Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
'` 914,070
~ 3 ~
.
., -- 1 --
DIGITAL FRAME SYNCHRONIZING CIRCUIT
.` ~
This invention relates to digital information
processing, particularly as such information is processed
~or transmission and/or recording such as in magnetic tape
recorders. In particular, it relates to systems and
methods for providing digital signals which are sel~-
clocking and which facilitates serial encoding information
into frames while minimizing bandwidth requirements.
With the advent of digital data communication,
transmission and recording systems, a number of schemes for
encoding data in digital form have been developed. While
early codes were not self-clocking, and therefore required
~ a separate clock or synchronization channel to ensure
reliable decoding, more recent and widely used codes such
as non-return to zero mark (NRZ-M) have been developed in
which a clock or bit sync signal is built into the data
code to enable self-clocking and the elimination of a
separate sync or clock track.
In NRZ-M recording, a transition is provided only
when a digital one occurs, and no transitions are provided
when digital zeros occur. Thus, a series of "l"s or "0"s
will essentially result in a shift in the DC level.
Because such a code has no way-to define a unit or bit
cell, it is not self-clocking, and clock information must
be added on separate tracks, with an attendant waste of
record medium or transmission equipment, as well as limit-
ing the ultimate density of recording due to potential skew
errors. Nevertheless, NRZ recording is the workhorse of
I~
'- -: . . . . ,, '
,
''' ~ , :' . .
~, ~
03
;~ ' "
- 2 -
the recording industry due to the efficient bandwidth
requirements and ready implementation.
Because random sequences of "l"s and "O"s can
result in pulse sequences having long equivalent wave-
lengths, other codes, such as phase modulation (PM) havebeen developed. In PM codes, the bandwidth is reduced to
one octave by providing an output for each bit, whether it
be a one or zero, thus also making the code self-clocking.
Since in PM codes, for example, a "0" may be represented as
a positive transition at the center of the bit cell, a
succession of either "l"s or "O"s may be seen to generate
a frequency fo=l/c, where c is the duration of a unit or
bit cell. Analogously, a successlon of 1-0-1-0 bits may
be seen to generate a frequency of fo/2, i.e., a frequency
having a period equal to twice the cell duration. The
possible generation of two characteristic frequencies has
resulted in this code sometimes being identified as 2F code.
In order to avoid problem with the detection of
the polarity of transitions, the Miller code, otherwise
known as the delay modulation (DM), modified frequency
modulation (MFM) or 3F code has also been developed. See
U.S. Patent No. 3,108,261 (Miller). In that code format,
"l"s are represented as transitions at a particular loca-
tion of the respective bit cells, such as at the center of
bit cells, regardless of the polarity, and "O'is are
represented as the absence of a transition at the particular
location of a cell, and the insertion of a transition at
the beginning of a cell if the preceding cell is also a
- . .
. . . ' ' ~
' ' , ' ~ ~ . . ' - :
. . , , :
- ~ .
~t~ 3~
-- 3 --
zero. Thus, in that system, a succession of "l"s or "O"s
will give rise to a first frequency fl=1/2 c. Similarly,
it may be readily appreciated that a succession of
1-0-1-0 digits results in the generation of a second
frequency f2=fl/2=1/4 c, while a succession of
1-0-0-1-0-0-1 digits results in the generation of a third
frequencY f3=2fl/3=1/3 c-
The three frequencies possible of being generatedthus give rise to the 3F nomenclature. The primary virtue
of the Miller code is that while the bandwidth of the code
is essentially the same as that of the NRZ code, self-
tracking capability is added, albeit at the expense of a
need to generate a 1/2 bit cell time, hence a 2f clock,
and the inability to recover requisite phase information
in order to properly decode the signal back to NRZ upon
playback until a 1-0-1 sequence is received.
