High density bias linearized magnetic recording system utilizing Nyquist bandwidth partial response transmission

TITLE SYSTEM FOR RECORDINGS

HIGH DENSITY MAGNETICS * LINEARIZED BY POLARIZATION * USING A PARTIAL RESPONSE TRANSMISSION WITH BANDWIDTH ACCORDING TO NY3UIST.

PRIORITY

USA DOM. BREV MARCH 1978

888130 DEL 20

Rome, II______Cili____

<img file="IT1112349B_D0001.tif" />

LU

Chiece-TkoA

THE

4 / 62SQ9

IRegister A

<img file="IT1112349B_D0002.tif" />

Protocol η * 21126 A / 79 utmSTtRO OF THE TRADE AND CRAFTS INDUSTRY

Provincial Office for Commerce, Trade and Crafts in Milan

COPY OF THE MINUTES OR THE DEPOSIT FOR AN INDUSTRIAL INVENTION PATENT

The year 1979 on the nineteen day at eleven minutes of the month of March fifty and the company SPERRY RAND CORPORATION jlxSigrwx of US nationality ^^ev '^ · Υ. 10019 (USA)

Via a mandate. Ineational Patent Office. C. GREGORJ and electively domiciled for the purposes of law in Milan - Via Dogana l at the authorized representative presented to me, the undersigned:

- Stamp application for the grant of a patent for industrial invention

MAIN; CXWTLEX »y (X having by title:

'' SYSTEM FOR HIGH DENSITY MAGNETIC RECORDINGS LINEARIZED BY POLARIZATION, USING A PARTIAL RESPONSE TRANSMISSION WITH BANDWIDTH ACCORDING TO NYQUIST

Inventor designat -

Priority of the patent application in: USAN 6SS 130 of 20 March 1978 supplementary to patent no. dep. granted on (application no.

accompanied by:

- Description in duplo of n. 75 pages of writing.

- Drawings, tables n, 7 in duptax examples, prow.

- Letter of assignment - Priority document and Italian translation (reserve)

- Authorization or deed of transfer, (reserve) xdQtehterszinnBc ^ iKDOcnanGDxteMljRVBatoanrwKesseÌBKMNMRRKiaKPiietSrmeVtocx

- Payment certificate on postal account n.00668004 in the name of the Tax and Concessions Registry Office

Rome of L. 341 .ooo ^ Mass by the Office Postal of Milan 32 on 19 * 3.79 η. 15θ

- Stamp duty of L. 2,000. The application, descriptions and drawings listed above have been signed by the applicant and countersigned and signed by me with the official stamp

The UFFICILE

Pietro Mesameo

<img file="IT1112349B_D0003.tif" />

ι

<img file="IT1112349B_D0004.tif" />

<img file="IT1112349B_D0005.tif" />

4/62809

<td>! OnJe MINISTRY OF INDUSTRY, TRADE AND CRAFTS</td><td></td>

<td>1 r Central Patent Office - ROME</td><td>21126 &lt;/ W</td>

<td>, 1 to Company</td><td></td>

<td>SPERRY RAND CORPORATION</td><td>. 2 Ί 2. i / H ';</td>

<td>organized and existing company according to the holm oaks</td><td></td>

<td>1 State of Delaware (US)</td><td></td>

<td>located at: 1290 Avenue of the Americas,</td><td></td>

<td>New York, NY 10019 (USA)</td><td></td>

<td>of US nationality</td><td></td>

<td>through authorized and domiciliary agent. PATENT OFFICE</td><td></td>

<td>ING. C. GREGORJ, Milan, Via Dogana 1 - request an at-</td><td></td>

<td>patent tested for industrial invention having</td><td></td>

<td>by title: «'' MAGNETIC RECORDING SYSTEM</td><td></td>

<td>HIGH DENSITY LINEARIZED BY P0LART7.7A—</td><td></td>

<td>ZIONE, USING A PAI RESPONSE TRANSMISSION? -</td><td></td>

<td>ZIAL WITH BANDWIDTH ACCORDING TO NYOUTST</td><td></td>

<td></td><td></td>

<td></td><td></td>

<td></td><td></td>

<img file="IT1112349B_D0006.tif" />

............----------------------------------------------------/·· ·· ··* ·· ·

RC-25044 4/62809 mb • * · · fc

Description of the invention having as its title:

<img file="IT1112349B_D0007.tif" />

HIGH MAGNETIC RECORDING SYSTEM

DENSITY LINEARIZED BY POLARIZATION, USING A PARTIAL RESPONSE TRANSMISSION

WITH BANDWIDTH ACCORDING TO NYQUIST.

on behalf: SPERRY RAND CORPORATION at: 1290 Avenue of the Americas,

New York »New York 10020 (USA) of United States nationality and electively domici * · -

<img file="IT1112349B_D0008.tif" />

• fcfcfc fcfcfc · _ · fc fc fc • fcfc in Milan »Via Dogana 1» at the authorized representative

Patent Office Ing. C. Gregorj

Filed the n. 2ll26 * (»+++++++++++

GENERALITY OF THE INVENTION

1. Field of the invention

The present invention relates to systems z) | storage maghetic, digital type, high density, of the type used, in general, for large capacity storage for computers.

Magnetic disk storage systems using the concept of recording and reading of the non-contact type represent specific types of such systems.

OFFICE tìkfcVETTI lng. C. GREGORJ

2. Description of prior technology

Current magnetic memory systems

<img file="IT1112349B_D0009.tif" />

• ···· • ·· · • *· · * • · ··· « ····· · · ..........

- 2 sa, digital, non-contact type, use the concept of saturated flow recording, in combination with limited codes in the sense of-.

the length, in order to reduce costs for each bit of information that is recorded while maintaining the ability to increase the online capacity of the.

&lt;memory systems. These codes, although they allow to alleviate the problems inherent in the processes of recording random data, of the loss, of the timing and of the interference between the flow transistors, operate at the cost of requiring the registration of a greater number of code bits compared to the corresponding ones data bits. Multi-level registration, i.e. involving the use of more than two levels,_F it is not used to increase the storage density of the data since the nonlinear characteristics of the magnetic interface do not allow this general recording format, reliably. However, in a specific way, it would be possible a ternary recording in a non-linear channel, through the use of three states represented, specifically, by the positive saturation, by the negative saturation and by a flow equal to zero. Since the zero flow level of the

<img file="IT1112349B_D0010.tif" />

<img file="IT1112349B_D0011.tif" />

<img file="IT1112349B_D0012.tif" />

·· · ··♦· » · · • « · ·

I ··· • ·· * • ··· a «» *

- 3 ternary recording does not allow to erase previously recorded data overwriting with new data would require a long erasing cycle or an additional erasing transducer. For various reasons, it is not possible to * use a separate erase transducer in disk archives. Furthermore, it should be noted that communications theory indicates that contemporary commercial systems have not yet obtained bits densities that approach the theoretical limit set by Nyquist.

regardless of what has been previously indicated, partial response signaling has been developed to increase the capacity of the communication channels with reference to the frequency of data transfer, these communication channels • being represented, for example, by telephone transmission systems. The concept of partial response data transmission has been described in U.S. Pat. 3,388,330, released June 1, 1968, entitled '' Partial Response Multilevel

Data System¹'and in U.S. Patent No. 3,492,578 issued January 27, 1970, entitled' Multilevel

Partial-Response Data Tr ariani ss ion ''. Although the partial answer has been taken into account ····· · ·

........

