logo    Trolley Scan(Pty) Ltd

The Supertag® story

Supertag ® is the name given to a multiple article scanning low cost RFID technology developed to scan the contents of a trolley(cart) in a supermarket.

The registered trademark Supertag was owned by CSIR (South Africa) in 1994 and might since have been sold to another party. There is no connection between Trolley Scan and the ownership of the trademark Supertag.

Supertag technology was invented by Mike Marsh and his team while working for the South African Government funded research group named CSIR. Mike Marsh went on to start the private company Trolley Scan but was not allowed to use the Supertag technology, and so he invented the Trolleyponder technology.

This page gives some of the history of the development of Supertag while Mike Marsh worked at CSIR.

The CSIR (formerley Council for Scientific and Industrial Research) was established by the South African Government in 1946 as an organisation to deal with South African-specific problems. In the 1980's there were 35 divisions covering issues such as Buildings, Water, Roads, Personnel, Mining, Telecommunications, Aeronautics, Computing, Mathematics, Chemical Engineering, Marine Sciences, Material Science, etc. In the 1980's the staff complement was approximately 5000 engineers, scientists and technicians.

A customer pushing the famous trolley through the scanner at
Menlo Park Pick 'n Pay.

In January 1994 the world was surprised to discover that a South African development was likely to change the future retailing and logistic systems.

State of knowledge

In order to get reasonable operating ranges of a few meters, the radio frequency system would need to operate in the UHF band. Although there were already radio frequency transponders on the market (passive - no battery), these were limited to magnetic coupling modes which meant very short range. Such devices were used to identify race horses (Trovan) to stop them being swapped for handicap purposes. In UHF a US company called Amtech had developed a transponder system to identify containers on a railway which had gained wide acceptance, and there were developments to use transponders for toll roads.

There had been a lot of interest in being able to identify items at a distance, but software outstripped hardware. A major software company in South Africa had approached CSIR about the availability of a transponder to be placed on assets at a bank so that the assets could be tracked if leaving the building, but no suitable device was available at that stage, even though they had spent a lot on the software development.

Prior to November 1990, the South African Government Research laboratories of the CSIR in the Division of Microelectronics and Communications Technology, had been approached by a motor manufacture to find out if it was possible to have a very low cost RF transponder to be used for just-in-time manufacture, the restrictions being that it would need to withstand the heat of the paint shop while being very cheap.

The challenge with developing a low cost transponder system was to make a system that had relatively simple radio frequency components. The most common transponder type system in use at that stage was the remote control unit used to open driveway gates and garage doors. Here the transponder generates an operating frequency, modulates the carrier wave with the data to be sent which is then transmitted through an antenna. The receiver receives the radio signal, and has a complex system to detect and decode the incoming signal. This involves the receiver generating a slightly different frequency from the carrier wave, mixing it with the incoming signal to generate an intermediate frequency which is amplified and narrow band filtered, and then passed through a detector to decode the data code. This means that the transponders are complex and expensive.

At CSIR we had been involved in the use of doppler radar speed tracking devices, which operate on the principle that the reader outputs a single frequency which hits the object being measured and is reflected to the reader. The movement of the vehicle causes the return frequency to be shifted due to the doppler principles and so the same frequency is used in the reader to transmit and decode the incoming signal, in most cases just needing a mixing diode and simple amplifier and decoder to have the answer at baseband frequencies. Our approach was to design a transponder that effectively behaved as if it was a moving item by shifting the return frequency, which we could do very simply. y

Our solution for the motor vehicle problem led to the development of a transponder that repeatedly broadcast its identity and used a "backscatter modulation" method. Such a transponder was built using discrete components. Extensive testing was carried out to detect such a transponder passing at speed mounted on a car as well as when the car was stationary.

This concept was later modified to provide a design for detecting the identity of miners working deep in the South African goldmines. Here the rebroadcast was delayed by a random backoff timer based on the principle that when identifying a few tags in a zone at a time, all the tags would have some uncluttered transmission time when they were not jammed by other tags. This principle is valid if there are only a limited number of sources in the beam at a time.

None of these concepts could handle many transponders in the reading zone at the same time, mainly because the tags all respond on a single frequency, many talk at the same time as they are free-running, and because the return signal is so weak compared to the energising signal they do not have onboard receivers which would be able to hear if other tags are talking at any instant.


The Supertag protocol
Supertag® protocol was developed by accident by Mike Marsh. While attending a management strategy meeting of the Division of Microelectronics and Communications Technology in Pretoria at a place called "Aarbeidsaamheid" in November 1990, he suddenly realised how a simple protocol could be implemented that would allow any number of tags to be read.