In addition to such systems for establishing bit
sync or self-clocking capabilities, it is also desirable
to utilize formats in which incoming data is partitioned
into blocks or frames of data such that error checking
code words, parity and the like may be inserted. Such
schemes likewise require the addition of a unique
succession of bits as a frame sync word to delineate each
frame. Prior art frame sync codes generally require
storage systems in which entire frames are delayed in
temporary memories upon playback and frame synchronizer
circuits "look" at the entire frame to determine the
presence of a particular alternating pattern (see U.S.
3~
-- 4 --
Patent No. 4,002,845). In other frame synchronizer systems,
a long pulse such as provided in Mi]ler Code by a succession
of "O"s has also been suggested, but is undesirable in that
it adds a signi~icant DC component which greatly expands
the bandwidth requirements. Similarly, a high frequency,
such as four or more multiples of a basic clock rate may
also be employed, again at the expense of system complexity
and greater bandwidth.
In the present invention, a frame synchronizing
circuit is provided in which a basic Miller encoding
circuit is modified to provide a frame sync pulse having a
duration equal to three bit cells, thus generating a
fourth, lower frequency, f4=1/6 c. Such fourth frequency
utilizes the available low frequency portion of the
spectrum without requiring any additional high frequency
bandwidths. The resultant frame s~nc pulse cannot result
from any normally allowed transition of "l"s and "O"s and
may, upon playback, be readily detected by means responsive
to the f4 frequency. The frame synchronizing circuit thus
comprises means for generating a Miller coded digital
signal consisting of a 1-0-0-1 sequence of digital bits
and for inhibiting a transition between the 0-0 sequence
thereof, whereupon a signal block having a duration equal
to three bit cells is generated. This signal block has
- 25 associated therewith a fourth frequency which cannot
naturally occur from any sequence of digital "l"s or "O"s.
The circult ~urther comprises means for inserting the
signal block into a formatted digital data stream at a
-- :. - . . . -
~ - ' " '.' ~ ' . ' , ' . , ' . ' :
l~i4~3~
predetermined location. Preferably, such a block would be
inserted at least once for every frame of a predetermined
number of bits formatted data to provide a frame sync
signal which can be subsequently readily and simply detected
and operated on by means responsive to the presence of the
fourth frequency to delineate each frame.
Figure 1 is a block diagram of a preferred
circuit for forming the frame synchronizing signal pursuant
the present invention;
Figure 2 is a set of characteristic signals such
as may be processed through the circuit of Figure l;
Figure 3 is a block diagram of a preferred
; circuit for decoding the frame synchronization signal;
Figure 4 is a set of characteristic signals such
as may be processed through the circuit of Figure 3; and
Figure 5 is a set of characteristic signals such
; as may be processed in an alternative embodiment for form-
; ing a frame synchronization signal pursuant the present
invention.
Figure 1 shows a bloc~ diagram of a preferred
embodiment of the circuit for forming the frame synchron-
ization signal according to the present invention. The
encoding circuit 10 is adapted to receive a digitally
encoded non-return to zero (NRZ) signal on lead 12 at one
input to an exclusive OR gate 14. The other input to the
; gate 14 is controlled by a signal on lead 16 from a record
controller ~nd time generator circuit 18. The circuit 18
is of conventional design and is not discussed in full
'
' . ' ' - : ', ' ' ' :
., - .: ~:
, j . . . .
3~1
\
-- 6 --
herein. The circuit 18 typically includes crystal clock
oscillators, shift registers and the like for generating
appropriate timing signals to convert a continuous stream
of digital bits into a run length limited code in which
the digital bits are partitioned into a succession of
frames, each of which contains a predetermined number of
bits and in which each frame is provided with appropriate
parity check words, error check words and frame synchron-
ization words. When the gate 14 is appropriately strobed
by a fO bit sync signal from the record controller and
time generator 18, the gate allows the NRZ signals on lead
12 to pass therethrough to a D-type flip-flop 20. The
flip-flop 20 is clocked by a clock signal at twice the bit
sync rate (i.e., 2fo) from the record controller and time
generator 18 on lead 22. The output of the flip-flop 20
is coupled on lead 24 to the toggle input of a J-K type
flip-flop 26. The J-K inputs of flip-flop 26 are con-
trolled by a frame inhibit signal on lead 28 from the
record controller and time generator 18, which input
signal occurs once for every frame and thus completes the
formation of the frame sync signals as will be described
hereinafter. A thus encoded signal complete with bit
sync and frame sync information appears at the output of
the flip-flop 26 on lead 30.