« » ·*·» ··♦· ·*·

<img file="IT1112349B_D0013.tif" />

• * ····. * ♦ ·· * · • «· • * * ·· · • ·· • · · · ···· ··· ..... • a ari» ·

- 4 for use in magnetic, digital storage systems, this signaling technique has hitherto been contemplated as applicable to memory systems using a saturated flow recording. It is believed that these systems considered in the prior art are subject to difficulties with regard to the recovery of the timing while in the case of the use of recording systems multi-level automatic gain control is obtained better than that required by contemporary practice and, in any case, The previously described problem associated with the deletion of old data is prevalent.

Furthermore, in these systems proposed by the prior technology, difficulties were encountered in modeling, or configuring the spectral frequency response of the channel to partial response signaling, particularly with reference to the partial response of Class IV which represents one of the most useful signaling classes partial response due to the fact that it allows obtaining the Nyquist frequency with a minimum channel bandwidth. Since partial response reporting requires a linear combination 'a ·. · *, · * · • ·

C.

· » 4 « ......

* « « 4

Ing *

- 5 pulses in the data flow for an exactly and accurately controlling intersymbol interference, the saturation non-linearity of the magnetic interface, aggravated at the high data densities, would tend to distort the partial response data combinations, making it difficult, if not impossible, the recovery of significant data.

according

Therefore,; to what has been previously indicated it is evident that the applications of the techno-, already earlier concerning the partial response signaling, to the memory systems involving the use of magnetic means, as regards the bi-data_t nari to be stored are limited to a binary data rate equal to the pulse repetition rate according to Nyquist for a saturated stream recording system with minimal bandwidth. As previously indicated, it can be noted that if ternary amplitude pulses are recorded using positive saturation, zero flow and negative saturation to represent ternary levels and achieve a 50% increase in the frequency of binary data using bandwidth allowed by the magnetic interlayer, a * 4 · »* • 4 ··» * ♦ · would be required,

<img file="IT1112349B_D0014.tif" />

• » 4 · •444 ·*·· • 44

<img file="IT1112349B_D0015.tif" />

··· • · · · · • · * • · » ·„*

- 6 - separate erase cycle or a separate erase transducer for erasing old data. These erasure techniques are particularly difficult in commercial type disk storage systems.

SUMMARY OF THE INVENTION

The present invention contemplates the use of a polarization or premagnetization signal to linearize the magnetic interface in a system for storing digital data operating on a magnetic medium with partial response signaling. Preferably, a partial Class IV response is used in which the system transfer function is shaped in such a way that the frequency spectrum of the pulses representative of the data has a sinusoidal lobe while in an enlarged Class IV partial response system, multiple sinusoidal lobes occur. The polarization allows the use of a pilot tone for an exact recovery of the timing independently from the data and, in a similar way, for an automatic control of the gain, independent of the data, of an accurate type, to alleviate the adverse effects of the distortions t

multiplication associated, for example, with variations!

• · · ···.,

<img file="IT1112349B_D0016.tif" />

· 0 ··· • 0 * - ’ » • · ···*

UFFlC. "&gt; 6fte'Ì ^« ri

In »C ^ EGORJ

- 7 in the "" floating "height of the head» associated with the channel. Furthermore, the channel linearized by polarization allows recording of the non-saturated type of pulses at multiple levels represented, for example, by ternary and · - quaternary impulses, with the consequent possibility of a &lt;significant increase in the density of storing information of the long system the track. In other words, the frequency of information transfer, expressed in hertz of bandwidth, is now significantly higher while still adhering to a minimum bandwidth system according to Nyquist. The polarization signal, as well as the linearization of the channel, allow to delete the old data, at the same time as the writing of new data, in such a way as to eliminate * the practical difficulties that are encountered in compliance with what previously described. The linearization of the channel by means of polarization also facilitates the configuration of the purity of the channel for the partial response in class IV and also facilitates the inclusion of the pre-emphasis on the recording side, to improve the signal / noise ratio, this ratio not being obtainable in commercial disk archives.

» ·»·4

Ϊ · ·· ·· and ::^:·· *4 »

• 4 • · * 44 * • •• a ··· · · · I aaa · • · a ·

SFHCiO BRijvfcEV

9 · &lt;®Μ

The preferred version of the invention also allows to convert the data into a new sequence of interleaved bi-pulses for the generation of the recording signal. This arrangement allows obtaining a spectral null slightly above the data frequency band for the insertion of the pilot tone.

It is foreseeable that the attempts of the prior technology in the use of a partial response with minimum bandwidth in digital recording systems of the magnetic type, will not lead to satisfactory results for a high linear recording density since the non-linearities of the magnetic interfaces tend to distort the appropriate linear combinations of the pulse width required in partial response signaling. By using alternating current linearization polarization with partial response signaling, linear recording densities were obtained four times higher than those obtained in accordance with the prior technology with similar type magnetic interfaces.

·* ··· ·

<img file="IT1112349B_D0017.tif" />

. ··· ·· · · · · « · 9 9 9

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be more evident

OFFICEÒ eBE ^ ETTI

O. Cì'GROSORJ * * fc • fcfc * · fcfcfcfcfc fc «fcfcfc from the analysis of the following detailed description, which must be considered in conjunction with the attached drawings, in which:

Figure 1 is a schematic block diagram of the recording portion of the digital recording magnetic system proposed by the invention;

Figure 2 is a schematic block diagram of the reading portion of the magnetic recording system, of the digital type, proposed by the invention;

Figure 3 is a representative diagram of the waveform of the interleaved bi-pulse signal used in the system proposed by the invention;

Figure 4 is a waveform diagram useful for describing the operating principle of the system proposed by the invention;

Figure 5 is a diagram of the frequency spectrum of a representative pulse of data shaped in the system proposed by the invention, in accordance with the signaling of the class IV partial response type;

figure 6 represents a diagram of the frequency spectrum of the bi-pulse sequence infc · ····

-10 tercalates used in the system proposed by the invention;

figure 7 is a schematic block diagram of the derivative equalizer proposed by¹ invention, of the type schematized in figure 2;

Figure 8 is a schematic block diagram of the transverse filter schematically represented in Figure 2;

Figure 9 is a schematic block diagram of the portion of the digital signal format generator schematized in Figure 1 which generates the sequence of interleaved and real bi-pulses:

it synchronizes the data format with synchronism and preamble signals; and figure 10 constitutes a diagram illustrating the timing of the operations of the circuit schematically shown in figure 9.

DESCRIPTION OF THE FAVORITE VERSION

The preferred version of the invention is contemplated for application to an archive, that is to say to a magnetic disk "file" for mass storage of computers. It should be emphasized that this field of application is purely exemplary in that the concepts of the invention • fc fc · * · * · * · ** .fcfc.

&lt;fcfc · · · * · ···· ···. · • * • ·· *. * ····· '· ····· · «

.......

. · . . · « ····· • :· • · · · · ·.»· · .· • · * · ·

..... ··*· ·· · · · · • . ... · «

• «fcfc 0 fc fc.fc · • fcfc

- 11 are applicable to any channel for magnetic recordings and reproductions. The theory of communications indicates that fundamental binary information densities along the recording path are theoretically potentially available, significantly higher than those obtained so far in commercially available disk archives, in disk recording systems of the non-contact type. The digital system for the magnetic recording of data proposed by the invention has characteristics considerably superior to those of the previous technology tending to the final densities expressed as linear density of the bits and is characterized by an increase of 3-4 times as density in comparison with commercial products contemporaries proposed by technology. Partial response signaling was used to achieve this significant increase in the linear density of the bits. Although this signaling configuration has in the past been considered for use in magnetic recording channels, it has been believed that these applications were valid only for a saturated stream recording format. Furthermore, it was believed that due to the non-ì

linearity inherent in the magnetic interface, these * * • · · · ···· «'»

-to , • · · ".· • · '· · " · ·

1Ipr iG0r attempts would not have been successful due to the high linear density of the bits.

