Note: Descriptions are shown in the official language in which they were submitted.
lV'~
Cross R~ference to Rel~ted ApplicatiGns
" .
The prescnt invention relates to an automated cod~
recognition system including the coded label, a type of font
for providing the coded label and the method of analyzing
the code, and, in particular, with respect to a binary or
,~ two level code employing, for example, background and appro-
priate indicia. Binary codes of this type may be used for
, personal or merchandise identification through optical and/or
', magnetic techni~ues.
' 10 The subject matter of this application is related
to applicant's copending application Serial No. 214,673
, filed November 26, 1974 which, in turn, corresponds to
applicant's U. S. Patent No. 3,891,831 issued June 24,
, 1975,
~, Semi-automated and fully automated sensing systems
, for code recognition purposes are quickly becoming a more
attractive consideration for point-of-scale operations,
; such as a chec~.-out counter application in retail stores
and supermarkets. They provide quick and accurate data
access for such items as : merchandise identification;
~^ merch,andise price, and; credit identification.
It has been found that it would be desirable for
i such systems to be versatile to accommodate small as well as
; large operations and thus to lend themselves to hand-held
,3 operator actuated scanners, in addition to the larger and
more sophisticated fixed scanner stations, by which merchan-
dlse is normally transported. In addition, the system
should be flexible to also accommodate the reading of coded
`~ labels generated with hand-operated printing devices that
may be used at retail outlets, as opposed to printed labels
that are derived from sophisticated printing machines nor-
~, mall~ having minimal tolerances in print variations. Even
cb~
;: ,.;. . . . : .. ~
1~'4()314
the latter occaslonal~y generate irregulari~ies caused by
problems lncurred ln cacll, thc plate making and printing
process.
It should be readily evident, however, that the use
of hand-operated instruments introduces problems which can
affect the overall efficiency of such systems. For example,
a hand-operated scanner in being moved over a coded ]abel
would be subject to speed variations, acceleration variations,
and/or angular velocity components in directions normal to
the code transverse direction. The hand-operated printing
; mechanism, on the other hand, would introduce variations in
!' print that might play havoc with the print tolerances for
~ which such a system is specified. These variations in print
telerance might also occur as a consequence of the wide
variety of coded media to which such a system should lend
r~
itself. Such coded media mi8ht include credit cards, labels,
17 tic-ets, packagee, etc.
:' '
,
-- 2 --
~vb/~k
1()40314
In lts method aspect, the invention relates to a
method of reading on a record medium a two level code repre-
senting at least one character of a set of characters, each
of the characters bein~ depicted as a group of substantially
parallel bars with spaces therebetween, collectively positioned
on a surface adapted to be informationally scanned along the
surface in a generally linear direction generally perpendicular
to all of the bars, wherein no one character has more than seven
,~, modules of information represented by the code and each character
. 10 is arranged in a configuration so that two consecutive modules
of a first level are not immediately followed by two consecutive
modules of a second level, the code for each character defined
j by at least 8iX consecutive transitional occurrences between
the two levels of the code, the method comprising the steps of:
scannlng the record medium to derive a time based electrical
signal denoting the transitional occurrences; measuring from
the electrical signal at least the first four periods between
alternate ones of the transitional occurrences; comparing the
mea~urements of each two overlapping periods to generate three
ratios, each ratio representing not more than one of three
~ possible values, the values being approximately one, two-thirds
I
and three-halves, and; decoding the values arrived at for the
three ratios to tefine a character. .
In accordance with one preferred embodiment of the
present invention, there is shown a code reading system
~ . .:
: . , .: '
,,,~
i~': ' , . '
~' .
: ' ;'
. 3
: Jvb/Jk
1040314
for r~ading a binary or two l~vel code by sensing the white
to black (W-B) and blac~. to white (B-W) transitions and
se~arately measuring the periods between successive W-B
transitions and ~)etween successive B-W transitions. ~s
the measurements of two overlapping periods are derived,
one of a series of comparisons of the periods of overlapping
W-B and B-W intervals is effected to generate, in ratio form,
intelligence comprising several of such comparisons which are
decoded to identify the binary code. Because comparisons are
computed over two overlapping time periods, each one defined
by alternate transitions, the large accelerations observed
~ith hand-operated scanners and considerable point variations
; can be tolerated. This is further enhanced by the simple
choice of the criteria employed which are: nearly one, smaller
't than one, and; greater than one.