The manner in which the circuit of Figure 1
processes incoming NRZ signals is most readily explained
in conjunction with the set of wave forms set forth in
~ Figure 2. As may there be seen, an incoming signal may
' ~, ' '~ . :
: - - ~ - - ~ . . .
~14~1
-- 7 --
comprise a succession o~ digital bits such as a succession
o~ the following digital bits~ 0~1-0-0-1-0-0-1-0
shown in curve A. The NRZ equivalent of such a succession
of bits as set forth in wave form B would thus be provided
at lead 12 of Figure 1. The digital bits within the first
six unit cells shown in wave form A and as NRZ encoded in
wave form B represent actual digital bits of incoming data.
The next four bits comprise a four bit sync signal consist-
ing of digital bits 1-0-0-1. These bits are inserted at
the end of a predetermined number of digital bits comprising
a given frame by conventional circuits~ typically including
shift registers, parallel to serial converters, etc. Thus,
for example, a digital 1-0-0-1 signal may be provided by a
quad two input multiplexer, four inputs of which are hard-
wired to provide a digital 1-0-0-1 sync word pattern. Thus,
when appropriately strobed, input data bits will be
temporarily stored and the sync word 1-0-0-1 outputted in
the appropriate spatial position. The bit sync clock at the
fundamental frequency fO as provided by the record control-
ler and time generator 18 on lead 16 to the exclusive ORgate 14 is shown as wave form C of Figure 2.
By an exclusive OR function in which the bit
clock is combined with the input NRZ signal on lead 12 in
gate 14, the NRZ signal is converted to a biphase or
; 25 Manchester code on lead 15 of Figure 1. Such a biphase code
is shown in wave form D of Figure 2. The input NRZ signal,
constituting high states for the digital "l'ts and low states
for the digltal "0"s are thus oonverted, analogously, in
.
.~ . ' ' ' ~ :' '
: :
,, ~.,.~.:
.
31
,
-- 8 --
the biphase or Manchester code such that digital "l"s are
represented as positive transitions in the center of each
unit cell which the digital "0"s are represented by
negative transitions in the center of each unit cell. Such
a signal may then conveniently be converted to a Miller or
3F code by applying the signal to a divide by two circuit
in a conventional manner. (See, for example, U.S. Patent
No. 4,045,613.) However, a close examination of the biphase
signal of wave form D reveals spikes outputted from the
exclusive OR gate 14 at the beginning of each unit cell
when the existing level of the signal is at a low state.
Such spikes are believed to be caused by inherent timing
errors between the input NRZ signal and the fO clock on
lead 16. Although such errors may be reduced by proper
design, it is believed to be virtually impossible to
eliminate them, and the resulting spikes may be sensed by
, the divide by two biphase/Miller conversion circuit,
` resulting in false output transitions. Preferably, there-
fore, the output of the exclusive OR gate 14 is coupled to
; 20 the D-type flip-flop 20 which is synchronously clocked on
.. ~ .
lead 22 with the bit sync clock on lead 16, but at a rate
2fo as shown in wave form E of Figure 2. Accordingly~ the
` input wave form on lead 15 is effectively sampled slightly
: ,
after each of the transitions, and thus provides a delayed
biphase signal on the output of the flip-flop 20 on lead 24.
