Optical Sensor Systems

Commercial timing systems
Post Reply
SuperDave

Optical Sensor Systems

Post by SuperDave »

If someone is seeing their system being susceptible to such light influences
This is exactly my point. If you could see it, you wouldn't need the timing in the first place.

It is not hard to imagine a video camera light illuminating the sensor at the same time that the car finishes and is blocking the timer's source light. The result is that the car still finishes and the timer still reports a time. But the time is wrong. The reported finish is different from the actual finish, undetectable by humans who assume that if the computer shows a time, the time must be correct. In the extreme it is of course possible for there to be no finish at all but that's unusual. It's more likely that the car's reported finish is an inch or two later than it actually was.

This error would not be there for a car that crossed without the video camera present. This error would be different if the video camera were held in a different position. For a forum that often deals with incredible nit picking when it comes to cars and their construction, this unquestioned acceptance of an enormous error source truly baffles me.

In some poorly designed circuits, a flash can over-illuminate the detector and the 'darkness' caused by the sudden loss of the flash illumination can cause the sensor to report an early finish.

Properly designed mechanical sensors don't have either of these problems. Even mechanical sensors that (as reported elsewhere in this forum) appear 'flimsy' or 'can bend' are far more reliable both because the potential error from the 'bend' is much smaller and because the software used with them runs all cars on all lanes and averages the times to find a winner. The result is that even if the sensor were 'bent' several inches (or even located several feet up the track), all cars would use that lane and sensor once and the error would disappear in the final calculation. This error elimination does not occur with optical sensors because the error is not constant.
User avatar
gpraceman
Site Admin
Site Admin
Posts: 4926
Joined: Fri Jun 20, 2003 12:46 am
Location: Highlands Ranch, CO
Contact:

Re: Has anyone done any of these track ideas?

Post by gpraceman »

SuperDave wrote:It is not hard to imagine a video camera light illuminating the sensor at the same time that the car finishes and is blocking the timer's source light. The result is that the car still finishes and the timer still reports a time. But the time is wrong. The reported finish is different from the actual finish, undetectable by humans who assume that if the computer shows a time, the time must be correct. In the extreme it is of course possible for there to be no finish at all but that's unusual. It's more likely that the car's reported finish is an inch or two later than it actually was.
I am by no means an expert on optical systems, but what I do know is that I've deliberately tried to fool timing systems by using flashes and bright lights and with the sensors recessed down inside the track, I have not been able to do so. I've even had times for the same car get a very tight standard deviation in doing so. To me, that means that light issues can be mitigated. Of course, you are welcome to disagree.
SuperDave wrote:Properly designed mechanical sensors don't have either of these problems. Even mechanical sensors that (as reported elsewhere in this forum) appear 'flimsy' or 'can bend' are far more reliable both because the potential error from the 'bend' is much smaller and because the software used with them runs all cars on all lanes and averages the times to find a winner. The result is that even if the sensor were 'bent' several inches (or even located several feet up the track), all cars would use that lane and sensor once and the error would disappear in the final calculation. This error elimination does not occur with optical sensors because the error is not constant.
It is obvious that you feel mechanical means of determining a finish are more accurate. Though, you seem to be alone in that. All other commercial timers that I have seen and worked with use optical sensors.

Personally, I do see a problem with mechanical sensor flags that do not stay straight up and down, as the order of finish can be affected. Many people choose to score using a point system (based on the finish order) and such a condition would be a problem. Some people also choose not to have racers run on each lane (for a variety of reasons) so the error in a bent flag would not be cancelled out. Your software seems to control these issues by only allowing times scoring and running cars in each lane. That is my preferred method of running a race, but that is not so for everyone.
Randy Lisano
Romans 5:8

Awana Grand Prix and Pinewood Derby racing - Where a child, an adult and a small block of wood combine for a lot of fun and memories.
docbar85
Journeyman
Journeyman
Posts: 30
Joined: Sat Mar 12, 2005 7:41 am
Location: N.C. , Nashville

Re: Optical Sensor Systems

Post by docbar85 »

The better sensors have a emitter and receiver that are tuned to each other, stray light has no effect on them.
SuperDave

Re: Optical Sensor Systems

Post by SuperDave »

The better sensors have a emitter and receiver that are tuned to each other

Generally true, more below.
stray light has no effect on them

Not necessarily.

Broadly speaking, there are three kinds of optical sensors for Pinewood use.

1. Simple sensors, like the 'affordable' system mentioned in other current posts, have sensors only, no emitters. They rely on room light as a source and rely on the car blocking that light. Changes in room lighting clearly affect these devices. As the other post mentioned, the user had to turn off one set of room lights in order to get results. At home, this is practical.

2. The most common arrangement is a per lane emitter facing a per lane detector. The light may be incandescent, laser, IR even UV, though most common is IR (infrared) mostly because the most common and least expensive detectors are most sensitive in the IR range. They are less sensitive to local lighting than 'simple' sensors but they are not immune. Burying them, tunneling them, etc. can help, but absent careful evaluation of the individual design, one should be very wary of accuracy claims. When the phone finally stops ringing in late March I hope to do some documented tests to show the depth (or as others might argue the lack of the depth) of this problem. As a former test engineer I hope to bring some expertise to the issue.

3. Much less common, in my experience unknown in the Pinewood Derby world, are what the previous writer calls 'tuned' . This is not 'tuned' in the sense of IR versus UV for example, but rather tuned in the sense that the emitter is pulsing rapidly and the detector is looking for those pulses at that frequency only. More specifically, the detector is looking for the absence of those pulses when the car crosses. Thus, theoretically, a steady light would not have the correct pulses and the detector would not be fooled. BUT if the steady light shines on the detector directly it can overload the sensing device (photodiode, etc.) causing its output to be steady (in saturation) and for all intents and purposes appear as a car crossing. In this case, burying the sensor can help greatly since the saturating light must then be much more directly in the path of the sensor. Careful (and more expensive) design of the sensing circuit, for example with automatic gain control, can also help stray light immunity. [This is why your TV remote works better in a dark room than a bright room.]

'Tuned' systems have two other not well understood problems in the timing, as opposed to remote control, world. First, the pulses must be very fast, on the order of several times the display resolution of the timing system. You can't have 1 millisecond pulses detect a 1 millisecond crossing without adding significant error. Second, the reaction time of the 'missing pulses detector' (a complex and more costly circuit than a simple photodiode/comparator) must be consistent, again much less than the timing system's display resolution. Commercially available industrial systems such as the SunX sensors (used by at least one provider of Soap Box Derby Timers), not only don't pulse fast enough, but have large and unspecified reaction inconsistency. This is not a big deal if you are counting bottles on a ketchup line. It is a big deal if you're trying to separate cars at the finish line.

Photo detection is simple, but like much of life, the devil is in the details.
Post Reply