If you want to read a lot of transponders at the same time, since all the transponders use a single communications channel, you need some method to manage the communications from the different transponders to clear up the communications channel so that the transponders can be identified in interference free time slots. This invention did it so successfully that we could read up to 500 transponders in a zone at read speeds of up to 70 per second.

Another benefit of the protocol was that the actual value of the identity number in the transponder was not used by the protocol. This allowed us to have many transponders with identical numbers and we would be able to count how many of those items with the same number were in our reading zone at one time. This meant that this development would be suitable for use in retail stores, where the same product item would have a common identity number, and we would be able to count all those items accurately. This was in fact what happens with the barcode number on retail items, where the number gives country, factory and product range. Supertag was a true barcode replacement technology.

When we invented the protocol, we were not looking for a solution for supermarkets, but were dealing with cars and miners.
With these attributes it seemed that this technology would be suitable for use in retail stores where it could be used to scan the contents of a supermarket trolley (cart) without the need for unpacking. We found that we could further modify the protocol to allow the transponder to be fitted with an anti-shoplift feature called EAS. This would allow the contents of the trolley to be read when passed through a scanner, and after reading turn on the anti-shoplift feature so that when read by a scanner at a doorway it would detect any items that were passing which had not been properly processed.

To build such a device would mean that many technical challenges would need to be addressed. We gave the project the name "Post Toasties" after the US cereal product range.

Our team set about to build this device and reader.

January 1991 the first patents were filed.

When filing a patent one needs to prepare to defend such a patent if challenged. Our lawyer required us to design a simple explanation of how the protocol worked in order to explain to a potential judge who might be involved in such a case, and a person who was most probably not an expert on electronic matters.
For our model we used:

Assume you have a school hall filled with students who are all deaf and blind. The person on the stage doing the counting is also blind. The students are given an identifier to call out:- say a colour. So say 30% of the students are told to call out "BLUE", while another 40% are told to call out "RED" and the rest are given other colours to call out. The students are told to call out their colour, wait some random time, and then call out the colour again................ (The rest of the model I will hold over).

An RFID reader system would comprise of a reader module attached to a computer system, and a transponder attached to the item being identified. Ideally the transponder would be very simple in mass manufacture, and would receive its power and would get its operating RF frequency for communication from the energising field from the reader. The reader would comprise of a transmitter that generates an energising frequency at the RF operating frequency, which would be radiated towards the transponder, and a receiver to receive the returned signal from the transponder and decode it into a computer compatible signal.

At that time we had major technology challenges. We had access to two different processes for the manufacture of an IC to implement the cheap transponder. There was a digital CMOS technology which could be used for low power digital circuits comprising of flip-flops and gates- this was available through the SAMES factory in Pretoria. We had access to a foundry at CSIR with the Plessey 1A process that could make analogue parts such as diodes and transistors and high power switches. However the manufacturing processes in the different foundries were so different that they would contaminate each other if brought together during manufacture. We needed high speed low loss diodes such as Schottky diodes to convert the incoming energy efficiently from the antenna into DC voltage to operate the logic circuitry in the transponder. We needed capacitors to be implimented in integrated circuits to store the energy extracted from the RF signal to power the digital circuitry. We needed very low power digital circuity in the logic implementation because energy availability would be very low at a distance from the reader. We needed a way of programming data into the transponder so that we could store data relating to the item to which it was attached.


After about 18 months of development, the prototype had developed to such a stage that it could be demonstrated inhouse. The management of the Division of Microlelectronics and Communications Technology saw potential in the system and we started to look for commercial partners. CSIR comprised of a staff that was focused on research and development and were not experts at commercialisation, so we sought a partner. Initially we dealt with SAIDCOR, a section of CSIR that had taken earlier inventions to commercial partners for industrialisation. This group handled legal contracts and progress monitoring but did not do further development.

A similar operation to SAIDCOR of the UK Government had become privatised and formed into a company called British Technology Group which listed on the UK stock exchange. They had experience in commercialising research technologies and when they saw the potential of Supertag we started discussions. This was in November 1992. In October 1993 CSIR signed an agreemnet with BTG to commercialise the technology. The agreement meant that CSIR had to absorb all the sunk costs and we split the future income from licence fees and royalties.

About this time we brought the specific project to the attention of the president of CSIR, Brian Clarke. As president he had to look after the issues relating to a staff of about 5000 and did not get involved in specific projects. He wanted assurance from people in industry that the project had merit, and so he approached the Managing Director of Pick n Pay, one of the largest retailers in South Africa of groceries, to evaluate the potential. Pick n Pay sent the commercial Director, Chris Davies, to see the project and he started to become part of the project and give us advice on issues. Pick n Pay had a large number of superstores, some with as many as 70 checkout counters in a row, and yet at the end of the month the buying habits of the South African public were such that they arrived at counters with two trolleys full of goods that then took a long time to process. Supertag could make a real difference.