Such a delayed biphase signal is shown as wave form F of
Figure 2. Each bit cell is now denoted as being delayed
ln time one-half of the 2F clock period, or equivalently
- . : .. . . .
. . . .. .
... : . . .. .. : .
. . - , - . :
.. .. . . . .. . .
3~
.
g
a fourth of a unit cell. This delayed biphase signal is
coupled to the J-K flip-flop 26 to achieve the divide by
two biphase/Miller conversion function noted above.
The J-K inputs of the flip-flop 26 are controlled
by the frame inhibit signal on lead 28 from the record
controller 18, in such a manner that the J-K inputs are
brought low sufficiently ahead of the transition occurring
in the Miller or 3F code signifying the two sequential "O"s
in the desired sync word. See the circled transition in
wave form G of Figure 2. The transition between the
successive "O"s in the 1-0-0-l sync word is thus inhibited
at the output of the ~-K flip-flop 26. This inhibition is
provided by frame inhibit signal such as shown in wave
; form H on the lead 28, wherein a single pulse occurring
once for every frame thus brings the J-K inputs to a low
state at the appropriate time. With the inputs of the
flip-flop 26 thus restricted, the transition between the
successive "O"s within the four bit sync word is inhibited
and the resultant 4F output on lead 30 as shown in wave
form I results. The transition at the circled portion
of that wave form is absent, thus resulting in a pulse
~
extending three unit cells in duration. This resultant
frame sync signal represents a fourth frequency or time
period which may thus be readily detected as set forth
hereinafter.
A preferred companion circuit 32 for detecting
the frame sync signal is shown in Figure 3. In this
figure, an input 4F signal such as provided after the
,
~ ' - ' ~ ' ' ,' ':
.
)31
-
-- 10 --
encoded signal has been recorded on a suitable record medium,
such as magnetic recording tape, and retrieved via a
conventional magnetic playback head is inputted on lead 34.
The input signal is coupled to a doubler circuit 36 which
includes a monostable multivibrator, so as to provide a
monostable output pulse for each "0" crossing of the input
3F signal. The output of the doubler 36 is in turn coupled
to a 3F or Miller/NRZ decoder circuit 38 on lead 40. The
signal from the doubler 36 is applied on leads 40 to a sync
detector network 42, to a phase detector 44 and to a feed-
back gate 46. A regenerated bit clock signal is further
provided to the 3F/NRZ decoder circuit 38 on lead 48, which
signal is used together with the signal on lead 40 to
convert the 4F signal back to an NRZ output signal.
The output from the monostable multivibrator
within the doubler 36 resets the sync detector 42 upon each
transition signifying a digital bit. The sync detector 42
preferably consists of a five bit counter 50 and an inverter
52. The decoder circuit 32 also includes means for
regenerating a 2fo clock signal, which is coupled to the
five bit counter 50 on lead 54. The manner in which this
signal is regenerated will be discussed hereinafter. As
will be seen in more detail in conjunction with the
discussion of Figure 4, the application of successive
pulses of the 2fo signal on lead 54 to the counter 50 during
the occurrence of a signal on lead 40 corresponding to a
frame sync pulse extending three unit cells in duration
will enable the counter to reach a count of five during
.
.' . . . , . ' ' '" : ,:: ~ '': ' ' '
: . . . :, . , ... : : :.. .-,: .- : ... , : : .. : . : . .: : : . - . .
:.:'' : -:: '' . .. '- ''': . ' ' . ' ' ~', '. ' , -. ' ~
: - .:. .. . ,. . ..... -,: ,. : . " ~ . : . . . .
. . .: :. - . . i ~ .
,., . . . :.: :
~14Q3~
, . . .