Consequently, the present invention uses, for example, a polarization field created by a very high frequency alternating current, to linearize the magnetic interface, thus obtaining a reliable improvement in performance, using, for the first time, an unsaturated flow recording, with partial response reporting.

The use of polarization also allows simultaneous, low noise cancellation of old data when new data is recorded.

The linearized interface allows the insertion of a sound tone for the recovery of the timing and for an automatic control of the gain. The linearized medium also allows a multi-level recording, with a consequent significant increase in the density of information storage. Furthermore, the linearized medium allows, in a convenient way, equalization both from the recording side and from the reading side, it allows the filtering, the pre-emphasis and the shaping of the channel to improve the signal / noise ratio in the format reading phase partial response signaling, without aggravating the distortion.

»Fc fc • fcfc * ·· • fcfcfc J • ·« · ¹

<img file="IT1112349B_D0018.tif" />

.··· * ··

Ί3 Referring now to the specific figure:

it can be observed that a schematic block diagram of the registration section of the system made in accordance with the principles of the present invention has been reported. The electronic section concerning the recording, i.e. writing, is controlled by a clock signal generator 10 of the system which, in the version currently considered, consists of a crystal controlled clock signal generator operating at a frequency of 80 MHz. The binary data to be recorded on the recording medium is provided by a binary data source 11 synchronized by the clock signal generator 10 of the system. In figure 4 some typical binary data provided by source 11 have been reported, these data having been identified by the explanatory notes shown in figure 4.

The binary data derived from the source 11 is converted into equivalent data, at several levels, in a binary / L-ary converter 12. It must be noted that the binary data can be converted into data having any number of levels, capable of to be reliably recorded • · ·

<img file="IT1112349B_D0019.tif" />

«

<img file="IT1112349B_D0020.tif" />

.♦·

0*0 0 • 00 • ···* • 0 000

- 14 on the magnetic interface and recovered from said interface, using the system's partial response signaling complex. It must be pure

It is found that the present invention could be concretized with a two-level recording and, in this case, the converter 12 would not be used. In the illustrated and described version "a binary to ternary converter is used to orderly arrange groups of three binary digits representative of eight states, in two amplitude pulses, of the ternary type" representative of nine states, one of which is not used.

Converter 12 provides a characteristic map of binary data to ternary data. Figure 4 illustrates the trend of ternary data provided by the converter 12 in response to the binary sequence illustrated in said figure.

THE

The ternary data is supplied at a frequency of 20 million pulses per second while the corresponding binary data is fed at a frequency of 30 Mbits / sec. Therefore, in the illustrated version, the binary data characterized by two levels represented, for example, by the binary level 1 and by the binary level 0, are converted into ternary data having three levels constituted, specifically, and by way of example, by the levels. ·. ·: * ••• Φ • · 0 0 »

0 ·· • r. 0 0 0 * ···· • 0 · 0 _ • * «

<img file="IT1112349B_D0021.tif" />

• · ·

- 15 them +1, 0 and -Ί. The conversion process carried out by the converter 12 is synchronized by the system clock 10. It should be noted that for the practical embodiment of the invention it is possible to use any appropriate binary / ternary converter 12, such converters being commonly used in classical partial response communication channels.

In the version currently considered, each ternary output pulse, derived from the converter 12, is represented by a parallel pair of binary digits. Therefore, the binary input sequence is uniquely converted into a sequence of parallel pairs of binary digits representative of the converted ternary data. In the preferred version of the invention, the ternary data can be recorded with an unsaturated stream, although it is evident that the ternary data is

Compatible with a saturation recording (involving the use of a positive flow saturation and a negative flow saturation together with a zero flow). ral registration in ì

saturation being used in accordance with the invention since polarization provides for the simultaneous cancellation of the previous data.

- 16 For convenience »the sequence of ternary impulses output from the converter 12 can be denoted by the reference while the various individual ternary impulses of the sequence are distinguished • · • ···· ·· * • * · * · ·· • * ·&gt; · * * · • ♦ · from reference B.

n

In partial response signaling systems it is generally known that the input data must be pre-coded in order to prevent the propagation of errors in the decision circuits operating on the received or reproduced pulse sequence. Consequently, the ternary pulses output from the converter 12 are fed, in the form of an input, to a pre-coder 13 whose operation is synchronized by the system clock 10. The precoder 13 in turn provides the sequence of pre-coded ternary impulses. The sequence derived from the converter 12 is precoded, on a module basis, in accordance with the rule for reporting przialé according to Class IV, in accordance with what is indicated:

C = fB + C „\ mod- £ η l η n-2 I in which it represents the multilevel data module. In the ternary version currently considered, (

the precoding equation is as follows:

C * (B + η ln n-2 mod 3 ··· • fcfcfc · * ··· fcfcfc

- 17 The values ​​in the modular series of ternary elements are considered to be equal to: (-1, 0, +1).

Figure 4 illustrates the precoded ternary data corresponding to the exemplary ternary sequence in said figure. The pulses in the various illustrated clock or timing ranges have been identified by means of various typographic characters, that is, by means of dots, dashes and small circles, by way of illustrative example.

It must be emphasized that various precoding circuits are known in the field of partial responses. In the currently preferred version of the invention, the sequence of parallel pairs of binary digits representative of the ternary pulses is fed, as input, to the precoder 13 which performs the previously defined module 3 arithmetic function, so as to obtain the sequence of pre-coded ternary impulses, in the form of further parallel pairs of binary digits representative of the same. The frequency of the pulses output from the precoder 13 is equal to 20 pulses / sec.

The precoded ternary data, derived from the precoder 13, are fed, as input, ** · * · • fcfc?. · • fc ·· · • fc · ····.

···· * * fc

14, which is synchronized to the system. The creator of digital 14 is used

- 18 to a creator of the digital signal format, indicated by the block 10 of the clock 10 signal format for controlling the bias signal for recording, for generating a pilot tone and to arrange the data in the desired format, to obtain a recording signal. The digital data arranged in the desired format are converted into a recording signal through a classic digital / analog D / A converter 15 which accepts the previously precoded parallel copying of binary digits and converts the same in the corresponding ternary impulses. The total recording current is represented by the sum of three components that can be considered as an information support signal, as a pilot tone and as a polarization. The components represented by the pilot tone and the bias signal of the recording current will be described in greater detail below. The component represented by the information support signal consists of three successive sections considered as synchronization section, preamble section and data section. In the range of »··· ··· ····::. :

• · fc fc ··· »· ** · •» · · · * ···

- 19 synchronization, the signal consists of a sine wave having a frequency of 5 MHz, in synchronism with the 20 MHz data time base and the same is used to solve a phase ambiguity in the recovery circuit; of the timing in reproduction, in accordance with what will be later described in greater detail. The preamble interval can be represented by any predetermined fixed impulse scheme to provide an '' indicator '' or '• flag'¹ for starting data. It should be emphasized that suitable traditional circuits contained in the creator of the formats 14 allow the digital representation of the synchronization and preamble signals to be generated; they are then multiplied temporally with the pre-recorded digital ternary data, according to what represented by the parallel binary pairs derived from the precoder 13 ·

The creator of the format 14, in combination with the digital / analog converter 15, provides the information support signal used for recording, in the form of a sequence of interleaved bi-pulses. Format creator 14 supplies control signals to converter 15, ί t * «* • *« «· *« * · ·· · «» * ♦ ·· • * ·· # f »

<img file="IT1112349B_D0022.tif" />

···· ····» ·»·

- 20 so that each ternary impulse having a value other than zero in each clock interval of 50 nanoseconds, is decomposed into a pair of subsidiary impulses having a duration of 25 nanoseconds, extending beyond the clock intel, this pair being considered as an intercalated bi-impulse. The first subordinate impulse of each bi-impulse has the same polarity as the corresponding impulse representative of the corresponding precoded ternary data and occurs in the first half of the clock cycle in which the ternary impulse occurs. The second subordinate impulse of the impulse also has a duration of 25 nanoseconds, while it is characterized by the same absolute value as the first sub- impulse, ordered while it has an opposite polarity with respect to the same, the second subordinate impulse occurring in the last half of the interval of clock of 50ranoseconds that occurs later.