The use of these three simple criteria is to some
de~ree based upon the code configuration to be scanned. By
a proper configuration, the criteria are easily arrived at.
~ In the preferred embodiment, the arrangement of the two level
20 code, each character in a set of characters, comprises of
three bars of varying widths and spacings where each width
and spacing is a multiple of a basic module. ~o bar or
space comprises of more than two modules and no two modules
of a first level is followed by two modules of the second level.
Further, because of the comparison made between
overlapping periods defined by two sets of alternate transi-
- tions, less criticality is placed on tight print tolerance
t~ afford the use of less expensive hand-operated devices to
accommodate substantially less than perfect imprints. This
lies in the fact that the overlapping periods will have been
e~ually increased by the locality, resulting into spread
pxint variations to generally maintain a similar ratio.
., .
~ _ 4 -
1~)4~3~4
Brief Description of Drawings
. Figure 1 is an enlarged view of a record medium carry-
ing the three ~ar code or representing a font of type used in
the present invention.
Figure 2 is a waveform diagram representing a char-
~cter nine to show the W-B and B-~7 transitions and the defined
periods Tl through T4.
Figure 3 is a block diagram illustrating a general
overview of the code reading system of the present invention. ~
Figure 4 is a diagram, in block form, of the circuitry
; used for th.e scanner processor 14 and signal ratio detector
15 shown in Figure 3.
Figure 5 is a diagram, generally in block form,
of the circùitry ùsed for the comparators 35, 36 and 37 in
i Figure 4.
. Figure 6 is a diagram, in block form, of the cir-
cuitr~ used for the signal decoder 16 shown in Figure 3.
. Figure 7 is a diagram, in block form, of the cir-
cuitry used for the control logic 18 shown in Figures 3 and 6.
Figure 8 is a timing chart illustrating the wave-
forms of selected signals generated throughout the character
analyzing system and also illustrating selected states of
the analog switch units and sample and hold units depicted
~n Figure 4.
Description of the Preferred Embodiment
With reference to the drawings, there is shown in
Figure 1 a three bar code positioned on a record medium 12
for the digits zero through nine. Also shown, is a two bar
coded character for start/stop commands. As illustrated, no
bar is wider than two modules or bits, and no space between
two bars in a character code is wider than two modules. It
is also noted, that in each encoded character, no two con-
cb/' ~ S
- ~40314
secutive modulcs of a first level (e.g. black) are followed
b~ two consecu~ive modules of a second level (e.g. white).
~t should also be unders~ood the three bar coded depicted
Could comprise a font of type from which the code is printed.
A start/stop sy~bol represented by two bars is also shown.
In the waveform diagram depicted in Figure 2, which
represents a scanned character nine, the three bars are rep-
resented by the black levels detected by a scanner and the
two spaces in between are represented by the white levels.
ThiS configuration in binary form can be represented by 1101101.
Four measurements are defined and performed during a scan
operation on a character which are based upon the W-B and
B-~ transitions detected. These comprise deriving the periods:
Tl extending from the first to second leading edges or W-B
transitions; T2 from the first to second trailing edges or
B-W transitions; T3 from the second to third leading edges
or W-B transitions, and; T4 from the second to third trailing
edges or B-W transitions.
In the decoding operation, the ratio of two successive
but overlapping periods are used to derive three ratios i.e.
