GB2137457A - Recording and reconstructing digital data - Google Patents

Abstract

A digital encoded signal (T) is filtered by a filter (11 Fig. 2a) to generate a series of pulses (V) from edges of said encoded signal. The pulses are recorded on a magnetic record carrier. The reproduced signal (W) has relatively low zero level shift making possible its reconstruction by a simple Schmitt trigger (14 Fig. 2a). <IMAGE>

Description

SPECIFICATION Recording and Reconstructing Digital Data The present invention relates to methods of recording and of reconstructing digital data and to electrical circuits operating according to the methods.

In the field of computers, particularly small computers for home or school use, means are commonly provided for recording digital programs from the computer RAM onto magnetic tape, often making use of a conventional inexpensive cassette recording machine. To this end, several methods of encoding a digital signal are known. A basic method is the so-called "tone burst" method in which the presence of an oscillatory signal indicates a "1" and the absence of such a signal indicates a "O". Such a method cannot achieve a very high density of data storage.

A second method (Kansas City Standard) utilizes a low frequency to represent a "0" and a high frequency to represent a "1". According to the version employed, from one to eight complete cycles of the frequency in question is or are recorded in each case.

A more sophisticated method is phase encoding in which a succession of "1 "s or "O"s are recorded with the direction of phase shift, orshiftfrom a high level to a low level ofthesignal, indicating the presence of a "1" ora "0". Thus, a positive travelling edge may represent at "1" and a negative travelling edge may represent a "0".

However, the time of the edges is all important since otherwise it would be impossible to transmit, for example, two "0"s in succession. To achieve this, a positive-going edge is interposed between the two negative-going edges and this positive-going edge is ignored by virtue of its time position in the signal. Thus, the maximum recordable density is limited by the requirement that these intermediate edges must be capable of being recorded between any two edges which are part of the digital encoding.

A further method which has become known is the so-called Sapporo City Standard. In this method, half wave edge width encoding is employed. A relatively wide pulse of positive or negative polarity represents a "0", and a relatively narrow pulse of positive or negative polarity represents a "1". However, this method is limited by the fact that it requires a definite pulse edge to define the end of any particular binary digit.

Furthermore, with this method the length oftime required to record and reconstruct a given number of data bits depends upon the data pattern. A succession of "0"s will require a longer time and slow down the transmission rate. Therefore, one cannot precisely predict the time required for a block of data to be transferred even if the exact data length in terms of bytes is known.

A method which avoids this latter disadvantage is the polarity-shift encoding method. In this method, an edge is inserted only where the data changes from a "1 " to a "0" or from a "0" to a "1 " and this non-return to zero (NRZ) signal is directly recorded. If a succession of "1 "s or a succession of "0"s are to be recorded, the pulse magnitude remains constant and the number of "l"s or "0"s is determined by a timing circuit. This method allows a much greater density of data to be stored on the tape. However, up till now its practical implementation has presented problems.

In one known method of implementing polarity-shift encoding, or non-return to zero encoding, to record a "1" the tape is magnetized in a first direction forthe entire duration ofthe"1", and the tape is magnetized in the other direction for the entire duration of a "0". Although this produces a satisfactory output from the play-back head of a conventional cassette recording machine, the output from the amplifier of such a recorder is unfortunately so irregular that it must be passed through a bandpass filter in orderto approximate the signal to the output from the play-back head. The resulting signal is not however entirely regular and may have a shifted "0" level which renders its decoding not entirely straightforward.The known method employs two to three differentiation stages as well as several delay lines, an analog multiplying stage and a combination of gates to reconstruct the data.

According to one aspect of the invention there is provided a method of recording digital data in which a digital encoded signal is filtered to generate a succession of pulses from edges of said encoded signal and in which said succession is recorded on a magnetic record carrier.

Preferably, the signal is filtered by a high-pass filter whose cut-offfrequency is between one to fourtimes the baud rate. A filtration Q-factor of about 0.5 is recommended. A second order filter is preferably employed.

Most preferably the cut-off frequency is two to three times the baud rate.