Figure 3 illustrates the generic bi-impulse at inter-;

The reduction obtained by the decomposition of each of the precoded ternary impulses having a duration of 50 nanoseconds, when the same have an amplitude other than zero. The polarity of each of the subsidiary pulses of the bi-pulse is selected * -

<img file="IT1112349B_D0023.tif" />

• · · .··

<img file="IT1112349B_D0024.tif" />

yefiCSMWtSVETTI

i.ng. ce &amp; asQRJ • ··· *

- · - * * «· born in accordance with the polarity of the ternary impulse, in accordance with what previously described. When the ternary impulse is equal to zero in conveying data, the bi-impulse will result • 4 4 4 4 4 · · · »

2· »4 · ’··· •

« • 4 »44·

44 4 • 4 *

<img file="IT1112349B_D0025.tif" />

4’

4 4 identically equal to zero while the subsidiary impulses will not occur. Figure 4 illustrates the sequence of intercalation bi-pulses resulting from the pre-coded example ternary sequence that has been illustrated and, from the analysis of this figure, it can be found that the intercalation bi-pulse can be modulated at binary, quaternary or L level, as optionally required in other appropriate versions according to the invention. The lines indicated with dashed lines, in the form of dots and circles, identify the particular bi-impulse deriving from the precoded ternary impulse identified in an analogous way * The voltage spectrum of the intercalation bi-impulse sequence was represented in Figure 6, from the analysis of which it can be found that the signal does not have a continuous component while the same is characterized by a better high frequency content within the limits imposed by a null point in the spectrum, thus reducing the requirements associated with a circuit of high-fre pre-emphasis »00» · * «0 *

0000*

-00*

00

0 0

- 22 queries in the data band, according to Nyquist. Above this band, the frequency spectrum is attenuated to a convenient null point;

for the insertion of the pilot tone. Furthermore, from the anatlysis of Figure 4 it can be found that the biim-, pulses are interleaved or engaged in such a way;

that the resulting sequence may present the same number of levels of the precoded multi-level pulse sequence from which said biimpulses are immediately derived. This is advantageous in the details of the generation of the sequence of interleaved bi-pulses which will be described below with reference to figures 9 and 10. In more general terms, it can be found that in data transmission systems represented, for example, by systems using a coaxial cable channel this plug-in feature facilitates the use of a non-linear transmitter for a maximum output power concentrated in the minimum data band, according to Nyquist, when the concept of partial response, binary or multilevel signaling is adopted. By applying this feature of the invention involving the use of intercalated bi-pulses, it must be noted that a main advantage is found

0 0 00 00000 ,00.

0 0 0*

000⁴

- 23 ··.

between in the higher high frequency content, with the possibility of obtaining a greater signal / noise ratio with respect to the attenuation of the channel with regard to the frequencies of high value, instead of in the corresponding spectral null, in which case a pilot tone can be used, or less, in a contemporary way. When a pilot tone is not used, timing and automatic gain control can be recovered • · · * »• · | • fcfcfc «fcfc ·· • fcfc • fcfcfc * from the same data, in a way known in the specific technology, although this is associated with a certain decrease in the effective data transmission frequency.

With specific reference to figure 4, it may be convenient to conceptualize the generation of the sequence of intercalated bi-pulses in the following way: firstly, the precodecyclic ternary sequence multiplies a periodic sequence of unit pulses having a period of 50 nanoseconds, for the production of the precoded impulse sequence. The precoded pulse sequence is involved with the intercalated bi-pulse of Figure 3 for the generation of the biim sequence—

The interleaved pulses represented in Figure 4.

It must be emphasized that the apparatus

THE

<img file="IT1112349B_D0026.tif" />

··· • ·· «

- 24 ⁿ° n virtually uses this procedure which, however, has been described to facilitate understanding of the principle of operation of the invention.

Although the version currently under examination has been described in terms of the use of intercalated bi-pulses, it must be underlined that the pre-coded multi-level signal, exemplified as constituted by a pre-coded ternary signal in figure 4, could be used directly for recording, through the digital / analog converter 15, eliminating the interleaved bi-pulse format. In this case, a spectral null of the data could be provided for the insertion of the pilot tone, by means of the low-pass filter 17 schematically shown in Figure 1,

The output derived from the converter 15 is fed, through a recording equalizer 16 and the low-pass filter 17, as input to a sum amplifier 20. The recording equalizer 16 compensates for the losses associated with the interface, by using a low frequency and high frequency preemphasis. In the version currently under consideration, the low frequency and high frequency preemphasis is applied below and

<img file="IT1112349B_D0027.tif" />

*· »· • · · ·

<img file="IT1112349B_D0028.tif" />

» • * • · ···· • « *

<img file="IT1112349B_D0029.tif" />

« ♦ , · » * » , •· #*

- 25 above 2 MHz respectively. The recording equalizer 16 can be concretized, as an approximation, within an almost linear portion of the characteristic representative of the trend of the phase as a function of the frequency tu in radians of the equation of the transfer function: i

OUTPUT / INPUT = Α / ω + βω + Ccu³by means of a classic network which resembles the one schematized in figure 7 "in which the concept of electronic signal integration is introduced. It must be underlined that in the classical recording systems, of a magnetic type, with a binary character, adopting the concept of saturation, this equalization is not possible due to the presence of the non-linear interface. By means of the new combination proposed by the invention, the equalization in the writing or recording section can be used to allow obtaining a significant increase in the signal / noise ratio by means of an optimal division of the total shape required for the system transfer between e- '.

the qualification in the registration section and that of the reading system. Filter 17 consists of fc »fc&gt; _&gt; fc ·» fcfc • «fcfc» ί · * · »· fc •» · * «

fc4 «fc

- - * lng. ΛΕΡ'ίΛΕΡβ ·· · ····

- 26 a low-pass, linear phase filter capable of considerably attenuating the high frequency components above, for example, at 12.5 MHz in order to prevent the generation of unwanted intermodulation products, the frequency of which is between 0 and 10 MHz during the recording process. *

An appropriate circuit assembly included in the creator 14 of the digital signal format generates a square wave in response to the signal generated by the clock generator 10 of the system, at the frequency of the pilot tone, this frequency, in the version currently considered, being equal to

1/3 MHz. A filter 21 allows to extract the fundamental of the square wave constituting the pilot tone in such a way as to provide the sinuous pilot tone consistent, as phase, with the preamble synchronization signals and with the representative data signals . The pilot tone derived from the filter is added to the signal output from the filter

17, in the sum amplifier 20.

The output of the summing amplifier 20 is fed, as input, to a linear recording amplifier 22 which allows to combine the polarization signal with the pilot tone and with the data representative signals. The component ·· ·· »·« · • ...

··· • « · · · -·♦+

- 27 polarization lens can be constituted, for example, by a sine wave having a frequency of 38.5 MHz such sinusoidal signals being supplied by a crystal oscillator 23.