R21 T2/Tl; R32 = T3/T2~ and; R4/3 = T4/T3. It should be
understood, of course, that the inverse of these ratios ~;
could be selected as criteria, if desired. With the particular
three bar code depicted in Figure 1, these ratios can readily
take one of three values: nearly one (N); smaller than one~S),
and; larger than one ~). A chart is shown below where for
each character of the set shown in Figure 1, there is shown:
a corresponding code pattern in ones and zeros in a second
column; period yalues derived in a third column (a time
unit one being equal to each ~odule unit transversed), and;
yalue criteria for each of the three ratios for each character
in a fourth column,
cb~ - 6 -
1'~40314
Character Code Pattern Tl T2 T3 T4 R21 R32 R43
0 100101 3 3 2 2 N S N
1 1001011 3 3 2 3 N S L
2 101011 2 2 2 3 N N L
3 101101 2 3 3 2 L N S
4 1011011 2 3 3 3 L N N
1101011 3 2 2 3 S N L
6 101001 2 2 3 3 N L N
7 1101001 3 2 3 3 S L N ,
8 110101 3 2 2 2 S N N
9 1101101 3 3 3 2 N N S
With reference to Figure 3, there is shown a block diagram
generally describing the code recognition system of the present
invention being used with a hand-operated scanner device 11
~or analyzing a coded record 12. The coded record contains a
selected combination of coded characters, from the code depict-
ed in Figure 1, for identi'fying the merchandise 13 to which
the record is affixed. Such a scanner 11 for discerning a two
level ~lack/white~ code, are well known in the art and need
not be described. It should be noted, however, that a scanner
~ensitive to a two level magn~tic code could also be adapted
,f,or use with the present invention.
The black and white two level signal, in the pr,esent
embodiment, is translated into a two level electrical signal
in the scannQr 11 by a suitable photodetector ~not shown) and
then fed to a signal processor 14 where electrical signals
indicative of transitions between the two levels are generated.
Th,e output of signal processor 14 is then operated on by signal
~tio detector circuitry 15 for measuring the periods Tl
through T4, then applying a series of ratio tests and then
deriving ~or each such tests, one of three values in terms of:
~early one (N); smallex than one (S), or; larger than one (L).'
cb~' 7
. ~ ,
~ 403~4
The latter values for each test are routed to a signal decoder
16 and upon values detected for each of three successive tests,
the decoded signal is fed to a digital output 17 where the
digit is displayed or stored. A control circuit 18, connected
~rom the ratio detector circuitry 15, controls the operation
of si~nal decoder 16 and digital output unit 17.
In turning to Figure 4, the signal processor is
shown to include an amplifier 21 connected from the scanner 11
and the output of which is differentiated by a dif~erentiator
22 to generate bipolar signals denoting the times of occurrence
of the white to black and black to white transitions. These
bipolar differentiated signals are fed to comparators 23 and
24 for comparison with threshold values +V and -V respectively,
to generate t~o sets of uniform pulses. `By using such an
arrangement, the scanning system is made independent of absolute
reflectivities of black and white so long as a suitable contrast
of the different colored bars is maintained. Accordingly,
it is readily evident that color combinations other than black
and white may be used.
A first set of positive pulses is emitted from com-
parator 23, denoting white to black ~W-B) transitions, and a
second set of positive pulses is emitted from comparator 24,
denoting black to white ~B-W) transitions.
Transition integrators 25 and 26 are connected to
a common fixed reference voltage V REF. These integrators
25, 26, generate a constant slope ramp and are respectively
reset b~ the trailing edge of signals W-B and B-W. The
outputs of integrator 25 are fed to a suitable analog switch
2~ and a conventional sample and hold unit 28 and the output
of integrator 26 is fed to a similar analog switch 29 and
~similar sample and hold unit 31. Analog switch 2l and sample
~nd hold unit 28 are triggered by theleading edge of the W-B
c~ - 8 -
. ,:: . ., , - . : .
1~4~14
signal to re~pectively turn off analog switch 27 and hold
in sample and hold unit 28, the voltage representation of the
integrated time value receiy~d froM integrator 25. Similarly,
~ the leading edge of the B-W signal pulse turns off analog
:~ SWitCh 29 and holds in sample and hold unit 31, the voltage
~epresentation of the integrated time value 29 received from
. integrator 26. By reason of the NAND gate Flip-Flop 30, it
~ill be clear that when one switching unit is off, the other
is on. Similarly, when one sample and hold unit is in a
sample condition, the other will be in a hold condition.
