Manchester encoding: opposing definitions resolved

by Roger Forster


| Introduction | Investigation | Historical background | Discussion | Conclusion | Acknowledgment | References |


Manchester encoding is widely used to provide clock and data information simultaneously via a single connection. There are two opposing—and incompatible—definitions of Manchester encoding where the rising edge in the centre of the data bit indicates either a logic '1' or alternatively a logic '0'. This paper resolves the differences and identifies the originator of the method.

Manchester encoding is named after the University of Manchester, where the first recorded use of it occurred in the late 1940's. It is a digital phase modulation encoding method which has the following advantages over NRZ (non return to zero) signals such as those used in conventional TTL/CMOS logic. These are:

The clock can be extracted from the data at the receiving data terminal equipment (DTE) by using a digital phase locked loop (DPLL). This makes the method suitable for use on single-core systems such as optical fibre and coaxial cable. In current engineering systems it is most commonly used on Ethernet and token ring local area networks (LANs). It does have two minor disadvantages:

A literature search for Manchester encoding produces two opposing factions as to how it should be generated.

Faction 1, centred on authors such as Stallings[1] and Halsall[2], states that the encoding scheme is based on the modulo 2 addition of the data and clock signals, using an exclusive-or gate. This method causes the output data to change state on the positive edge of the clock. Faction 2, with authors such as Tanenbaum[3] and Sanderson[4], states that the scheme is based on the modulo 2 addition of the data and clock but where the output data changes on the negative edge of the clock.

In each case there is always a change of the output data state, to facilitate easy bit synchronisation of the data, by providing a level change in the middle of the data bit. For Faction 1 this means that for a logic '1' this change is in a positive direction and for a logic '0' in a negative direction, see Fig. 1a. For Faction 2 it is vice versa: the change in the centre of the bit is in a positive direction for a logic '0' and in a negative direction for a logic '1', see Fig. 1b.

Fig. 1 Manchester encoding: (a) faction 1 definition; (b) faction 2 definition

Fig. 1 Manchester encoding: (a) faction 1 definition; (b) faction 2 definition

Investigation

These opposing definitions can cause a problem to the conscientious student who reads around the subject as they will find this conflicting information. This is because it is obvious that a Manchester-encoded data stream generated by a DTE using the first method will not be received successfully by a DTE that decodes using the rules of the second method, as the received data will be inverted. A modification to Manchester encoding, called differential Manchester encoding, overcomes any data inversion caused by the transmission path but requires a more complex encoding and decoding process than conventional Manchester encoding.

This problem of the two opposing definitions was brought to the attention of the author in late 1999 by a student comparing the textbook recommended for their course unit with the author's lecture notes. Upon investigation it was noted that Stallings stated that his method, although the opposite of Tanenbaum's, was in accordance with the IEEE 802 and other recognised standards. From communications with Stallings and Tanenbaum it was determined that the originator of the method was a postgraduate student at the University of Manchester during the early 1950's but the definitive reference to Manchester encoding was not identified at that time. A literature search using the Internet and on-line databases yielded references that were split between the two methods[5, 6]. The IEEE 802.4 specification (section 1.7.1) quite clearly states that a high-to-low transition represents a logic '0' and a low-to-high transition represents a logic '1'.

A single reference was found that pointed towards what might have been the original paper[7]. A search of listings of UK theses for PhDs taken at Manchester University in the early 1950's also failed to identify conclusively the originator. The author of the original reference was finally identified as a result of a letter published in IEE Review[8] requesting help from the IEE membership. Numerous replies identified Dr. G. E. Thomas as the original author. In fact the first reply coincidentally was from Dr. Thomas himself, almost two weeks before the author received his own copy of the magazine.

Historical background

Thomas graduated at Manchester University in 1948. He then became a postgraduate student and finally a member of staff there, working under Professor F. C. Williams on early computers, in particular the Manchester Mark 1 computer.

Williams introduced Thomas to phase modulation for digital encoding, which Thomas believes that Williams had used during World War II at the Telecommunication Research Establishment (TRE). Thomas described the phase modulation technique, as used for writing to and reading from the high-speed magnetic backing store of the Manchester Mark 1 computer, in a number of papers[9, 10], his MSc dissertation and finally in his PhD thesis[11]. A drum provided the main store for the Manchester Mark 1 computer, which used a CRT store as the random access memory (RAM) device.

Fig. 2 F. C. Williams (courtesy University of Manchester)

Fig. 2 F. C. Williams (courtesy University of Manchester)

Hence, although Thomas did not invent the system his are the original references and the first published practical implementation as far as the author can determine. The scheme adopted for the Manchester Mark 1 computer drum was for a negative-going edge in the middle of the bit[9, 10].

A phase modulation system was chosen in preference to NRZ or RZ (return to zero) schemes due to the balanced nature of its write currents—NRZ and RZ would have given unbalanced write currents. In the Manchester Mark 1 computer a logic '1' was represented by a negative-going pulse and a logic '0' by a positive-going pulse. Thomas and other members of the Manchester Mark 1 team have been unable to shed any light on why that scheme was used as opposed to the inverse method used by the IEEE 802 standards.