It is not necessary for the polarization to be coherent, as a phase, with the information support signal and with the pilot tone but, however, it must be noted that this polarization signal must have a sufficiently high level and frequency) high so that any residual non-linearity of the support means of the;

recordings can cause distortion in the- or equalized signal reproduced, even for data configurations that are found in the worst case. The polarization signal is fed- »

Connected to the amplifier 22 through a gate circuit 24 which allows, minus, the passage of the polarization signal under the control of a signal derived from the creator 14 of the format of the digital signals, when required, during the recording interval . The output from the recording amplifier 22 is fed to the recording head, so as to allow the recording of the signals on the recording support medium. Must be underlinedf ·· * «·« ··· »·· ♦ ···· • * ·· • M ******

28 to the fact that the bias signal transferred through the circuit 24 can be alternatively amplified separately from the linear amplification of the output of the sum amplifier 20 while the resulting output currents can be added together in the recording head.

With specific reference to Figure 2, it can be noted that a schematic block diagram of the electronic assembly of the reading system of the version of the recording system, of the magnetic type, proposed by the invention has been reported. The upper row of components forms the processing section of the analog signals representative of the data of the reading circuit complex while the lower row of components allows to perform the functions of recovery of system timing, sampling, decision and logical conversion into data tracks.

The signal that is read by the head is fed to a preamplifier 30. The component represented by the pilot tone is extracted from the reading signal by means of a filter 31 of the pilot tone which applies the pilot tone to an envelope detector 32 for purposes of automatic gain control and a PLL phase lock ring (phase

0 0 * 0 00000 • 0000 s * ·· ·· * •• 0 · * 000 o F f ^{£ TTl}

In »« «« »« »··· *

- 29 - ·:.:. : ··· locked loop) 33 for the recovery of the system reaction time * The circuit 33 consists of a classic envelope detector which provides a signal in accordance with the recovered envelope of the registered pilot tone. Therefore, it can be found that when variations of the ford occur;

gno deriving, for example, from variations in the height of '' buoyancy ** of the head, the amplitude.

of the envelope signal varies in proportions, direct with the gain variations found in the cor-. frequency of the pilot tone frequency equal to 13 1/3 MHz.

The preamplifier 30 supplies the signal representative of the reading data to a low pass filter 34 which has a flat passband

I ì

with a linear phase, this filter being used, in the system currently under examination, to attenuate the spectral components above 18 MHz, so that subsequent processing must not introduce significant noise or should not cause the introduction of non-linearity .

The filtered signal is fed to an automatic gain control amplifier (CAG) which receives a gain control input from the envelope detector 32.

*· · ···· ·· • · · ··* • · • · · ···· • ··*

444 ···· .··

- 30 In the system currently under examination it has been determined that the gain variations, expressed in decibels (dB) of the interface, are substantially proportional to the frequency within a wide bandwidth. The behavior of this system is believed to derive mainly from changes in the head's buoyancy height.

the

Ideal gain control to compensate for these frequency dependent gain fluctuations would be overly complicated and, as a result, a simpler automatic gain control circuit is used in the version currently under consideration. This is possible because in the version currently considered, although the signal energy extends from 0 to 10 MHz, the signal energy is dominant in the vicinity of 5 MHz for operation with partial response in Class IV. in addition, it should be noted that the required gain variations are small. Consequently, the automatic gain control amplifier 35 uses a gain control function for the entire signal, corrected for a frequency of 5 MHz, thereby providing a good approximation for ideal control of the broadband gain. The amplifier 35 of

444 ··*·· ···· · ···· ♦

•4 44 4 ····’ · 4 4» ·

»44 4 · ~ 4 · 4 4 4 • · • fc ·· fcfcfc ··. ··

<img file="IT1112349B_D0030.tif" />

fcfcfc • · ·· fcfc »····· • ·» · automatic gain control takes advantage of the gain control law given by:

FCFC »

• fcfc • · «fcfc

<img file="IT1112349B_D0031.tif" />

• fc · · * ·

<img file="IT1112349B_D0032.tif" />

where X represents the input to the amplifier

35, derived from the envelope detector 32. Therefore, it can be found that the amplifier 35 provides the desired gain control as an exponential function of the variations from the pilot tone envelope. The positive constants and are adjusted according to specific parameters:

of the system, while it represents any convenient number for the realization of the automatic gain control circuit. The constant C₂ is selected according to an experimentally derivable ratio equal, in this case, to 13-1 / 3/5, between the variation of the gain, in dB which occurs at the frequency of the pilot tone which, in this practical embodiment in accordance with the the invention is equal to 13 1/3 MHz and that which will occur i

at the center of the data band whose value is, in this case, 5 MHz.

The output signal from the automatic gain control amplifier 35 is passed through a derivative equalizer 36 and through, fc fcfcfc fc * · · • fc

Uff ing.

0Λ. · ^ Ερ © «. · ** ···· a transverse filter 37 in such a way as to provide for spectral shaping and phase correction. These components are used to measure the signal spectrum and to obtain the desired contour of the partial response in Class IV with linear phase. The derivative equalizer 36 has a transfer function which approximates an almost linear curve representative of the phase trend as a function of the frequency in radians ω:

<img file="IT1112349B_D0033.tif" />

*»· ·»·* • » · · *

· »»»♦ » »0 » &gt; » 0 « «

0^β • 0 0

<img file="IT1112349B_D0034.tif" />

OUTPUT / INPUT = (D - EU&gt;² + FU⁴) with the coefficients D, E and F regulated in accordance with. the parameters of the system, in such a way as to compensate for the variations that occur in the i functions

interface transfer, resulting from track-to-track variations in the wavelength, at any given frequency. Branch equalizer 36 compensates for interface losses at short wavelengths and also compensates for the effects of the beam. It can be i

used a registered frequency brushing to adjust the coefficients D, E and F, so that the output of the device in response to the frequency brushing allows to obtain a flat frequency response. Afterwards verκπβπΒ »* ·

- 33 further details will be given regarding the traditional network comprising the derivation equalizer 36, referring specifically to. figure 7.

The transverse filter 37 provides for an amplitude and phase equalization, that is to say with a shaping, or spectral shaping. Filter 37 provides an equalization from 0 to 20 MHz, within a frequency resolution of approximately 1.5 MHz.

The transverse filter 37 is adjusted in such a way that the pulses representative of the data that are transmitted through the channel are converted into the appropriate form corresponding to the partial response in class IV, as shown in Figure 5. The circuit allows to correct both the phase reset and the system width response, operating in accordance with what will be evident to those skilled in the art. The details of the transverse filter 37, concerning the structure and the adjustment thereof, will be reported below with specific reference to Figure 8.

The spectrum associated with the partial response in is

Class IV represented in figure 5 / constituted, ···· • ··· • · * · ideally, from a sine-sinusoidal lobe from 0 to 10 y ¹

MHz, that is to say according to [ω / (2.10)], and we go 34 ¹ *3 ···· ·· « ·« ··

<img file="IT1112349B_D0035.tif" />

zero at a frequency higher than 10 MHz. It should be underlined that alternative versions could be used within the scope of the invention, having a partial response frequency spectrum comprising two or more sinusoidal lobes. For example, two sinusoidal lobes from 0 to 10 MHz can be used, with a¹null at a frequency of 5 MHz for the insertion of a pilot tone. Alternatively, a sinusoidal lobe can be used in accordance with the partial response in Class IV but with the pilot tone inserted at the marginal zero of the band a

MHz.