I The analog switches 27 and 29 are connected in common
$ to three amplifiers including, a first amplifier 32 having a
predetermined gain delta (~), a difference amplifier 33 and
: a summing amplifier 34. The output of amplifier 32 is also
j fed to each, the difference amplifier 33 and summing ampli-
; fier 34, so that these amplif1ers will re~pectively provide
gain Yalues of one minus delta ~1 -d~) and one plus delta
~1 + y), assuming the gain characteristic of each of the
amplifiers is one. Thus, whichever of the analog switches
27, 29 is open to yass a signal from one of the integrators,
the signal passed is acted upon by the difference and summing
a~plifiers to generate two output signals, each respectively
¦ decreased or increased by a value y . Together these signals
~epresent an output having a tolerance of 1 + ~.
The lower level signal from the difference amplifier
~ 33 is applied to the LL input of each of three comparators 35,
l~ 36 and 37. Similarly, the upper level output of summing ampli-
~ fier 34 is applied to the UL input of each of the three com-
¦ parator units 35, 36 and 37. A third input to each of the ~
comparators 35, 36 and 3~ is connected from the sample and .-
hold units 28 and 31. Three different lead paths are used for
this latter connection and in each is included a gain amplifier
'''
C`~)~ 9
~.,. .,, ~ ,. ~ . . . , - -
1~40~14
of predetermincd value based upon the particular code configura-
tion. With the con~iguration depicted in Figure 1, values
corresponding to one, two thirds, and three-halves are employed
to assist in determining which values each of the three r~tios
has, for eventually decoding the binary coded information.
Amplifier 38, having a gain factor of one, is connected to
co~parator 35. ~mplifier 39, having a gain of two-thirds,
is connected to comparator 3G. Amplifier 41, having a gain
of three-halves, is connected to comparator 37. ~ typical
example of a comparator unit 35 through 37 will be described
with reference to Figure 5, wherein comparator 35 is shown
to co~prise a pair of analog comparators 42 and 43 connected
in common to the data input of flip-flop 44. In operation,
when the summing amplifier 34 output to the positive input of
amplifier 42 is more positive than the amplifier 38 output, then
plus 5V output will be effected. Similarly, when the ampli-
fier 38 output to the positive input of amplifier 43 i5 more
positive than the difference amplifier 33 output to tne minus
output of amplifier 43, then a plus 5V output will be effected.
When both amplifiers exhibit a plus 5V output, they effectively
act like an AND gate to enable flip-flop 44 to generate an
~NSIPE signal denoting a nearly one ~N) ratio. Conversely,
when amplifier 38 output is not within the bounds created by
the output of amplifiers 33 and 34, i.e. amplifier 38 output
~s ~ore positive than amplifier 34 ~1 + ~), or more negative
th~n amplifier 33 (1 ~y ), the level at flip-flop 44 D input
will be zero volts, enabling a negative response ~NSID~)
from flip-flop 44.
With reference back to Figure 4, an immediately
previous period T held in one of the sample and hold units 28,
31, is operated on by each of the amplifiers 38, 39 and 41 to
respectively modifv the held sign~l with each of the three
c~/ - 10 - :
4V314
multiplication factors shown. The modified signal will only
cause one of the comparators 35, 36 and 37 to indicate that
the modified signal is ncarly the same as that from one of
the an~log units 27 and 29, and an output will only be gener-
ated from that comparator. If the held information is equal
to one, then onl~ the signal path including amplifier 38 will
be within the bounds determined by amplifiers 33 and 34.
This will enable comparator 35 to produce an inside output
denoting the detected ratio as nearly one. On the other hand,
if the information from one of the sample and hold units is
less than one, amplifier 39 will operate on tile signal so
that it will lie between the LL and UL levels of comparator
36, to enable comparator 36, denoting the detected ratio is
less than one. If the information fr~m one of the sample
and hold units is greater than one, the signal will be operated
on by the amplifier 41 so that the resulting siqnal will lie
between the LL ànd UL leyels of comparator 37, to enable com-
parator 37, denoting the detected ratio is greater than one.
