BACKGROUND OF THE INVENTION Storing information magnetically on moving media involves encoding clock and data pulses in a fashion that provides a reliable decoding means. The encoding method generally defines the decoding means.

In three-part encoding, the polarity of the clock pulses are the opposite of the polarity of the data pulses. The scheme makes the decoding process easier than, for example, a data stream encoded using MFM. An additional problem with the three-part encoding scheme is knowing, in the decode mode, which polarity pulse represents a clock bit and which polarity pulse represents a data bit. Once the polarity of pulses has been established, the decode process becomes trivial.

SUMMARY OF THE INVENTION A unique, recognizable, encoded pattern was developed that could easily be identified in a stream of data without an established convention for data pulse polarity and clock pulse polarity. The pattern identifies the polarity convention used for encoding the data to follow.

The pattern contains a known code violation. In the preferred three-part encoding method, the data transition for a given bitcell must occur either in the first half of the bitcell or in the second half of the bitcell. The data pulse may not occur at the 50% or mid point, of the bitcell. If it did, the decoder would not be capable of . reliably decoding the data.

For the beginning of the unique recognizable pattern, a burst of constant frequency is recorded. This being a code violation since pulses occur at the 50% point between adjacent

pulses. After this burst of constant frequency, a significant period of time is allowed to pass before the next three pulses are recorded. This time being preferably chosen to be eight times the period or 16 times the pulse time of the constant frequency used for the previous burst. The last three pulses, also a code violation are recorded at the same burst frequency at the beginning burst.

The data pulse alignment takes place as the first pulse has been read, or decoded, after the long period between pulses is detected. The polarity of this first pulse after the long period can be assigned, by convention, to be, for example, a data pulse. From this point, in the recorded data stream, pulses of that polarity can be decoded as data pulses and those of opposite polarity can be decoded as clock pulses.

At the end of the data stream, a complimentary pattern can be recorded such that, if the reproduce device is approaching the encoded data in a backward direction, the data and clock pulses can be aligned, or assigned the proper polarity, prior to reading real data. At this point, based on the three-part encoding scheme, the data can be properly interpreted.

LIST OF FIGURES The invention may be understood by reference to the accompanying drawings, of which: Fig. 1 is a diagram showing a typical three-part encoded data stream;

Fig. 2 is a diagram showing the preferred encoded data stream for data alignment;

Fig. 3 is a block diagram of a circuit to decode the encoded data alignment pattern.

DETAILED DESCRIPTION OF THE INVENTION Fig. 1 illustrates a typical three-part encoded bit cell where, in this example, the clock bit transition 100 is negative going when read in the forward direction (left to right), and the data bit transition 102 is positive going when read in the forward direction. The bitcell is completed by the start of the next bitcell as the clock for the next bitcell 104 is encoded. Fig. la shows the detail of the data pulse, indicating the nominal time (t) for the time duration between the encoding of the data pulse and encoding the next clock pulse.

The data alignment pattern is illustrated in Fig. 2. The leading burst pattern 200 contains constant time duration pulses. The time duration of these pulses, the time between one transition 202 and the next 204, is comparable to the time duration of the encoded bit as shown in Fig. la. The long time duration period 210 after the constant time duration pulses 200 identifies the pulse stream as being a data alignment stream. The duration of 210 is preferably chosen to be longer than a typical time period in a data stream such that a digital or analog state machine could identify this time duration as being unique. The pulse transition 220 immediately following the long time duration 210 designates the transition polarity of a data bit. In this illustration, when traversing the encoded pattern in the forward direction, from left to right, the polarity of the data transition for the forthcoming data stream is to be positive. This consequently identifies the clock transitions for the forthcoming data stream to be negative. The two transitions 222 and 224 following the data direction designator 220 are provided to realign the encoding

pattern such that the first encoded bit cell 230 can be properly decoded.

Fig. 3 illustrates a typical circuit block diagram to be used for detecting the data alignment pattern and aligning the data stream such that the clock and data transitions are properly decoded. The decoder circuit 300, reference Application Serial No. 206,407 filed June 14, 1988, Fig. 2, produces a pulse 302 to indicate a positive transition has been read and a pulse 304 to indicate a negative transition has been read. The pulses are used by a S-R type flip-flop 310 to create a single pulse stream 312 where by a positive transition pulse sets the flip-flop and a negative transition pulse resets the flip-flop. The timer module 320 measures the time between pulses on line 312 such that when the time of the current pulse exceed, for example, 8 times the pulse time of the previous pulse time, an enabling line 322 is set. After 322 is set, the next transition pulse, positive 302 or negative 304, will latch either flip-flop 332 or 334 respectively. If flip-flop 332 is latched, positive transition pulses will be designated as decoded data bits and conversely, if flip-flop 334 is latched, negative transition pulses will be designated as decoded data bits. AND gates 340, 342, 344, and 346 in combination with OR gates 350 and 352 serve to route the proper polarity pulse lines 302 and 304, to the decoded output clock 360 and decoded output data 362 lines.

While the invention has been described in detail by specific reference to preferred embodiments thereof, it is understood that other variations and modifications may be made without departing from the spirit and scope of the invention.