*··· • ··· · • ··**

The output derived from the transverse filter 37 is fed to a low pass filter 40%; represents the final filter of the system, having a flat and linear phase bandwidth up to a frequency of 10 MHz, this filter being able to attenuate input components greater than 11 MHz. the circuit 40 also includes a filter + tuned to a frequency of 5 MHz to provide the previously described synchronization signals which are conveyed along a line 41 for reasons which will be described in greater detail below. The impulsive signal, representative of fc fc lng- • $. «»^Tec”^{U β} .· · ····

- 35 The data, filtered and equalized, is fed¹ along a line 42, this signal having been represented in Figure 4, as a reproduced equalized signal. The specific composite waveform • fcfcfc «fcfcfc

<img file="IT1112349B_D0036.tif" />

fcfcfc * • fcfcfc fcfcfc

<img file="IT1112349B_D0037.tif" />

• fcfc • fc ·· illustrated is formed by the impulsive components indicated by the curves in dots, dashes and small circles corresponding to the components identified in the same way in the sequence of interpulse pulses—!

the sides deriving from the precoded ternary impulses I ì

representative of the data, in accordance with what previously described. It must be stressed.

the fact that creator 14 of the digital signal format controls the digital / analog converter ι

co D / A 15, in such a way as to allow obtaining. of the interleaved biimpulses previously described.

The equalizers and filters 16, 17, 34, 36, 37 and pre-emphasize, equalize and form the frequency spectrum of the pulses representative of the data so as to obtain, together with the linearized magnetic interface, a close approximation to the ideal spectral configuration , sinuous type, of Class IV represented in the figure;

5. This spectrum is identical to that of any impulsive component of the composite signal fed along line 42 and represented, with its own im »1" _Λ

- 36 duct "" in figure 4.

As previously described, the phase lock ring PLL 33 is sensitive to the pilot tone, so as to provide for the recovery of the system timing. The phase lock ring

PLL 33 acquires and follows the filtered pilot tone, having a frequency of 13 1/3 MHz in the example currently considered and provides a reproduction clock signal at 40 MHz which is divided by 2 in a clock signal divider 43, so as to obtain the pulse frequency required for sampling the analog signal representative of the data on line 42. The clock signal divider 43 consists of a divider by 2 circuit used to provide the required clock sampling signal having a frequency of 20 MHz. Since at the beginning of a data range, the phase of the signal at the output of the clock signal divider 43 has the same possibility of being characterized by one of two fixed values, a phase recovery generator sensitive to the synchronization signal present on the line 41 is used to eliminate phase ambiguity. Phase 44 reset generator positions the clock signal divider ····

·....·

<img file="IT1112349B_D0038.tif" />

<img file="IT1112349B_D0039.tif" />

• fc.fc • fc.fc ·· *

<img file="IT1112349B_D0040.tif" />

<img file="IT1112349B_D0041.tif" />

• · · • fc.fc • fcfc · ···· ·· ·· * 4 * · * • 4 · _ 27 - τ'9 *

... ϊ · 'to a predetermined initial state, m accord- ·· ♦ ·:&gt;

do with. the points of passage for zero, from positi- * 'Z

4 • 4 • 4 4 · vo to negative or from negative to positive of the signal

·. · • * sinusoidal synchronization, in accordance with ··. ·· • 44 »· what will be evident to experts in the specific sector.

The reproduced equalized signal present on the line 42 is fed to a classic sampling and storage circuit, or maintenance 45 which is timed by the output derived from the divider circuit of the clock signals 43. The sampling and storage circuit 45 samples the reproduced signal, at the beginning of each nanosecond interval, recurrently, at the frequency of the ternary pulses representative of the data, as previously described and in accordance with what is illustrated in Figure 4 . The threshold decision devices contained in the sampling and storage circuits 45 provide, by means of the use of high-speed voltage comparators containing bistable circuits or latiphes '', a constant sequence of IfJ signals at the admissible output levels, in agreement with the partial response configuration used. In the ♦ tf! ····· · ··· ·· ♦ version. .... .

^4

4 4

In »C.-SS® currently considered, the previously described ternary registration input translates into obtaining five reading levels of integer value represented · in specific node · by 0» +1 and +2 in the project of reporting to partial response in Class IV. The sequence 1FJ which must de • * 4 4 · • 4&gt; 44 · *

4 * 44 ·. ·· * * z · ···· «

• « ••4« • ·· · ···· • · • · ·

<img file="IT1112349B_D0042.tif" />

4« • · ··

<img file="IT1112349B_D0043.tif" />

rivare by the waveforms shown in the figure

4, has been considered as "sampled and stored composite signal". It must be underlined that the five possible sampled and memorized levels are converted, through said threshold devices, into equivalent binary logical representations in parallel, for the further processing before the final recovery of the basic rare biA data. thresholds previously described, are commonly used in communication systems involving the use of sampled data. In this case, the thresholds are established in a known way, so as to obtain a minimum probability that the noise and the distOr—

The system may cause a discordance of the sequence of the integer values ​​^ F ^ with the result desired by the signaling system design, so as to eliminate the formation of errors in the recovery of the bits

<img file="IT1112349B_D0044.tif" />

»00

- 39 The output derived from the sampling and storage circuit 45 is fed, as input, to a digital decoder circuit 46 represented in figure 2. For a general L-level system, the decoder 46 recreates the L-ari data originally recorded, from the desired sequence of reading the (2L-1) partial response levels, by means of the elementary module arithmetic rules specified in the US patent if

n. 3,492,578, In the illustrated ternary version e.

described, the conversion performed by the decoder.

it's the following:

(Ρ 1 mod 3 V where F ^ represents the element at full level in the sequence IfJ, that is to say intended as corresponding to the impulse originally recorded, representative of the data indicated in B_n in figure 4, the ternary output deriving from the digital decoder 46, resulting from the example waveforms shown in figure 4 and presenting the modular series of values ​​(-1, 0, +1) has been represented, in figure 4, as '' ternary output data ''. The ternary output data indicated in figure 4 are identical to the ternary input provided by the converter 12 and represented in figure 4, owia4 »0« 0 • »0 0 ·« *

0 00 4

0·* *·· *00*

0

<img file="IT1112349B_D0045.tif" />

•4 • · · 0 0 4 0

0 * • 0* *····

00·· • 4 * 0 · ·

0· 0 0 0 4

00000

4440

0 0 4 ····

<img file="IT1112349B_D0046.tif" />

mind in the absence of noise and prohibitive distortions. A module encoder of the type used in accordance with the invention is well known in the field of partial response and, in the version currently considered, provides the ternary exit pulses according to a binary representation in parallel pairs.

• 9 ·« *

<img file="IT1112349B_D0047.tif" />

« 9 « · ··· · ·· • » · • ··* 4

<img file="IT1112349B_D0048.tif" />

The output derived from the module decoder

46, of the digital type, is fed, as input, to an L-ary / binary converter 47 which, in the version currently considered, has been concretized as a ternary / binary converter. The converter 47 operates in reverse mode to the converter 12 in the recovery of the fundamental binary data from the storage system, originally provided by the binary data source 11 for the storage of the bits therein. It should be noted that the.

circuit complex is included in the converter 47: to operate on sequential pairs of the parallel binary representations representative of the ternary data in such a way as to obtain a schematization in corresponding triples of data of the source that provides the fundamental binary data, using a function of inverse schematization with respect to to that adopted in converter 12.

···· • · ··&gt;. »·. • ♦ s * ···· ll is

With specific reference to figure 7, it can be noted that the details of the classic derivation equalizer 36 used in the practical embodiment according to the invention have been reported therein. The input to the equalizer 36, coming from the automatic gain control amplifier 36, is fed, through a regulation delay circuit 50, to a multiplier 51 sensitive to the signal representative of the coefficient, indicated in D. The input is also fed to a branch circuit 52 which provides a linear phase approximation to the second branch of the same, with respect to time. The output from the branch circuit 52 is fed, through a regulation or compensation delay circuit 53, to a multiplier 54 which multiplies the signal by the coefficient E. The output from the bypass circuit 52 is also fed, as an input, to a branch circuit 55, similar to the circuit

52, which provides the second derivative, with respect to time, of the signal output from the circuit 52.