As will be apparent, the ratio formed by the signal
at either of analog switches 27, 29 with the immediately previous
$ignal from either of sample and hold units 28, 31 to be detect-
ed will only cause a ~alue of nearly one, greater than one or
less than one. This, in effect, only allows one of the com-
parators 35, 36 or 37 to be enabled with each ratio d~termination,
As will be discussed in greater detail hereinafter, a strobe
$ignal occurring at t~le time of the two level transitions
~our through six are used to sample all the comparators at
~ach of three times to generate the values detected for R21,
R32, and R43-
3~ The signals from the comparators 35, 36 and 37 are
then fed to the signal decoder 16 which is described in detail
in Figuxe 6, where there is generally shown a shift register
cb,'
ld~E4~314
45, a character decoder 55 and storage register 57. The
shift register 45 comprises a series of memory units 46
througll 54 with three vertlcal columns of three units, each
whereb~ the three units 46, 47 and 48 of the first column
are positioned in parallel to respectively receive and store
the values from the output of comparators 35, 36 and 37
representing the decision made as to one of the three ratio
tests R2l, R32 and R43. For example, if thc ratio test R2l
.- indicated a nearly one decision, tl~en memory units 4~ would
be enabled by comparator 35 to denote the nearly one (N)
output, whereas memory units 43 and 44 would maintain their
zero outputs.
Immediately prior to a decision made on the second
ratio tests R32, the information on memory units 46, 47 and 48
is respectively shifted to memory units 49, 50 and 51. In a
. similar manner, after a decision has been made as to the ratio
-. test R32 and immediately previous to making the ratio test
, information from memory units 49, 50 and 51 is respect-
' ively sh.ifted in parallel to memory units 52, 53 and 54,
whereas the information in memory units 46, 47 and 48 is
, similarly shited to memory units 49, 50 and 51. Thus at
the ena of the three ratio tests per character, shift register
45 provides an indication of each of these three ratio test
decisions.
As is shown, the character decoder 55 is connected
to the shift register 45 so that the outputs of memory units
46, 49 and 52, which are coupled from the nearly one (N)
comparator 35, are connected to a first input of character
decoder 55. Similarly, the outputs of memory units 47, 50 -
and 53, which are coupled from the smaller than one (S) com-
parator 36 are connected to a second input of character decoder
55 and the outputs o~ memory units 48, 51 and 54, which are
c~ ~ 12 -
1~4()~14
coupled from t~e larger than one (L) comparator 37 are
connected to a third input of the character decoder 55.
The ratio Yalue information stored in memory units 46 through
54 is e~peditiously shifted out of the shift register into the
character decoder 45 subsequent to occurrence of the third
strobe signal of sixth transitional occurrence, allowing each
of the memory units 46 through 54 to be reset for processing
of the next character to be read. The character decoder 55
will provide, in a manner well known in the art, a binary
coded decimal digital output which upon occurrence of a decoder
impulse from control logic 18, will be transferred on leads
56 to a storage register 57.
Control logic 18 controls the system operations to:
speci,fy a left-to-right and right-to-left scan; decoder;
strobe~ and denote end of scan. This is expeditiousl~ accom-
plished by use of a control logic circuit such as is illustrated
in Figure 7, where a suitable transition counter 61 is connected
from OR gate 62 with B-W and W-B pulses and also from NOR
gate 63 with a stop/start signal denoting the beginning/ending
of the reading of a selected group of characters on a label
12 by way of flip-flop 64 and 65 which are respectively connect-
ed to leads within comparators 36 and 37 such as that denoted
~t 40 in Figure 5. In the present invention, a left-to-right
scan across the dual two bar start/stop code shown in Figure
1 would signal the commencement and then ending of reading
a character in a forward direction by virtue of a two-thirds
ratio signal received at flip-flop 64. As may be apparent,
~ right-to-left scan would signal tne commencement and ending
of reading a character backward by virtue of a three-halves
ratio signal received at flip-flop 65. As is noted, the
zero output of each of the flip-flops is connected back to the
D lnput of the other so that when one is enabled, the other is
cb/ - 13 -
314
inhibited.
When neith~r of flip-flops 64, ~5 are enabl~d, their
one out~uts ~hrou~h ~iOR gate 63 wi]l m~int~in transition counter
61 as well ~s a character counter 66 in a cleared condition.
