Split times

[Chief Geek]

Sensor versions 1 and 2 were used in 2012. V1 & V2 used the same basic (salvaged) components with V2 being a simple refinement that was less fussy and needing slightly less TX power.

Sensor V3 are built with new parts (many are free samples) and are designed so I can build batches of 1, 5, 10, etc. quantities.

Design-wise, all the goals in my 2/1/13 post have been met. After the other components are designed, I'll work on circuit boards and quantities.

Range:
V1 & V2: 25 ft
V3: 40+ ft.
Sensors can be farther from edge of road and, hopefully, distract fewer drivers.

Response time (minimum beam-blockage-time that reliably registers):
V1 & V2: 25 milliseconds
V3: 1.2 msec
Should be able to see #36's and #3's helmets below 100 mph.

Precision (max variation between the fastest and slowest observed response times)
V1 & V2: 20 msec
V3: 0.2 msec
Proper speed traps are now easy.

Cost:
V1 & V2: Don't know the "real" costs (much salvage, few $, many hours). All one-offs.
V3: Both TX & RX sides are much cheaper and made from off-the-shelf parts. Power/batter consumption is much lower (4x lower on TX side)

Interesting technicalities (interesting to me, at least):
- IR detectors only detect IR light that's pulsed at a certain frequency. This is why sunlight and temperature doen't impact them much. 38kHz is very common for remote controls & 52 kHz seems to be the standard for 3D TV glasses.
- Most IR detectors also have an automatic gain adjustment function. This is great for receiving short bursts of "data" from TV remote controls but terrible for "continuous-detection" jobs like car sensors. Non-auto-gain sensors are available and intended for things like garage-door-opener-safety-sensors, safety light curtains, etc. Discovering this difference made many sensor improvements much easier.
- Model numbers for very powerful IR LEDs and their driver circuits are easily found on forums for adult, survivalist-style, outdoor-laser-tag. (Me either.) Builders of home-brew outdoor-laser-tag "sniper rifles" are very well versed on these subjects. (The interweb certainly is a very large and very strange place.)

[KevinGale]

Paper!? How about the grease pencil and plexi-board leanded aganst a leanto?

Yeah, I remember that! And a big piece of cardboard and magic makers at least once I remember. That didn't work so well since erasing was impossible. That was the time that made me think, "I can write some software to handle this"

[walterclark]

Hi Paul. Thanks for the update...and thanks for doing this. Its pretty cool.

Now if there were just a way to modulate the data onto the wire - without having to replace ours with something that could reliably carry DSL, even with CAP, the lower speed upstream band requires 160KHz. Not that ours has been tested and cannot, its just that it has so many spots where the impedance must be a mess due to damage and repairs, I cant envision it being able to for any distance. Even a "good" wire is only expected to work up to 18,000' - the official ADSL upper limit....And we would need to add DSL filters at each headset box.

[Mopar 151w2]

We've talked about new wire. IF we find a source. Should we consider 3 or 4 conductor, in light of all the data we want to throw around?

[walterclark]

We've talked about new wire. IF we find a source. Should we consider 3 or 4 conductor, in light of all the data we want to throw around?

I think a big issue with more than 2 conductors is bulk. 2 twisted pair would permit a dedicated checkpoint line an a dedicated data line but such a cable requires a jacket to keep it together so it would come out about the size of CATV / cable TV coax. 2+ miles of that would require a HUGE reel.

Not to mention other issues - those common with many cables: inflexibility (and shape memory) making putting it down and retrieving it a lot of work, difficulty tapping and repairing, and if the conductor itself isnt designed for repeated flexing, fatigue failure. - the big issue with 2 adjacent twisted pair and analog comm: crosstalk, shielding can correct this but it makes it just that harder to handle and repair.