The output from the branch circuit is fed, as input, to a multiply f'AtnT'P Sfi the fin Al n T'ir'P'Vfi. rnmo πτ'λγιτ'ί ri ccvrinAr-i in_ ···.

• · · .·· ····

UFFISÌ ^⁶*^'¹

Ina45 ·? »⁸⁰ gresso, the representative signal of the coefficient

F. The outputs derived from the multipliers 51, 54 and are fed, as inputs, to an adder circuit 57 which supplies the output signal of the device. In accordance with what has been previously described, the transfer function of the derivation equalizer 36 approximates the almost linear expression between phase and frequency in radians ω:

• · · ·; ·· »·· ···· • z · ···· •« ···· • · • · «

<img file="IT1112349B_D0049.tif" />

• · «····,« fcfc ··

OUTPUT _, 2 "2,

INPUT SA) + F (w) in which the coefficients D, E and F are adjusted in accordance with what previously described.

The regulation, or compensation delay circuits 50 and 53 are included to equalize the transmission delays in the three paths of the branch equalizer 36, so as to prevent phase distortions that could otherwise derive from the presence of such unequal delays .

With reference to Figure 8, it can be noted that the details of the transverse filter 37 used in the preferred version of the invention have been illustrated. The input derived from the branch equalizer 36 is fed to a delay line 60 equipped with intermediate sockets, this line having a spacing between the fc

- 43 between the sockets taken equal to τ, said spacing / being related to the interval of the fundamental pulses and to the bandwidth of the system, in a known way. The j

I ι

outputs present at the sockets of the j

The delay line 60 is supplied, as inputs, to the respective multipliers 61 whose second inputs receive the signals representative of the coefficients C ", C" ..... C "corresponding to the sockets,

GIN of delay line 60. The outputs derived from j

The multipliers 61 are combined in a circuit i

The sum 62 which provides the output of the device!

I sitive. The transverse filter 37 has the following transfer function.

• fcfc • fc · · · fcfc ··· ···.

• * • ···· • fcfc • · ·

N

<img file="IT1112349B_D0050.tif" />

EXIT

ENTRANCE

The k = or transverse filter 37 is tuned by adjusting the values ​​of the coefficients Οθ-Ο ^ wherein, in the example currently considered, N = 2 ^ et = 25 nanoseconds. The approximate tuning of the filter can be obtained by adjusting the coefficients for the maximum apertures in a traditionally displayed eye configuration, formed

on the reproduced equalized signal, schematized in figure 4. The fine tuning can be obtained by minimizing the errors observed in a

- 44 recovered pneudo-random sequence. It should be noted that in the quality »versions of the invention

I ι

commercial, the cross filter 37 can be replaced by a. classic fixed type LC filter or any other type of filter designed in such a way as to have the same characteristics.

Referring in detail to figures 9 and 10, it can be seen that in the figure _t the details of the portion of the device which creates the formats 14 for the digital signals, used in the generation of the sequence of interleaved bi-pulses, have been illustrated while the figure shows the timing with respect to the circuit schematized in Figure 9. The bi-pulse generator circuit shown in figure 9, includes a time division multiplexer 70 having two inputs marked by the references 0 and 1_ selectively connected to the output thereof, in accordance with a control signal ^ coming from a multiplexer (MUX) through a line 41. The position control signal;

MUX represented in Figure 10, indicates that inputs 0 and 1_ are alternately connected to the output once during each pulse interval which, in the version previously described, ··· ....

· -.Ata.

- — * ··* ·· · * «· ♦ &gt; · ,··

JC GfeéflSJ

- 45 has a duration of 50 nanoseconds. Therefore, input 0 is connected to the output during the first 25 nanoseconds of each interval while input 1_ is connected to the output during the second 25 nanoseconds of the corresponding interval. The output derived from the precoder 13 is fed to the input 0 and, in compliance;

as previously described, this signal comprises a pair of binary digits representing the ternary impulses. It must be underlined that from the precoder 13 to the input 0 of the multiplexer 70 binary digits are fed in parallel in a number required to represent the number of levels in accordance with the L-ario system implemented. These digits in parallel can be considered as forming a word

W. The application of sequential words by the precoder 13 has been illustrated as "output of the precoder" in Figure 10.

The word formed by parallel binary digits, output from the multiplexer 70, is fed, to a displacement register formed by three stages in parallel, schematized by block 72, said register being synchronized at a frequency of 2 / T, wherein T represents the word interval, if »4 * 4 • 4 4« 4 4 · 44: · 4 4 4 4. ·: ·

4*4 *

.· • 4 4 .·· ···· * 0 ♦ ··· *···* ·

004

- as shown in Figure 10, this interval, in the version currently considered, being equal to 50 nanoseconds. Therefore, the displacement register 72 introduces a delay of 3/2 word intervals, as required to input the bi-pulses, in accordance with what is shown in Figure 10. The delayed parallel binary digits, coming from the shift register 72, are fed to an inverter of code 73 ί which provides the inverse word “w which is appropriate for providing the reverse polarity portion of the bi-pulse, as previously described. The output of the code inverter 73 is fed to the input terminal 1 of the multiplexer 70, so as to supply the appropriate command signals to the digital / analog converter 15 for

I the generation of the sequence of interleaved bi-pulses, in accordance with what is illustrated and described above.

Therefore V provides the code it causes

I the generation, by the digital / analog converter 15, of the appropriate level and polarity pulse for the first portion of the bi-pulse intercalated with the code inverter 73 and the displacement director 72, of the delay type which provide » «« »» • ···· · ** · • 4 • * · ♦ ·· · »00

0 0 &gt;·* 4 0

<img file="IT1112349B_D0051.tif" />

···

MI »·· β

- 47, χ · χ

W_k starting from W ^. The word W_k translates to an impulse of the same level but with opposite polarity with respect to that caused by W ^. This, together with the switching carried out by the multiplexer, generates the desired intercalation, in accordance with what previously described and illustrated with reference to Figure 4.

The binary output word intercalated, as shown in Figure 10, is fed to a second time division multiplexer 74 which combines the configuration of the synchronization signals and the configuration of the preamble signals with the word considered, in accordance with what previously described. The configurations of the synchronization signals and of the preamble signals are obtained by means of the generators 75 and 76, respectively, operating in the traditional way. The output of the multiplexer 74 is fed, as input, to the digital / analog converter 15, for the generation of the signals

The recording analogs in accordance with what previously described, for recording on the corresponding recording medium.

From the above description of the preferred version of the invention, it will be apparent

44444 • · 4 44

44 « .**· »··· • · • 44 :·· • 4&gt; .4·

4444

4 44 ····, ·· ng. · * ···

- 48 evident that by means of the use of a se—.

polarization signal to linearize the magnetic interface and through the incorporation of a partial response signal in the linearized recording channel, of the magnetic type, it is possible to obtain a 3-4 times increase in reliable linear recording, towards the systems at the same time, commercial, currently available on the market. The system proposed by the invention has far surpassed i technology

earlier, towards obtaining the final theoretical registration densities in the field of communications. It was believed that the linear impulsive densities obtained could not be reliably obtained using a partial response format, in the absence of the linearization polarization. Furthermore, the inclusion of the linearization polarization allows the use of a multi-level recording and this allows to increase.

further the binary information stored in each recording impulse cell. Polarization also provides simultaneous erasure of old data and allows the use of a pilot tone for time recovery and gain control purposes »The linear system • · • ·· · · * ···· · · · · ··· · · · 5 · ··· · · * ··· · * • * ·· »

• · _* ·"·· · ·· * " · to ·

..... ....