When we came to the press launch, Chris Davies arranged for us to use the Menlo Park Pick n Pay near CSIR, and he provided us with the first two plastic trolleys available in South Africa. These had been sent to Pick n Pay for evaluation as all trolleys at that stage were metal. We needed a non-metal trolley as we were to scan the goods while they were in the trolley so needed the radio signals to penetrate the sides of the trolley. We had made a wooden trolley but now used the new plastic trolleys.

In December/January in the Southern Hemisphere, all is quiet as most of the staff are on long holidays. Our programme had decided that our press launch demonstration was ready and that we should launch early in January. We approached the Vice-President of CSIR, Dr Daan Toerien, to get authorisation and support from the public relations office for a launch in early January. We then tried to find a news team that would be interested in covering the launch, but were sent to commercial film companies who wanted a lot of money to make a film of the launch, more than our programme had allocated in its budget. Eventually we persuaded a news organisation to do the film (Reuters), and we agreed with them that it would be embargoed for three days to allow us to get together press packs etc.

The demo was done at Menlo Park Pick n Pay and a news film was made, as well as us taking stills. That afternoon everything blew open as the news team broke the embargo and put out the film. Notes from my digital diary show that Reuters estimated that the film was seen by 300 million viewers on 650 broadcast stations (including BBC, SKY, CNN). Then came the printed media wanting interviews, photos and demonstrations.

As a result of the coverage we had contact from about 1500 companies around the world who saw the demo and wanted to be involved. We recorded their phone numbers and addresses, but were not able to give them the time they needed as we were a small research organisation. We had numerous visits from the Chiefs of South African manufacturing industries wanting to be involved. However industrialising a new technology requires a certain compatibility in understanding between the manufacturers and researchers, and this project was not in a stage to hand over to a company that was not equipped to industrialise a new technology.

Over the next four months up to 3 technical teams visited per day to study the technology.


The Famous trolley
The Trolley scanner
Inventor with demo tag
Part of the development team at the Supermarket

In April 1994 Mike Marsh left the development group to start his own small engineering consulting company after nearly 20 years of service at CSIR. Winnie Mandela had been appointed as the new Deputy Minister of Technology by the new South African Government. A four year restraint of trade was placed on his skills so as not to compete with CSIR.

BTG took over negiotiations with prospective licensees and announced in May 1994 that ICL had taken an option to a license. In October 1994 the third license was signed.
In February 1995 they announced the fourth license option to major industrial company. Later they licenced the technology to a US company for US$10 million.


People involved in developing the electronic identification system

Members of the TEAM that made RFID possible and gave the world a demonstration of a trolley
Mining Systems Programme
Mike MarshInventor, development team leader, programme manager and negotiator
Andre LenarcikDeveloper of RF systems initially for the vehicle project, and then for the trolley project. Designer of antenna systems, and the RF components of the transponder.
Clinton van ZylDesigner of the microprocessor and receiver components for the reader. Involved in the initial testing of the vehicle system and later of the trolley system.
Andries van SchalkwykBuilt the phase lock loop and receiver components and assembled many tags and the demo system.
Martin OosthuizenBuilt the microprocessor components of the receiver, many tags and assembled the demo system.
Paul MekisichBuilt the transmitters together with the notch filters for rejecting interference.
Martin WalkerBuilt the initial discrete component transponder.
Werner BremerModelled radiation patterns for inside metal container.
Trish RitterWrote the software used for the trolley demo till.
Paw Paw CabritaBuilt the initial wooden trolley system, and the mechanical components of the scanning system that was used in the Pick n Pay demo.
Jack MarxMade the mechanical components of the receivers and power supplies of the vehicle and Supertag system.
... and about another five members of the programme 
IC Design Programme
Hendrik van Eeden Converted the original discrete design to a low voltage MOS process in a foundry in Europe. Wrote a software simulator to show initial simulation regarding error rate.
Lourence CronjeAssisted in conversion of discrete design to first multichip implimentation.
Roy AtkinsTook over non-digital design aspects and preparation of specifications after key staff left.
Microelectronic Applications Programme
Mark Carson, Gideon Gouws, Mario Marais Collectively developed a method of attaching integrated circuits to cheap antenna substrates in volume applications.
CSIR Information Services
Murray Dell Comprehensive mathematical model of the performance of a group of tags

Further reading
Details of the patents
The effect of operating frequency on RFID performance
Pictures of early development models

Images from Press launch

Trolley Scan (Pty) Ltd © / info@trolleyscan.com