The problem we have with finding wire is that we cant use just anything that carries a signal. For instance the CATx cables. They are everywhere and I am sure we could buy a bunch of it cheap if we wanted. But the stuff is stiff as all get out and the wires are small and fairly brittle because they are made to be installed in a cableway, wired to wall outlets and left alone. Maybe the perfect wire is that designed for studio floor and field microphone use. Very flexible - both the jackets and the conductors, so they are easy to handle. The biggest issues are cost and cable diameter. There are very small gauge mic cables but they would probably not survive the first attempt by a car to stretch them in an off.

Along those lines the Mogami W2552 would be an excellent cable in many ways. It is very flexible and though it is 25ga, the spiral shield provides a lot of protection against damage. Its overall diameter is 3/16" so it is smaller than the typical coax but still pretty large and the cost is $0.40 per foot on 650' reels. So maybe 20 reels would do us and that would cost about $6000. Mogami makes a similar cable that is much smaller with 35ga wire but I am sure it is too delicate for our use. The cheapest small diameter (1/8" or less) new wire I have found is Belden 8451 and that is $0.020 per foot. I am not sure about the flexibility/memory because it uses a foil/polyester shield and they tend to be stiff. It is listed a mic cable but suggests console wiring for a use where stiff is not such a problem.

Surplus wire is something that I would have too actually look at and handle to judge. Hardly any that are available in large quantities and low prices (like $0.010 per foot) come with any useful information so I wouldnt buy it without touching it first.

[sachilles]

What about synchronized timers, rally style? Each broken beam records a time, calculate the difference. Probably too much work on the back end, but could work for splits. Just spit ballin'. Love what you guys have been doing, not implying that more needs to be done or anything like that.
My other thought is self service. Have you seen those ir lap timers. You put an emitter at a start line, driver puts a timer with receiver, every time you pass an emitter gives you a time. Put an emitter at every check, and let the driver view his own splits real time en route. They are pretty cheap. Something like this.http://www.sumomoto.com/ezlap_lap_timer.htm

[KevinGale]

As I posted in another thread there some possible big increases in safety if we can find some way to send data. It can be really low bandwidth. Serial port baud rates would even work (like RS485).

The safety increases come in if you can attach multiple split time sensors to the wire. Control and/or Finish can then see where the cars are on the hill and can be warned when the cars get too close together. But this requires a way to send the data in something near real time. A delay of a few seconds would be OK but a delay of a minute makes it useless.

I've even thought of just converting the wire to be data only. So the headsets would be sending voice as data. That's a big change and I'm not up on the technical issues involved. Although that might allow use of only two strands.

[Chief Geek]

Over the past year, I've done some web-searching about how to get data over a long but simple analog audio line. One thing that may work, if someone that understands analog electronics can help me, is a very low frequency modem. Low frequency, like below human hearing range.

Human hearing bottoms out at 20 - 40 Hz, depending on your age. While I'm sure he could, I doubt Walter made the headset boxes to work down that low. Having spent many hours on the wire, I'll bet it's freq. range isn't too different from a telephone. According to wikipedia, the bottom end of an analog telephone's range is 300 Hz.

What if the "sender" generated a carrier freq of 200 Hz and modulated it up and down 20 Hz for '0's & '1's? It would have to make it's freq transitions somewhat carefully to prevent creating "noise" with the audible range, maybe by transitioning only at zero-crossings. Maybe, this would be "invisible" to the headset users.

I've already gotten my arduino chips to generate freqencies in this range, and I've been tinkering with a FFT "spectrum analyzer" library that runs on these $4 chips.

Even if all of that worked, the data rate would be really slow, like 1 char per second. Does anyone think a data rate that low would be useful?

Paul

[walterclark]

There are no hipass filters on the headsets or mics as they are not needed.

To deal with something as close in to the voice range as a modulated 200Hz signal would require something like active cancellation. Ordinary filters could not be made steep enough to do much good.

[Chief Geek]

Walter

Might it work if our carrier was farther from audible? 20 Hz? Even if it's possible, the bit rate would be truly tragic. Hmm. On to the next stupid idea.

Paul