As soon as either a start left signal (two-thirds) enables
flip-flop 6~ or a start right signal (threc-halves) enables
flip-flop 65, transition counter is clocked to COUIlt ~-w and
W-B pulses by way of OR gate 62. In addition, thc direction
of the scan is communicated to the character decoder 55 so
th,~t reciprocal of the ratio R34, R23 and R12 can be obtained
for decoding using known techniques.
Each time transition counter 61 reaches a count of
five to indicate six transitions has occurred, a carry pulse
ena,bles the character counter 66 to denote the completion
of scanni,ng a complete character. The latter carry pulse is
~lso used for effecting acceptance of the decoded character
from the character decoder 55 by the storage register 57, by
~ay of output lines 56.
Counter 61 outputs are connected to decoder 67 so
that upon occurrence of the fourth, fifth and sixth transition,
signals will be emitted by decoder 67 which is a typical
gating'circuit. The signals denoting the fourth, fifth and
si~th transitions are routed through the OR gate 68 and con-
~titute the strobe pulses. The fourth transition pulse from
decoder 67 which is also connected from the output of character
counter 66 is also supplied to one input of each of the AND
gates 69 and 70. The input signals to flip-flops 64 and
65 denoting the S and L ratios are also respectively connected
to AND gates ~9 and 70. In effect, this allows either AND
gate 69 or 70 to be enabled at the end of a count of a select
number of characters, so that upon detection of the stop bar
code upon the fourth transition count, the proper AND gate 69
cb/ - 14 -
... . .. . . . . . .
4~31~
or 70 is enabled to supply an enabling signal to either AND
gate 71 or 72 and simultaneously clear flip-flops 64 and 65.
Upon the presence of this latter enabling signal at either
~D gate 71 or 72, should this match the originally enabled
flip-flop 64 or 65, the start and stop bar code will be
matched to enable one of the ~D gates to indicate ~y way
of OR gate 73 the end of the valid character label has been
read.
Operation
Operation of the invention may be best explained
with reference to the waveform diagrams illustrated in Figure
8. Assuming that the scanner 11 is crossing a tri-bar code
representing the character one, an inverted electrical signal
output representing a white to black signal from the scanner
11 is shown at waveform ~a) in Figure 8. From the differentia-
tor 22 are derived, as is depicted at waveform (b) in Figure 8,
positive di~ferential spikes during the black to white transi-
tio~s which are operated on in threshold comparator devices 23
and 24 to generate W-B pulses and B-W pulses shown at waveforms
~c~ and (d) in Figure 8 respectively.
As may be seen at waveforms (e) and (f) of Figure 8,
positive transition integrator 25 and negative transition
integrator 26 will cause ramp pulses to be generated and which
are respectively reset by the ~-B and B-W pulses fed to the
~eset inputs of the integrators 25 and 26. The W-B pulse
triggers sample and hold unit 28 to cause storage of the peak
of the ramp voltage generated during the period Tl. At the
same time, analog switch 29 is then put in an on condition
during the latter position of period T2 (subsequent to Tl)
to directly pass the ramp generated signal from integrator
26 to amplifiers 32, 33 and 34, Assuming that the pea~ level
reached by the ramp signal during the transition period T2
cb/ - 15 -
? - . . ` - .~- ,
'~' ' .' '' " , '', ,' , '' ' . ' ,,
lV~ 3~
is "X" as illustrated at wave~orm (f) in Figure 8, difference
amplifier 33 will gen~rate a signal level x minus delta and
sum~ing amplifier 34 will generate a siynal level X plus delta ~-
and respectively supply these signals to the inputs LI. and UL
of comparators 35, 36 and 37 to define "windows" having toler-
ance levels of X + delta. At the same time, the stored peak
r~mp signal from period Tl in sample and hold unit 28 is fed
to the inputs of ~mplifiers 38, 39 and 41, whicll act to vary
the magnitudes of their respective outputs according to pre-
10 selected gain values at the respective ratios to the input of
comparatorS 35, 36 and 37. 7
Since the periods Tl and T2 are of equal durations
in this instance, the ramp signals generated by the integrators
25 and 26 Will be approximately the same. By providing the
plus and minus delta tolerance levels in each of the comparators
35 ! 36 and 37 to account for lnsignificant variances, it will
be seen that indic~tion of nearly one (N) will b~ arrived at
in co~parator 35 to generate an inside signal, whereby the
signal levels emitted from amplifiers 39 and 41 will lie
20 outside the X + delta window causing only and INSIDE output.
