Wien, Tel

5 7 8 2 30

B E L G I UM

Groenpol Blgique S A
Bru xell. Tel 02/25 16 36

DEMOCRATIC R E P U B L I C
O F T H E CONGO
Oesco d1 shul thess

ZUrich

Toi 25 44 a

DENMARK
S E M C O IVS
Glo'itruc.>. Te l

45 7 1 22

EASTE R N E U R O P E
General Radio Companv
( 0'11erseas)
Tel

4 7 7 0 20

ISRAEL

F I N LA N D

Into O / Y
H e l s l n k 1 . Tel.

' F RANCE
General Rado France
P a r is, Tel

023 5 7 7 4 , L y on

General A.id10 GmbH


Murrhen. Tel 4 0 1 8 1 7

H m b u r g , T el 4 5 06 56

G R E EC E
Mario-s Dolleggio

R p r sentanon
s
Athens. Tel 7 1 0 669

I R A '
Berk eh Co
Tehran. T el

L to
6? 8 2 94

Gem ral Radto C o m p a n y

( 0\l'ersea s ) . 1 cl 4 7 7 0 20
Gen eral And10 Company
lU K ) L 1 m 1 hci, Tel 2 2 5 6 7
services are

' ITALY

CJ enera l R a d i o I talia S . p . A .
M i l a n o , Tel. 29 3 1 2 9

LEBANON

ProJectS

G E R M ANY

EIRE

R epair

1 1 123

E a s t ronics L t d
Tel A'll iY. Tul 44 04 66

Beirut, Tel

2 4 1 200

N E T H E R L AN D S
G roenpol lndust:riolo
Verkoop N V.
Ams1crdam, Tel 020/64/474

N O R WAY

Gvsta'll A. Ring A/S


Os lo. Tel 46 68 90

PO R T U G A L
Ce1sa Serras
L1sl>oa

Q)
ci
0
'O
E a.: 0
u
"'
0.

R E P U B L I C OF SOUTH
A F R ICA

,,;

Associ ated E l ectro nic,; ( P t y ) L t d

Johannesburg, T e l

724 5396

SPA I N

H n.pano Elec.tronica S A
T el . 233 . 1 6. 0 1

Madrid,

SWEDEN
F ir m a Johan

Sol na , Tel

Q)

Lagercra n t z K B

......

08/83 0 7 90

SWIT Z E R L A N D
S-cyl fer

Zi.irich,

&

Co A . G .
Te l 2 5 54 1

' U N I T E D K I NGDOM
General Radio Company

( U K ) L11Y11 ted
Bourne E 11 d , Bucki nghar"1'15h 1 r e

Te l 22 5 6 7

YUGOSLAVIA

General Rac;ho Company


(Overseas)
Tel

47 70 20

0
0
<X: .
a:
_J

c::
Q)

Q)
c.
z X

LU

available at these offices.

www.americanradiohistory.com

....

<J)
ffw
<J)
::)

I
u
<{
<J)
<J)

<{

a
a:

0
u
z

0
u

'O
'O

"'

'O
c

"'

E
"'
"'

..,

u
0
u

'-

;"'
..
"'

-
c
"'

..

0.
0
c

0.

,..,
"'

c
0
;:

"'

'O

"'

"

>

.._

"'
!'

'O
0

0.
N

'O
c

..

> E'
0 e
e
'0

Q)

"'

.i::

Q)

0
0

"'

>
c
"'

0.

"'

E x'

0
u

0
.c

.i::

...,

8 B
"'
"

"

..

0.
"'

.,;
"

'O
c

"

i;;

"'

.!!!

"'

a.
"
0
c >
0
":::
o;
.c
0
!!!

"'
"'

'O
'O

"'

"'

c
0

Вам также может понравиться