According to a further aspect of the invention there is provided a method of reconstructing digital data recorded according to said one aspect of the invention in which said magnetic record carrier is passed over a play-back head to generate a first signal which is then amplified to provide a second signal which is processed to reconstruct said data, wherein said second signal is processed by signal level responsive means such as a Schmitt trigger circuit or the like whose output provides said digital data.

For a better understanding ofthe invention, and to show how the same may be carried into effect, reference will now be made, by way of example, to the accompanying drawings, in which: Figures la and lb are waveform diagrams illustrating filtering of a digital signal for recording; Figure 1c shows examples of filtering circuits; Figure 2a shows a block schematic circuit diagram of an asynchronous multi-clock mode recording arrangement; Figure 2b shows waveform diagrams relating to the circuit of Figure 2a; Figure 3 shows a more detailed circuit diagram of the circuit arrangement of Figure 2; Figure 4a shows a circuit diagram of an asynchronous single clock mode of recording; Figure 4b shows waveform diagrams relating to the circuit of Figure 4a; Figure 5 shows an alternative filtering circuit; and Figure 6 shows a further pre-amplifier circuit.

Briefly summarized, the present invention contemplates processing the original digital data waveform by means of filtering to produce a relatively sharp pulse of certain characteristics from each positive or negative edge of the original data signal. In otherwords,the original data signal is filtered such as by a conventional RC filter of appropriate time constant. The invention further contemplates reconstructing the data by means of a simple Schmitt trigger circuit or the like.

The filtering action is illustrated in Figures la and 1b. Waveform A of Figure 1a illustrates a waveform of original digital data. Figure la further illustrates the filtered waveform B as well as corresponding recorder output waveform C following filtration by a first order passive RC filter. The front slopea ofthe recorder output signal is rather soft which increases the possibility of decoding error when using a Schmitt trigger since the time at which the Schmitt trigger circuit is activated varies according to the pulse amplitude. Also a degree of unstable overshoot is present at the rear slopes b of the pulses.

A second example depicts waveforms D and E related to a second order passive RC filter. The Q4actor ofthe RC filter is 0.5. Here, the filter output waveform D contains a certain degree of overshoot while the preceding example exhibits a very gradual transition of the filtered signal. Due to this overshoot, the recorder output signal E now appears with different characteristics. Firstly, the front sloped of pulses of positive or negative polarity is much steeper and is preceded by a small degree of pre-shoot c. Secondly, the polarity of the reproduced signal is reversed with respect to the previous example. Lastly, the intermediate portion e of the signal between the two sharp edges is of constant polarity and does not have a shifted zero level.In other words, the reproduced signal possesses ideal characteristics for operating a Schmitt trigger circuit.

Athird example illustrates the waveforms F and G in respect of a second order active filter whose Q-factor is 1/\/2 (largerthan 0.5). The filter output F presents a higher degree of overshoot. Like the second order passive filter, the recorder output G has steep rising and falling edges d but without pre-shoot. The signal, however, contains overshootfatthe rear ends ofthe signal pulses which extends into the pulse of opposite polarity, thus reducing if not taking away the Schmitt trigger margin. An active filter of higher O-factor will only emphasize the overshoot and the reverse overshoot and cannot therefore produce a recorded signal suitable for triggering a Schmitt trigger circuit.

From above considerations it can be now summarized that a second order high-pass filter whose Q-factor is of the order of 0.5 or thereabouts is suitable for reconstruction of the original data by means of a Schmitt trigger circuit.

Figure 1 b illustrates waveforms H, J, K, L and M obtained through filters of various cut-offfrequencies and the corresponding recorder output signals. The filtered signals are on the left and the recorder output signals on the right in each case. The cut-off frequencies fc are as follows where Z is the baud rate: waveforms fe H 0.5Z J Z K 2Z L 4Z M 8Z Although the range may vary slightly depending upon the recorder characteristics and types of magnetic tapes used, it is found that a cut-offfrequency of one to fourtimes the baud rate is likely to produce a recorder output of acceptable quality.When the cut-offfrequency is too lowforthe baud rate, as in the case of 0.5 times the baud rate illustrated as waveforms H, the recorder output becomes asymmetrical accompanied by a noticeable degree of "zero" level shift according to the recorded data pattern. Furthermore, the slope of the waves becomes too mild resulting in inaccurate reproduction of the original data.