Fig. 3 Dr. G. E. Thomas (courtesy Dr. Thomas)

Fig. 3 Dr. G. E. Thomas (courtesy Dr. Thomas)

A history of the Manchester Mark 1 computer can be found at the Web site of Manchester University's Department of Computer Science[12]. Cunningham[13] has written about Professor Williams and Thomas has published an article on the drum[14]. The work at Manchester was subject to numerous patents at the time, including GB707637 and GB707634, which cover the phase modulation scheme. These patents and others can be viewed at the Patent Office Web site http://www.patent.gov.uk/dbservices/index.html using the esp@cenet system under 'Patent Services'.

Discussion

The author's initial thoughts as to why the two methods should be in existence focused on the concept of the mark and space definitions used on telegraph and early data communication systems versus that of conventional logic levels. In mark/space terminology a mark—a logic '1'—has traditionally been represented by a negative voltage and a space by a positive voltage; this is the representation for instance in the EIA RS232E or the ITU(T) V24/V28 specifications. This would give rise to the opposite method of generation when considering conventional logic where a logic '1' is a positive voltage and a logic '0' is 0 volts.

Web sites on Manchester encoding, such as that of Optimized Engineering[15, 16], and a Philips Semiconductors application note[17] do not show the actual voltage levels, which was the author's first clue to the conundrum. If we look at the IEEE 802.3 or IEEE 802.4 standards then it is clear that the author's initial thoughts are probably correct, however only members of the original IEEE 802 committee would be able to shed further light on the topic.

Conclusion

From the literature search it is most likely that the primary difference between the authors in the two factions concerns conventional logic versus that of mark/space logic. A logic '1' is always represented by a transition in the middle of the data bit from the voltage level representing a logic '1' to the voltage level representing a logic '0'. Hence the two waveforms shown in Fig. 4 represent the definitive status of Manchester encoding.

Fig. 4 Definitive definition of Manchester encoding

Fig. 4 Definitive definition of Manchester encoding

Acknowledgment

The author wishes to thank Dr. Thomas and the many other correspondents who replied to his letter in IEE Review[8].

References

  1. STALLINGS, W.: 'Data and computer networks' (Prentice Hall, 1999 6th edn.)
  2. HALSALL, F.: 'Data communication, computer networks and open systems' (Addison Wesley, 1996, 4th edn.)
  3. TANENB A. S.: 'Computer networks' (Prentice Hall, 1996, 3rd edn.)
  4. SANDERSON, P.: 'Encoding'. See http://www.cs.smsu.edu/~pete/csc465/notes/encoding.html, February 1998
  5. KUROSE, J. F., and ROSS, K. W.: 'Ethernet'. See http://gaia.cs.umass.edu/kurose/ethernet/ethernet.htm, 1999
  6. 'Introduction to the Philips audio protocol'. See http://tvs.twi.tudelft.nl/cases/audio.html, Delft University of Technology
  7. BROOKER, R. A.: 'The programming strategy used with the Manchester University Mark 1 computer', Proc. IEE, 1956, 103B, suppl., pp.151–157
  8. FORSTER, R.: 'Made in Manchester', IEE Review, March 2000, p.42
  9. THOMAS, G. E.: 'Magnetic storage'. 1st Cambridge Computer Conf., 1949
  10. WILLIAMS, F. C., et al.: 'Universal high-speed digital computers: a magnetic store', Proc. IEE, 1952, 99, Pt. II, p.94
  11. THOMAS, G. E.: 'The design and construction of an electronic digital computer'. Manchester University, 1954
  12. 'History of the Department of Computer Science'. See http://www.cs.man.ac.uk/Visitor_subweb/history.php3
  13. CUNNINGHAM, M. J.: 'F. C. Williams', Eng. Sci. Educ. J., April 1994, 3, (2), pp.55–64
  14. THOMAS, G. E.: 'Recollections of "Magnetic drum storage for the Mark 1 (1948-1951"'. See http://gethomas.members.beeb.net/Manchester/manchester_drums.html
  15. 'Signal encoding in Ethernet/802.3'. See http://www.optimized.com/COMPENDI/EN-Pers.htm (Optimized Engineering Corporation, 1999)
  16. 'Manchester signal encoding'. See http://www.optimized.com/COMPENDI/EN-SigEn.htm (Optimized Engineering Corporation, 1999)
  17. 'AN070—Verilog implementation of a Manchester encoder/decoder in Philips CPLDs'. Philips Semiconductors application note, 1997

| Introduction | Investigation | Historical background | Discussion | Conclusion | Acknowledgment | References |


IEE: 2000


This article also appears in Engineering and Science Engineering Journal, 2000, 9, (6), p.278-280

The author is a Senior Lecturer in the Faculty of Technology, Southampton Institute, East Park Terrace, Southampton, SO14 0YN, UK. He is an IEE Member.