|at the 2006 World Champs|
This time the Transitrace timing system did NOT perform flawlessly!
A dual system was used without manual backup, as allowed by new the rules. The system was lent by the CMBL club of Landres, France, through Jean-Paul Perret. A new version of the TT software was used, V2.00. (Soon to be released.)
On repeated instances both the primary and secondary systems failed to detect the passing model. When this occurred, it often lasted for the entire flight.
To begin with, the organizers were not active in setting up the system so that the official practice day could be used to verify the system and get the operators trained, as was previously agreed. Only late in the afternoon a tent, desks and electrical power became available, and operators in attendance. Therefore, the first contest day started with only very limited experience with the system.
Already flight 5 on the first day, 19th July, was a miss by both the primary and secondary systems. The contest was interrupted, and I was called in to analyze the situation. It turned out that on two occasions inside the timed flight two laps in a row had been missed by the primary system, and most of the flight was missed by the secondary.
The software is able to cope with a missed lap (which could always happen when a model is flying low), but not two in a row. A study of the log file showed that both the lap 0 and lap 9 passings were registed, so by taking the difference of these time stamps, the correct speed could actually be calculated manually. This procedure was not accepted by the jury, though. At the same time, two private Transitrace systems were in use, and they had recorded the flight without problems.
It must be underlined that the operators (and software + PC computers) had no role in the missed flight, but the sensors simply failed to detect the model.
Given the background of very little training on the previous day, the FAI jury decided that the first round should be interrupted, and postponed to the next day. Those who already had posted results were given the option to have their result stand, or start all over on the next day. Only one competitor chose to let his result stand. So, the remaining time of the first day was used for practice, and the operators got a fair amount of training. The system seemed to work fine.
On the next day, 20th July, it all started smoothly, but again a complete flight failed to get recorded. Yet again, two private systems had worked fine. It was suspected that the official sensors, positioned around 40% of a lap counter-clockwise to the private ones and about 1.5 metre apart, could be disturbed by the sunlight. (Although the sensor light openings were actually in shade.) The decision was made to move the sensors around 30% of a lap clockwise. After this, the round went on, with the affected competitor given a repeated first attempt. No more problems occurred.
When I took a closer look at the French sensors, I found a mistake in their assembly. Inside there is a "blinders" sheet which limits the light path. This was not properly attached, and could get misaligned in transport, causing the field of view to be reduced. This probably had not happened, but I replaced the sensors with two of my own for the remainder of the contest.
The second contest day, 21st July, went fine, but in the third round, on 23rd July, three more flight got unregistered, and the competitors were given repeated attempts. Clearly, all attempts at finding a solution had been unsuccessful.
Only an educated guess can be made to understand the cause of the failed registrations. Something in the conditions at Valladolid seems to be problematic. The system works on the principle that a passing model gives a dark contrast to the sky. With the requirement of an aperture large enough to catch both low and high flying models, and the need to have the sensor positioned well outside of the landing area, the model will occupy no more than 0.5 % of the field of view. At 650 metres altitude, the sky is darker than at sea level, and in the strong sunlight there could be solar reflexes in the model which counteract the drop in light caused by the model obscuring the sky. Clearly the sensor position with respect to the sun is more critical than expected for the prevailing conditions. It seems impossible to predict which sensor position is the most favourable.
If a dual system is to provide the intended redundancy, the sensors must be separated by a distance larger than the 1.5 metres used here. This will reduce the likelihood that an unfavourable sun and/or sky condition will affect both sensors at the same time. A separation by 10 to 15 degrees of the flight circle sounds reasonable, corresponding to 4-6 metres at a 22.5 metre radius.
This solution was never tried, as I felt it would be too big a change in the middle of the contest. If the system had been used on the practice day, it is possible that the problems could have been spotted, and such a solution been found for the contest.
Transitrace Log Files19th July, Primary | 19th July, Secondary ( Only flight 2, by 27 Benavent, is an official Round 1 flight. The flights following flight 4 are practice.)
20th July, Primary | 20th July, Secondary ( Round 1.)
21st July, Primary | 21st July, Secondary ( Round 2.)
23rd July, Primary | 23rd July, Secondary ( Round 3.)
Official F2A Results (PDF file)