On the other hand, if the cut-offfrequency is too high, overshoot pulses will appear and cut across the "zero" level as illustrated as waveforms M (8 times the baud rate). A certain amount of overshoot, however, can be allowed as illustrated as waveforms L (the case of 4 times the baud rate) where the overshoot or aftershoot portion stays at the same polarity. A cut-off frequency fc of 2 to 3 times the baud rate is likely to produce the most desirable results, but acceptable results may be achieved with a value of fc between 1 and 4times the baud rate. Waveforms K illustrate an ideal case with relatively sharp slopes and a small amount of preshoot.

The filtering requirements may be summarized as follows: 2nd order high-pass filter with 0=0.5 and cut-off frequency fc preferably in the range 2Z < fc < 3Z (where Z = baud rate) although the range Z#f#y4Z is acceptable.

Figure 1c shows a suitable passive RC filter where preferably R.=R2=R3=R, C2=C3=C and C4/ < C1 < C/2.

Approximate values of C and R may be obtained from the formula fc = 1/(27rRC) although the actual cut-off frequency of the high-pass filter will be shifted to some extent owing to coupling with the low-pass filter section. A final determination of the values can be made in the actual environment by referring to the waveform chart presented in Figure 1 b. Since the passive RC filter attenuates the signal greatly, an active filter may also be used as shown in Figure 1d. In this case,however, the value of feed-back resistor R4 should be carefully selected. Avalue of R4 between 2 to 3 times the value of Rs may produce acceptable results. Resistors R6 and R7 are optional to add gain. fc = 1/2irV(Ri.C1.R2.C2). Preferably 2R24R143R2.

Thus, the filter produces edge pulses corresponding to the rising and falling edges ofthe data present to be recorded. By means of the filter, the DC component of the data is eliminated and the edges are transformed into sharp edge pulses of both positive and negative polarity corresponding to rising and falling edges respectively.

By use of this filtered signal, digital data may be recorded upon a conventional magnetictape cassette, using a conventional magnetic tape cassette recorder. This is applicable to, for example, the problem of recording digital data from a RAM of a personal computer using a conventional magnetic tape recording machine for permanent storage. A computer-cassette recorder interface may be constructed utilizing this filtering technique and encoding the data by polarity-shift encoding.

Afirst circuit arrangement suitable for use in such an interface and employing an asynchronous multi-clock mode of processing is illustrated in Figure 2a.

The block circuit diagram of Figure 2a employs a universal synchronous/asynchronous receiver/transmitter (USART) which is represented by the block 12 and is a product of Intel Corporation, USA. Full details of the USART may be obtained from the manufacturers, but basically the USART receives serial data streams and converts them into parallel data characters for a processor. Similar communication LSls are available for various microprocessor families such as the Zilog Z80 SIO (Serial Input Output) designed for the Z80 microprocessor and the 6850 (ACIA: Asynchronous Communication Interface Adapter) for the Motorola 6809 microprocessor, etc. The methods and circuits presented in this specification can be easily modified for applications with such communication LSls.The USART can not only receive serial data but can also accept data characters from the processor in parallel format, convert them to serial format and transmit. The USART will signal the processor when it has completely received or transmitted a character and requires service.

From the transmitter data pin TxD (pin 19) of the USART, data T is taken to a filter circuit 11 which performs the filtering action as discussed with reference to Figures 1 a and 1 b. From the output of the filter 11, a data signal V is taken to a suitable input, e.g. the microphone input of a conventional cassette recorder and is thereby recorded on magnetic tape. The recording level is of course set to a value as high as possible whilst avoiding significant distortion. A setting at about half the maximum may be found acceptable.

From the transmitted data output TxD of the USART, in the asynchronous transmission mode, data bits are transmitted in serial form in a specific format. This format begins with a start bit, followed by five to eight data bits, then an optional parity bit, and finally one or more stop bits.

Following recording on the tape, the data may be played back and extracted for example from the headphone output socket ofthe recorder machine. Since the level of a headphone orearphone of a recorder is usually too low to operate directly a Schmitt trigger circuit composed of an FET operational amplifier, an amplifying stage 13 is included before supplying the reproduced data W to a Schmitt trigger circuit 14. The output X ofthe Schmitt trigger circuit is taken to the receiver data input RxD (pin 3) ofthe USART 12. although the amplification stage 13 is necessary, the waveform maintains a stable "0" level as a result of the correct choice of filtering factor as described with reference to Figures la, 1b and 1c.

