Verifying and diagnosing scoring machines

[edew]

Now that the new timing chips (or boxes) are soon to come into play, I have some questions for the armorers concerning how scoring machines are tested.

Yesterday at a local competition, there were several instances in saber where rather normal simultaneous hits would result in a single light (consistently for one side) or delayed lighting up of the two lights (red, then green) consistently. Well, it wasn't completely consistent, but when the action was "simultaneous attack" but the scoring machine had red-then-green several times, I was disconcerting. As the referee, I felt I had to overrule the timing that came up on the scoring machine because the actions didn't corrorborate with the way the lights switched on.

So the question is, how can one go through a complete diagnostic check of the scoring machines to make sure that they're working, especially if there might be a low power source coming out of the transformer or if the machines have heated up over the course of the day? Maybe there's high resistance on the cables. Maybe the rust on the blades might add resistance. Is there a diagnostic machine (preferably hooked up to a computer) that can run through a series of tests to make sure that the machine is up to spec?

Most of the other parts, reels, floor cords, body cords, epee/foil tips etc., can be easily checked by anyone with rudimentary knowledge of electrical circuitry. But how does one check things that should be working in the 5-400ms ranges with possibly very low signal-to-noise ratios? I'd hate to be the person on the side where there's no enough juice to switch on the light and have the referee make a justified, but incorrect, call.

[DHCJr]

Eric asks a good question and I wish I had a good answer. At most competitions, there is no time to check and even major competitions, I have been at, they don’t even pretend to have the latest timings, i.e. 2-weapon boxes. Many competitions also have received enough machines from various sources to maybe have 2 extra boxes. You work with what you have. I do have a tester I built, but it is manual. Also, at most competitions, I sometimes don’t have time to check the boxes. Many times the promises machines show up the morning of the competition and we are working on controlling the fencer’s equipment.

The FIE is just now getting into the 20th century. I know up until ’97 and it may have been later, they tested the machines with a breadboard arrangement very similar to an old-style telephone switchboard. They would literally move patch cables around to change timing and resistance, and then hit a button for each test.

This tester, I built about 15 years ago. For me, I don’t have the time. It would be nice to have, but because this is very much a niche market, the Armorers themselves make the only good testers. There are no good commercially available testers of any type.

[fencerX]

[...]delayed lighting up of the two lights (red, then green) consistently. Well, it wasn't completely consistent, but when the action was "simultaneous attack" but the scoring machine had red-then-green several times, I was disconcerting. As the referee, I felt I had to overrule the timing that came up on the scoring machine because the actions didn't corrorborate with the way the lights switched on.

Red-then-green?

I don't know about scoring boxes being tested, but I do know that when large populations of humans have been tested, the population's detection rate is higher and reaction time is invariably faster to red lights than green lights of the same luminous intensity. Research has shown it takes a green light of close to three times the luminous intensity of a red light before reaction times and miss rates are comparable.

In US the standards for traffic signals specify minimum intensities and require green signals have 2.0 times the luminous intensity of the red signals and the yellow signals have 4.6 times the luminous intensity of the red.

[DieterS]

According to the FIE rules (Annex B to the Material Rules, A.1.a.6) regarding scoring machines,

"No priority must be indicated between a hit registered againt one competitor and a hit registered against his opponent."

As a result, we (Eigertek) have taken pains to ensure that the lights turn off simultaneously.

Unfortunately, there is another commercial scoring machine that turns the lights off in the order that they were turned on. Several customers have asked me whether we could make the Eclipse also do this to help them referee. I explained to them that the rules forbid this. Not convinced, they asked me to check with the FIE. I contacted Dan DeChaine of the SEMI commission for confirmation or clarification.

In response, Dan wrote me that

1. There is not mention in the rules anywhere that requires the sequential light behavior.
2. He has never seen an FIE conforming machine that exhibited the sequential light behavior.
3. He has never heard of any such behavior proposed to the FIE.

Regarding the testing of scoring machines, the last I heard, there are only two large FIE devices (known as a "borderaux" if my spelling is correct) in the world that can test the FIE conformity of scoring machines. We actually built one of our own to test the timings of the Eclipse and other scoring machines. The lack of availability and expense of building one is probably why the FIE requires that conforming T2005 (Leipzig) machines must have a sticker on them and that the timings cannot be changed.

Dieter
Eigertek

[keith]

If it is a question of differing resistances on each side then a standard multimeter should allow you to check - although I guess the resistance is likely to shift quite a bit due to the contacts in the reel as it is extended/retracted.

But I didn't think any of the changes would affect how the box senses hits (current/voltage in the circuit). It may be due to the shorter timings showing up bad lames that weren't a problem before - there is now less chance for a blade to sweep across a good spot on a bad lame before you're timed out. A bad spot in a reel could have the same effect by spiking the resistance for long enough to delay the hit registering - even a few milliseconds could have an effect given the shortening of the timings.

[JEC]

If a floor cable has a 1 ohm resistance while the other cable has 100 ohm resistance, wouldn't that introduce a slight delay systematically?

If a floor cable (or anything else before gets to the scoring box) has a capacitance that needs to be charged first before conducting current, wouldn't that introduce a slight delay systematically?

[EricS]

If a floor cable has a 1 ohm resistance while the other cable has 100 ohm resistance, wouldn't that introduce a slight delay systematically?

If a floor cable (or anything else before gets to the scoring box) has a capacitance that needs to be charged first before conducting current, wouldn't that introduce a slight delay systematically?

The parasitic capacitances in the floor cables, although higher than a coax cable, are compensated for by the output drivers in the scoring machine.

In response to the original post, I'd carefully check the floor cords, reels, and such before blaming the scoring machine.

If the total resistance in the electrical circuit from the B/C line of one side to the A line of the other side fluctuates around the threshold, it could cause similar behavior to what was described in the original post.

[TheArmourer]

Slightly OT from the original question - sorry.

The problem you're getting into there is that the resistance of bodywires and weapons fluctuates.
A bodywire can work (tho possibly eratically) when it's as high as 40ohms (in good condition it should be about 0.5) and it can easily creap up with a bit of sweat. And in my experience groundwires don't get over resistive nearly as often as fencers own kit.

[sabreman]

The machines should have the ability to make a real time or dump a diagnostice trace via some type of interface (RS-232 on a lap top pc). This is pretty much standard practice in the hw world. I work for Sony in the tape streamer engineering division and with all our products you can make a real time or dump a diagnostic trace. The dump trace would include fw revision and detailed internal fw status information. The real time trace would show error messages / warnings and user friendy status information (presented in away that non techies could understand what happened). If the machines have small LCD then this info could be displayed there too. If security is an issue then this feature can be password protected.