These outputs from comparator 35, 36 and 37 are
then transferred to memory units 46, 47 and 48 of shift
register 45, upon occurrence of the leading edge of the first
strobe pulse (or fourth transition pulse per character) as
sho~n at waveform (b~ in Figure 8. The strobe signal is
applied to comparators 35, 36 and 37 to allow each of the com-
parators to make its decision at the moment when the peak value
of the signal defining the window at inputs LL and UL has
been reached. Clearly, during each strobe operation, only
30 pne o~ the comparators will generate an INSIDE output. Upon
occurrence of the trailing edge of the strobe pulse, switch
29 ~s turned of~ and switch 27 is turned on. At the same
c~ - 16 -
1'~4~
time, sample and hold unit 28 will now be in a sample state
and sample and hold unit 31 is put in a hold condition to main-
tain a peak value representative of period T2. At the peak
ramp signal level Y arrived at during period T3, diffqrence
ampllfier 33 will produce a Y minus delta level signal and
summing amplifier 34 will produce a Y plus delta level signal.
The Y + delta window set in each of comparators 35, 36 and 37
is compared with the immediately previous stored signal in
sample and hold unit 31 having the peak voltage level X arrived
at during the period T2. It is readily evident that the signa]
X in the hold unit 31 from period T2 is roughly fifty percent
greater in magnitude than the peak signal Y of period T3.
The two-thirds gain factor amplifier 39 will modify the
j signal level X from hold unit 31 to produce at comparator 36
a signal which lies between the level Y + delta at the I.L and
UL window. Comparator 36 will then, upon being sampled by the
strobe signal, provide an output indicating a ratio of two-
thirds or less than one, as a decision for R32. Immediately
prior to dumping this information into the shift r~gister 45
at the trailing edge of the strobe signal, the information
in memory units 40 through 48 is transferred to memory units
49 through 51 at the leading edge of the strobe signal.
In a similar manner, at the end of period T4 the
output of integrator 26 will be a level X which is passed via
analog switch 29 to apply to the UL and LL inputs of compara-
tors 35, 36 and 37 a signal X + delta. This signal is compared
with the immediately previous stored peak signal level Y from
the period T3 in sample and hold unit 28. Signal level Y has ~-
a magnitude of approximately fifty percent less than'the level
X, Thus, amplIfier 41 has such a gain factor ~three~halves)
to provide an output between LL and UL units to cause com-
par~tor 37 to emit an INSIDE signal during strobe which is
cb/ - 17 -
.
4~
indicative of the determination of the ratio three-to-two or
less than onc (L).
These three successive decisions: nearly one (N),
smaller than one (S~ and larger than one (L) will be respectively
stored in the memory units 52, 50 and 48 of the sllift register
45. Subsequent to the occurrence of each sixth transition
signal for each chara~ter transition counter ~1 in Figure 7
will emit a decode output, denoted at waveform in Figure 8,
to enable storage reyister 57 to receive a binary coded decimal
digital output on leads 56 from the character decoder 55, which
output is indicative of the character one.
It will be observed that the criteria of nearly
one ~), smaller than one (S) and larger than one (L) is con-
veniently arrived at through the signal ratio detector cir-
cuitry by using ratio values of one, two-thirds and three-halves.
The particular configuration of the tri-bar code designed for
the characters zero through nine, using a prescribed module
width, clearly lends itself to the use of such ratio values.
Specifically, it is significant to note that no code pattern
çontains two modules or positions of a first binary level
~hiCh are followed by two positions of a second binary level.
It is again emphasized that the particular choice
of modular widths and relations, as well as the specific
ratios adopted, allow clear-cut determinations to be made
and simplifies the decision process by having a small number
of values for each ratio and by providing for sufficient
27 latitude in period variances in the comparing stages.
c~ 18 -