Also connected to receive the output W of the amplifier 13 is a window comparator 16. This produces an output only when the amplitude of the input falls outside a certain "window" of amplitude. Thus it gives rise to one pulse for each edge of the original data signal. The window comparator output Y having rising edges is supplied to a signal detecting circuit 17, in which is provided a retriggerable monostable multivibrator with clear (type 74LS123) the output Z of which remains high as long as input signals are detected within a predetermined interval. The output Z of this monostable multivibrator is supplied to a polarity lock circuit 15 which upon receipt of a high output signal from the detection circuit 17 releases the Schmitt trigger circuit.

When there is no signal for decoding and the output of the signal detecting circuit falls to the low level, the polarity lock circuit 15 applies a high voltage to the non-inverting input ofthe Schmitttrigger circuit forcing its outputX to maintain the high level. Such a procedure is necessary to keep the "receive data" input RxD ofthe USART at a high level when there is no signal requiring decoding. This is a necessary requirement for use of the USART.

Figure 2b shows the waveforms T, V, W, X, Y and Z as well as clock pulses delivered by an oscillator 18 and divider 19.

Having established the presence of a data signal on its input, the USART starts its data assembly operation when the received data line falls to a low level indicating the beginning of the start bit. This resets the internal clock counter ofthe USART as shown in the bottom line of Figure 2b and the input data is then sampled at the programmed clock interval until the stop bit is detected. Detection of each start bit resets the internal counter so that successive sets of data can be correctly read. The oscillator 18 supplies a timing clock via the divider 19 to the receiver clock input RxC (pin 25) and the transmitter clock input TxC (pin 9) at a speed appropriate to the selected clock rate of sixteen times or sixty-four times the baud rate.

Figure 3 is a more detailed circuit diagram ofthe arrangement of Figure 2a where it can be seen thatthe filter circuit 11 designed for a baud rate of 8Kto 16K comprises three 4.7K resistors, two series capacitors each of 0.001 microfarads, and a shunt capacitor of 5000 picofarads. The head amplifier is formed from an integrated circuit type LM358, the Schmitt trigger is constituted by an amplifier type CA3130, the window comparator is formed from two amplifier types CA3130, and the signal detecting circuit 17 is a retriggerable monostable multi-vibrator (type 74LS122).

The polarity lock circuit is composed of a transistor type 2SA564 and a resistor. As described earlier, if the output of multivibrator 17 (74LS122) (pin 8) becomes low indicating absence of recorded signal, the transistor supplies plus five volts to the non-inverting input of the Schmitt trigger circuit 14 (CA3130) (pin 3). Thus, the receiver data line (RxD) of the USART can be kept at the high level.

In some cases, the polarity of the recorder output is reversed. For this reason a polarity changeover circuit is also provided in Figure 3.

It is also possible to program the USART for the so-called "single clock mode" in which the clock rate corresponds to the baud rate. In this mode, the USART's internal counter cannot synchronize itself with the receiver data input. One therefore needs to synchronizethe receiver clock input externally with the data signal (receiver data).

Referring now to the circuit of Figure 4a and corresponding waveforms of Figure 4b, two methods of achieving this synchronization are suggested: a) As shown in the upper portion of Figure 4b, a relatively wide pulse can be generated, e.g. at pin 5 of a retriggerable monostable multivibrator (type 74LS1 23), at every falling edge of the receiver data signal. The pulse width is adjusted to half of one bit data width. This pulse is used to reset the receiver clock generator circuit (divider counter type 74LS393). By supplying the reset pulse and the receiver clock generator outputto a NOR gate, a synchronized receiver clock can be produced. As illustrated, the NOR gate is constituted by an OR gate (type 74LS32) and an inverter (type 74LS04).

b) Alternatively, as shown in the lower portion of Figure 4b, a very narrow pulse can be generated by the monostable multivibrator at every falling edge of the receiver data signal and this pulse can be used to reset the receiver clock generator circuit, whose output (pin 8) provides the receiver clock directly.

