High accuracy lane timer (temp variation, weather, etc)

My kids race Soap Box Derby (not Pinewood) and we have been using the same relay based lane timers since the late 70's. These timers work quite well and provide an accuracy down to 0.001 and are generally fairly accurate over a varied temperature range (we race in just about anything from 30F with Snow to 110F with rain)

Before anyone points this out, I've looked at the Pinwood Derby stuff at http://www.miscjunk.org/mj/pg_pdt.html already. Several times. I've also noticed a lot of posts about the inaccuracy of the oscillator over varied temperature ranges.

My goals are as follows: Differential time between 2-3 lanes (2 here on our track, but some other tracks like Akron use 3 lanes) Overall elapsed time for a given lane, sampled at 4 intervals (top, 1/3 down, 2/3 down, bottom) for each lane Extremely accurate event given varied temperatures Added bonus would be speed at each interval, should be a simple calculation of length of car / time spent blocking I would think

Why am I worried about the lane samples and accuracy? Simple, we like to try different things; positions on the track, torque, camber, toe, etc... under different circumstances; wet track, dry track, rain, snow, sleet, etc... Since it isn't uncommon (happens at least once a race) to see differences of 0.001 accuracy matters when trying to tune your car (or driver).

Any guidance would be greatly appreciated. I realize I'll probably have to change out the existing oscillator with a TCXO or something similar. Options? What works, what doesn't?

Oh, and we use a low power laser (think cheap laser pointer) from the center of the track to the outsides to determine each lane. I'd love to find a way to do this with a single source across both lanes, or dual sources from either side, but I'm not sure that is actually possible since you can't really tell what car passed and exited when with such a setup.

PS: I should mention that the distance from start to finish is anywhere from 900-1100ft using Cat5 cable to carry the signal/power to all sensors. As I know that line distance makes a difference we also ensure that all lines are the same length with minimum coil ;)

Thanks in advance, - Jeremy

I did Soap Box races in my younger years too but we didn't have multiple lanes, just one lane and time measurement.

Anyway, you get a much greater accuracy (by the factor 1000) if you use an Arduino with a crystal oscillator instead of the ceramic resonator that current UNO/Mega2560 have. Take a Duemillanove or some Seeeduino for example and you'll get an accuracy that's more than enough for these races. Although the crystal may differ by some ppms (part per million) at different temperatures that doesn't make a difference at these distances.

Always nice to hear from/meet other racers. Now everything is down to a kit, so minor tweaks and driver experience makes or breaks your runs. I'm guessing in those days you still built your own car?

Since I have one lying around I thought of using the SparkFun Arduino Pro 328 (https://www.sparkfun.com/products/11113)

It has a XTal Resonator setup that is supposed to provide 16Mhz, but the details on the actual components is a little sparse. Think this would still be within acceptable tolerances?

Like the earlier suggest, I agree that you should use a crystal rather than a resonator for your arduino (make your own), but I would also suggest that you encase the arduino/crystal along with a heater (incandescent bulb) and temperature feedback, and that you include code to maintain a temperature of say 115-120F (your max plus 5-10 degrees)

Keeping it temperature compensated is the best way to ensure timing accuracy.

Another possibility, would be to accurately measure the clock frequency at multiple temperatures, and simply include temperature measurements and compensate in software for inaccuracies..

It has a XTal Resonator setup that is supposed to provide 16Mhz, but the details on the actual components is a little sparse. Think this would still be within acceptable tolerances?

No, it's accuracy is in the permilles so it's not suitable for your needs (it's similar to an UNO). A Seeeduino (http://www.seeedstudio.com/depot/Seeeduino-V30-Atmega-328P-p-669.html) is a quite modern Duemillanove clone which still offers a crystal oscillator.

Always nice to hear from/meet other racers.

I wouldn't call me a racer, that's more than 35 years ago.

I'm guessing in those days you still built your own car?

Yes, just the wheels were standardized and provided by the race officials for all teams. I don't remember but I think there was also some weight limit to ensure some safety standards.