Behind the Scene

All teams competing in the 1997 International Pedal Prix were aware that there was a failure to accurately record vehicle passes from about 2pm to 7pm on Saturday. As manager of the lap counting team I feel that it is necessary to explain the nature of the failure and the efforts that were made, in very difficult circumstances, to solve the problem and to provide accurate race results for 19 hours of the 24 hour event.

The lap counting team arrived at the track on Friday afternoon at 3pm.

It took until 6.30pm to set up the lap counting interrogator unit, the recording computers, the Novell File server and LAN network wired to a 10 baseT hub.

Testing of the equipment and the cards that were to be fitted to cars begun at 7pm. It was necessary at this time, to check that each of the 126 cards did respond after passing the interrogator and that this response matched the name, number and class of the vehicle. This data had been set up by prior to the race using the program Setup.

As cards were being checked it was obvious to me and the team, that at times, cards were not being recorded accurately.

The team began to look for the source of the problem. The following is an accurate record of attempts to solve this problem.

• Both the 240 volt lead and the serial data lead from the interrogator were removed. The 3 serial leads were re soldered and replaced. Both ends of the 240 volt cable was re soldered.
  This did not solve an intermittent problem

• For the first time the team had used a Macintosh 520c Powerbook as an independent backup computer for the collection of data.
  It was thought that there may be problems with the serial input of this machine compared to the previously used SE 30. It was thought that there may be an impedance problem and conflict between the two serial ports.
  I asked three team members, Robert Ardill, Grantley Roper & Julian Scott to drive back to Adelaide at 10.30 to pick up another SE 30 from Aberfoyle Park High School. I asked them to save all the data from the hard disc in the SE 30 to the school network, remove the data show interface, insert the ethernet card, reformat the hard disc, install a suitable system and finally install the ethernet driver software. This they did, and they arrived back at Murray Bridge at 2 am on Saturday morning.
  The two SE 30's were then setup and further testing resumed. The team experienced a degree of success be were not confident that the problem had been solved.
  The last team member went to bed at 4am.

• At 6.30am I rang Alf Grasso of Integrated Silicon Design at home and I explained the problem and our attempted solutions. I thought at this time that the problem was related to the interrogator unit.
  Alf dressed immediately and drove to Adelaide were he and a technician built another interrogator unit out of spare parts. This together with the testing of the unit took some time.
  Alf arrived at Murray Bridge with the Backup unit at about 10.30am on Saturday.
  New data serial leads had to be made so that the Macs could communicated with the new interrogator and configure it to receive data.
  This all took time and at 12 noon the backup unit was in place alongside the old unit.

• At this stage the original interrogator unit seemed to be functioning correctly and from 8am to 12 pm the lap counting team had fitted cards to all vehicles and checked that they were reading correctly.

• At 1.10pm Saturday, the race started. The interrogator unit seemed fine and a physical check as cars passed revealed that all cars were recorded.

• At 2.00pm the interrogator failed. Cars were not being recorded in the normal fashion, but the backup unit was receiving raw serial data into a terminal communications package called Z-Term.
  The nature of this data was different to the nature of the serial data collected on the failed unit because of the differences in firmware between the two Interrogator units. This meant that the existing software could not be used to record cars and generate reports.
  At least the collection of data adverted the need to go to a manual system. This would have been an absolute disaster for all concerned. Teams were informed of the problems and told that data was being collected and car passes recorded.

• By 2.30pm Alf Grasso had driven the failed interrogator unit in pieces.
  He suspected a component failure in a small RS 232 serial chip.
  Desperate attempts made locally in Murray Bridge failed to locate the part and at 3.00pm Alf returned to his workshop in Adelaide.
  A chip was located, installed and the unit successfully tested.
  By the time Alf arrived back at the racetrack is was 6.45pm.
  The unit was replaced and the system restored at about 7.00pm Saturday.

• The task now remained to match the data collected during the failure with the correct car numbers. Remember that this data was in a different format and did not match the original data collected.
  To their credit Alf Grasso and Jason Curl developed an algorithm that would convert the data. Jason then sat done and developed a program in MS Visual Basic that would read the car file with the old card number, convert it using the algorithm and then match the collected data to the car. In this way the laps for each car from 2 to 7 could be determined.
  Because it was raw data, no individual lap times were recorded.
  Jason finished this task at about 11.00 pm and the results were published.

• It was some time before we realized that laps recorded during the failure by the backup unit were not accurate. It was obvious that some cars had not been counted while for others there were laps added that were not completed. The added laps for some cars was a result of the merging of data files that were saved in stages, during the failure.
  The lap counting team and ultimately the Race committee, in the interest of fairness to all teams, decided to disregard the results during the failure. This was communicated to teams.
  I must point out that while the majority of teams accepted this decision, a number of teams where extra laps added and these laps had not been completed, felt that they had been cheated. This certainly was not the case.

• The decision to disregard the laps during the failure was certainly correct. I hope that the above clarifies and justifies this and I hope that the teams who displayed considerable aggression towards the lap counting team can now show a level of appreciation.
  The Team was under intense pressure and their ability to perform in these circumstances is a credit to them. I sincerely thank them all for their efforts.

Special Thanks

The programming skills of Jason Curl, with the assistance of Zia Khoskho, must be recognised.
Thanks Jason and Zia.

I believe that the effort made be the team member Peter Kelly is also very significant.
Peter spent considerable time in the past two months, developing a new and significantly superior report program to generate race results. the results generated directly in HTML code are "Internet Ready".
This in fact means that next year the results could be published on the Internet throughout the race. It was through his efforts that the results were published so quickly after the race on the Internet.
The overall Pedal Prix Web Site is a credited to the team, particularly Peter Kelly.
Thanks Peter.

Finally I cannot speak too highly of the efforts made by Alf Grasso of Integrated Silicon Design.
Without his involvement there would have been no results
Thanks Alf to you and Integrated Silicon Design.

Brian Lee
Lap counting team manager

Back to results page