It is also possible to carry out the invention using circuitry which does not employ the USART.

The filtering circuit used for recording may also be constructed as shown in Figure 5. Thetwo RC pairs which constitute a 2nd order high-pass filter are directly coupled. Here, the cut-offfrequency, fHc, can be calculated as follows: fHC = 2.67/2g7 > RHCH As described earlier, the fHc may be selected from the range of Z#fHc#4Z (Z=baud rate) A terminating low-pass filter is also recommended whose cut-off frequency, hc, is in the range defined as follows: 0.7Z/AsfLcs5.6Z/AZ where A=fHc/Z.

The value of RL and CL can be calculated from the formula: fLe = 1/2irRLCi (without load) In both cases, the load to be connected must be taken into consideration and the value of components should be adjusted accordingly.

An ALC circuit may be included in front of the Schmitt trigger.

Since variation of amplitude of played-back signal, due to characteristic of magnetic recording media, etc., tends to affect the Schmitttrigger response, it is advantageous to maintain the level ofthe signal, which isfed to the Schmitt trigger, in a workable range.

The pre-amplifier in Figure 3 may be replaced by the circuit shown in Figure 6. Here, an LED-CdS photo coupler is used as a feed-back element. The resistance ofthe CdS cell is used for determination of amplitude. If the signal exceeded a certain predeterined level, the LED emits light to the CdS. In responsethe CdS resistance will decrease lowering the amplitude of the section. Thus, the output level can be maintained at a predetermined level.

A A high-pass filter may be added in play-back section if desired.

Afirst order high-pass filter may be added in the pre-amplifier section to further minimize the "0" level shift and peak shift of the play-back signal.

The cut-off frequency of the high-pass filter may be selected from the range given below: 0.5ZSfHcSZ It will be apparent to those skilled in the art that the methods and circuits described herein according to the invention are applicable notonlyto NRZ (non-return to zero) data, but also to digital data encoded according to other standards such as mentioned in the introduction to this specification (but not the tone burst method).

Otherforms of reconstruction circuit will of course be necessary, but reliability of recording and reproduction may be improved.

Claims (16)

1. A method of recording digital data in which a digital encoded signal is filtered to generate a succession of pulses from edges of said encoded signal and in which said succession of pulses is recorded on a magnetic record carrier.

2. A method according to claim 1 in which said data is NRZ data.

3. A method according to claim 1 or 2 in which the signal is filtered by a high-pass filter.

4. A method according to claim 3 in which said filter has a cut-off frequency in the range of from one to four times the baud rate.

5. A method according to claim 4 wherein the cut-off frequency is in the range of from two to three times the baud rate.

6. A method according to any one of claims 3 to 5 wherein the filter has a Q-factor of about 0.5.

7. A method according to any one of claims 3 to 6 wherein said filter includes a low-pass section.

8. A method according to any one of claims 3 to 6 which the signal is filtered by a band-stop filter.

9. A method according to any one of the preceding claims wherein the signal is filtered by a passive filter.

10. A method according to any one of claims 1 to 8 wherein the signal is filtered by an active filter.

11. A method of reconstructing digital data recorded according to the method of any one of claims 1 to 10 in which said magnetic record carrier is passed over a play-back head to generate a first signal which is then amplified to provide a second signal which is processed to reconstruct said data, wherein said second signal is processed by signal level responsive means such as a Schmitttrigger circuit or the like whose output provides said digital data.

12. A record/play-back arrangement for digital data comprising: filter means for generating a succession of pulses from edges of a digital encoded signal; means for magnetically recording said pulses; means for playing back said pulses to provide a first signal; amplifier means for amplifying said first signal to provide a second signal; and signal level responsive means connected to receive said second signal and arranged to provide an output representing said NRZ digital data.

13. An arrangement according to claim 12 wherein said signal level responsive means is a Schmitt trigger circuit.

14. A method of recording digital data substantially as hereinbefore described with reference to the accompanying drawings.

15. A method of reconstructing digital data substantially as hereinbefore described with reference to the accompanying drawings.

16. A record/play-back arrangement substantially as hereinbefore described with reference to the accompanying drawings.