-- 11 --
the third NRZ unit cell period. The only time that such
five 2F clock periods could take place during ad~acent
transitions would be during such a sync period; otherwise,
the occurrence of a transition on the lead 40 indicative
of another digital bit would reset the decade counter and
thus would prevent the counter 50 from providing an output
signal. Since an output from the counter 50 can only be
thus provided if no rese~ signal is provided during five
2fo pulses, the peculiarity of the frame sync signal is
detected. The output of the counter 50~ indicative of a
frame sync signal is coupled to the inverter 52 and is
provided as an output frame sync signal on lead 56 to
control peripheral equipment on an output terminal 58, as
well as to provide a frame sync input signal to the bit
sync generator 60, to control the phase of the bit sync in
~` a manner to be described hereinafter.
The basic bit synchronization and clock regenera-
tion portion of the decoder circuit 32 utilizes a phase
lock loop network shown as the blocks including the phase
detector 44 together with a loop amplifier and filter
circuit 62, a voltage controlled oscillator 64 and the
feedback gate 46. The feedback gate is desirable inasmuch
as transitions in the input 4F signal occur at 1, 1-1/2
;::
and 2 bit cell spacings. The gate 46 couples the feedback
signal from the voltage control oscillator 64 to the phase
~ detector 44 on lead 66 only when an lnput pulse from the
; monostable multivibrator on lead 40 is available for phase
comparison. When the appropriate phase is present, the
;
,. . .. ,. , . ,... -,. ........ . .. -, ,. - . :
-
., . , , . . : .
...
31
- 12 -
reset signal is coupled through the phase detector 44 to
the loop amplifier and filter 62. The signal is thus
amplified and filtered to provide loop stability and to
remove undesirable high frequency components. The thus
filtered signal is then coupled to the voltage controlled
oscillator 64 to provide control of the operating frequency.
Since the reset signals on lead 40 from the monostable
multivibrator occur at twice the normal bit rate, the
output from the oscillator 64 thus comprises the 2fo clock
signal which is fed into the five bit counter 50 on lead 54
as discussed hereinabove. Further, the 2fo signal is
coupled to the feedback gate 46 on lead 68 where it is
gated to lead 66 to enable the comparison with the input
signal on lead 40. The 2fo signal on lead 68 from the
voltage controlled oscillator 64 is also coupled to the bit
sync generator 60, which is a divide by two circuit,
thereby providing a bit clock signal at a frequency fO on
the output lead 70. This signal is also coupled to the
3F/NRZ decoder on lead 48 as discussed above. The bit
- 20 clock generator, or divide by two circuit, is desirably a
J-K type flip-flop. As such a flip-flop is not phase
sensitive, the frame sync signal on lead 56 is provided to
this circuit such that the proper phase relationship is
established between the primary data signal on lead 72 and
the reconstructed bit clock signal on lead 70.
The decoder 38 is conventional in design and
- does not constitute a direct part of the present invention.
Such a decoder typically comprises a series of shift
, .: '
., : ' ' ' '
.
' .', ' ' . ' , ,
31
- 13 -
registers and timing circuits so as to perform the inverse
conversion from the 3F signal into a standardized NRZ output
signal on lead 72.
The manner in which the signals are thus processed
in the circuit shown in Figure 3 is more readily appreciated
in view of the wave forms set forth in Figure 4, which
figure is further desirably viewed in con~unction with the
signals as encoded and as shown in Figure 2. In Figure 4~
the same digital bits presented in Figure 2 are shown to be
provided on wave form A as a 4F input (wave form B). The
4F input thus corresponds to wave form I of Figure 2. As
the 4F input signal is processed through the monostable
multivibrator of the doubler 36, an output is provided in
which a transition occurs at each zero crossing, as shown
in wave form C. As that signal is processed through the
: phase lock loop circuit, including the phase detector 44,
loop amplifier and filter 62 and voltage control oscillator
64, a 2fo signal is regenerated on leads 54 and 68 as shown
in wave form ~. The 2fo signal is divided by two within the
bit sync generator 60 to provide the fO signal shown in
wave form E. Slmilarly, when five 2fo pulses on lead 56
are counted by the five bit counter 50 without a reset
signal on lead 40 causing the counter to be reset, a frame
sync signal is provided on lead 56 as shown in wave form F.