·· · · · · · · · · • ··· I • fcfc fcfc _UFFicujeREiifi.Trt

Ina. C. '&amp; FepRJ ***** · • fcfc · also facilitates the pre-phase in the recording section, so as to allow an increase in the signal / noise ratio. The partial response format uses an amplitude threshold detection to replace a classic peak pulse detection.

A primary purpose, in the design phase, in recording on discs is to increase the surface density of storing the information bits. This can be achieved by increasing the number of tracks in a radial direction or by increasing the linear density of the bits for each track, or • •• fc * • fcfcfc »• fcfcfc * • ·

<img file="IT1112349B_D0052.tif" />

• fcfcfc ·· ♦ · • · · • fcfcfc increasing both the number of tracks and the linear density j. For any given radial density of the tracks it becomes increasingly difficult to improve the linear density due to the inherent characteristics of hysteresis and demagnetization, of a non-linear character associated with a saturation-type registration, with a traditional structure, which entail the onset of certain problems represented, for example, by a downstream shift of the bits. The invention in question has made it possible to significantly overcome this disadvantage associated with the prior technology;

ι causing a linearization of the main sources. · ··· • fcfcfcfc »··«

- 50 distortion poles, these sources being able to be corrected by equalization, rather than allowing the maintenance of this non-linear character due to the fact that the latter negative characteristic could only be partially compensated. The magnetic interface used in the practical embodiment of the invention allows to provide, in general, a linear density of kbpi in an instrumentation of the prior technology. Using the invention described above, operating with this interface, it is possible to obtain linear densities of the bits comprised approximately, between 12 and 16 kbpi.

Although the present invention is applicable to any recording system on magnetic media, it must be noted that in order to obtain a reliable improved performance as regards the cost per bit and the number of bits per unit of track length, the The invention can be advantageously incorporated into the technology relating to magnetic disk archives currently preferred for mass storage for computers.

The invention has been described in terms of the use of a sequence of bi-pulses • * ·· · ···.

fc • fcfc · ₄ • ·«· ^] !

.·»* ···* ·&gt;'♦·&gt;

»·♦ ;^η3

- 51 intercalated but »nevertheless, it must be noted that it is also possible to adopt other data decompositions in other types of waveforms with multiple intercalated pulses, in order to obtain the spectral advantages or other advantages allowed by the invention» in accordance than previously indicated.

Furthermore, it must be emphasized that the precoded or multilevel binary data can be provided directly by the digital / analog converter 15 for recording within the confines of the partial response format.

The fact that the. signal formed by a sequence of multiple intercalated pulses, should possess properties that are desirable from the spectral and / or temporal point of view, possibly including a frequency spectral zero for.

insertion of a pilot tone.

It is possible to realize an alternative version of the invention by using a limited code ί as length with respect to the signal representative of the data that has been recorded, instead of using [a pilot tone for the recovery of the timing and Der the control of the gain. The properties of the limits limited in length are well known in the prior art, such systems being used * 444i&gt;

tl ect * * -4 4 4 * QTj »4 4 _to 4

4 4 4 * * *»4 • 4 9 ··*· • 4 44 · *44* · 4

4 4 4 4 »·· • ••·

<img file="IT1112349B_D0053.tif" />

- 52 ti to allow reliable recovery of the timing and reliable measurements of the channel gain, regardless of the original data provided by the source. The alternative version may or may not use interleaved bi-pulses, or the like, while remaining within the confines of the partial response format.

Therefore, it should be emphasized that the invention is not limited to the use of a bi-pulse sequence. In general, a multi-pulse intercalation sequence can be used for the advantages that the waveforms offer, or none of these waveforms can be used. For the specific purposes of the claims set out in the appendix, the signals circulating in the system have been described as represented by pulses. In accordance with what has been previously described, it can be observed that in the various portions of the system, these pulses can form appropriate words, typically conceived as consisting of parallel groups of binary digits representative of the pulse amplitudes.

Although the preferred version of the invention has been exemplified as using a partial Class IV response, in accordance with what

Infl.

· · · ·

<img file="IT1112349B_D0054.tif" />

···

- 53 described in U.S. Pat. 3,388,330 e

No. 3,492,578, it must be emphasized that the invention is also valid for versions incorporating a generalized partial response constituting a technique known in the specific sector also as adapted transmission or as equalization of the module transmitter. The generalized partial response uses a nature precoder.

more general than the precoder used in the

The previously described version, an example of;

this decoder having been described in Robert Priee's memory entitled 'Nonlinearly

Feedback-Equalized PAM vs. Capacity, for Noisy Filter

Channels '', Proceedings of the 1972 International

Conference on Communications, pages 2212-2217 »IEEE publication. However, in other respects, the format remains similar to that of the preferred version previously described. Alternatively, if a precoder is not used, the decision circuit of a resulting version should include a classic feedback equalizer of the feedback type in accordance with what is described, for example, in the previously defined IEEE memory. It must be emphasized that the precoder, the decision circuit and the deco • * · ··

<img file="IT1112349B_D0055.tif" />

•4 4 .·· • •4 «44

The dificer proposed by the invention can include elements variously suitable for a partial response, for a generalized partial response or for an adapted transmission, or for a: decision-making equalization of the feedback type or combinations of these techniques known in the specific sector.

Without departing from the spirit and purpose of the invention, it should be noted that the recording of partial response data described in the course of this discussion is characterized by the existence of a "null" at the Nyquist frequency of the amplitude spectrum of the impulsive components of the data read in the.

Waveform representing the reproduced equalized signals represented in Figure 4. The Nyquist frequency is mathematically defined as equal to half the pulse frequency and the possession, by the read pulses, representative of the data, of a null value, i.e. a spectral zero at this frequency makes it possible to operate the system at the minimum bandwidth according to Nyquist, in accordance with what is defined in figure 5 of the preferred version previously described. The width l · * '* 4 ····

- 55! ii rt ii'ti · '.

··· ·* * • · »

..... Ύ / • · fc • · · · *; · ·· «*« of minimum band according to Nyquist is defined as equal to half the frequency of the pulses but, however, in the specific technology it is known that the signal to Partial response defined during the present discussion can also be realized by adopting bandwidths slightly or substantially higher than the minimum of

Nyquist or by adhering to the condition of presence of a zero at the frequency of

Nyquist. Consequently, other versions involving an excess bandwidth, according to the invention, must be considered as falling within the spirit and purpose of the invention described in the course of the present discussion and

I in which an operation with polarization and / or with the use of multiple intercalation pulses is adopted.

It must be emphasized that the partial response format described in the course of this discussion does not necessarily have to have a spectral null at zero frequency although it possesses the spectral null previously indicated at the frequency of

Nyquist. For example, the spectrum may include a sinusoidally shaped lobe having an amplitude j ·· • · • · · • 00 *

thick

- 56 maximum at zero frequency, this lobe being generally shaped subsequently, in accordance with what previously described. In particular, the spectrum can include more than one spectral null within the minimum bandwidth according to Nyquist and can have a slightly or substantially excess bandwidth, in accordance with what previously indicated. Therefore, the terms' 'sinusoidal lobe *' or '' sinusoidally shaped lobe ''

I intend to include configurations of the type generally described during the present discussion, with specific reference to Figure 5, as well as configurations having a maximum amplitude at the frequency equal to zero.

Although the invention has been described in its preferred versions, it must be underlined that the words that have been adopted must be considered descriptive rather than limiting while it must be underlined that various modifications and variations can be made to the invention in question , within the spirit of the claims set out in the appendix, without departing from the true spirit and the true purpose of the invention, in its aspects * · 0 · 0 • «00 0 ~ ·· ♦ ···· wider • · * 0