The 4f signal as appropriately decoded within the decoder
38 is then provided as an NRZ output on lead 72 as shown
in wave form G.
., . , .,, , ~ : . , ' .
-: : . . ,: , . : .
,. .: :..... . . . . . . . ..
, - , :: . , ~ :
. . .
03~
i
- 14 -
The present invention enables a further
advantageous feature over that provided with conventional
3F or Miller code systems in that it enables the sync
detector to be reset each time a pulse occurs from the
monostable multivibrator within the doubler 36. Thus,
unlike that required with a Miller code where one
necessarily must wait until the occurrence of a 1-0-1 pulse
sequence in order to determine the correct phase of the bit
clock, in the present invention one need look only at a
; 10 single bit, since the phase information is already provided
via the phase detector 44. Further, the need for redundant
circuits to detect the proper phase signal as is necessary
in Miller decoders is eliminated.
In digital recording~ one normally wants to
~ 15 maintain the DC component of the recorded signal, but does
:: not want a long string of successive "O"s or "l"s to shift
the DC level. Thus, for example, in the embodiments of the
present invention set forth above, the frame sync pulse,
extending three bit cells in duration, may be sufficient
to result in an undesirable DC level shift. Accordingly,
~ in a further preferred embodiment shown in Figure 5, an
.(~ .
~ eight bit frame sync signal, as opposed to the four bit : .
.: . .
frame signal shown in Figures 1-4, may be provided. In
such an embodiment, an eight bit succession of digital bits
1-0-0-1-0-0-1-0 may be provided, as shown in wave form A of
, ~,:
Figure 5. The NRZ equivalent signal for such a train of
digital bits is shown in wave form B. As converted in a
manner similar to that shown in Figures 1 and 2, the
. : :
';
,, . . . . . , ...... . . , . . . . , , . ~
: : . . . . .. . . . .
l~t~03~
, .
resultant 3F signal will be as shown in wave form C. The
frame sync signal is then provided by inhibiting both pairs
of 0-0 transitions, such as by providing an eight bit frame
inhibit signal, wherein, as shown in the circled portions
of wave form D, two inhibit pulses are provided. As
further shown in wave form E, these in-hibit pulses result
in a pair of frame sync pulses, each extending three unit
cells in direction and o~ opposite polarity. Any DC shift
as may result from the first three unit cell duration pulse
are thus averaged out by the second but opposite polarity
three unit cell duration pulse.
In conventional Miller code systems, a full
length word is generally provided for frame synchronization
functions. In the embodiments set forth in Figures 1-4,
only four bits are required for the frame sync function,
thus making additional bits availab]e for use in controlling
auxiliary functions. For example, the additional bits mày
be used to signify the tape speed during recording such that
'~ upon playback, timing control signals may be appropriately
~ 20 modified. Alternatively, other functions such as analog
i, .
ranging signals and other clocking or function control
~ signals may be provided on the additional bits without
,. .
requireing further bits to be added to a data frame.
The foregoing description assumes the insertion
of a frame sync word once for each frame. It is similarly
within the scope of the present invention that a frame sync
signal be inserted at other places within the data stream,
such as only providing such a signal once every lO frames
~1~4~3~
- 16 -
or even less often, depending upon the stability of the
given data system.
While the circuit described above is desirably
utilized in a magnetic recording apparatus such as an
analog digitized recorder, the synchronization circuit has
similar applications in a variety of instrumentation
circuits and/or information processing applications.
Accordin~ly, while only a limited embodiment of the present
invention has been shown and described in detail, it will
now be obvious to those skilled in the art that many
modifications and variations which satisfy many or all of
the objects of the invention but which do not depart from
the spirit of the invention as described by the appended
claims will be included within the scope of the present
invention.
. . .
.
. ~ . .
.
... - .
,
. ,