-------------------- Begin Original Message --------------------
Message text written by "Rick Brown"
"Ok, I'll bite. For all you tech types here are some of the sources of
timing errors. I'm sure this list can come up with others:
1. Variations on what part of the car trips the lights.
2. Variable trip delays between start lights and stop lights.
3. Capture window granularity associated with polling for light trips.
4. Variation in time based frequencies, eg, a crystal oscillator is more
stable than a cheaper ceramic resonator and crystals themselves come in
various grades.
5. Drift and variation caused by voltage and temperature fluctuations over
the course of the event.
I suspect that our current system is much more accurate than the old Radio
Shack encoded, white light timing head based system that actually had a
relay in each head with all the variation attendant from a mechanical relay
closing. I bet that system was only accurate to around a tenth of a second
and bet our current system is accurate to around a hundredth of a second or
a smidgen better. For those of who think it's accurate to a thousandths of
a second; dream on, and I don't mean Steven Tyler.
-- Rick"
-------------------- End Original Message --------------------
Rick, Please tell us about why the JAC timer is not accurate to .001 of a
second.
Long ago our EE guy, Bob McCullouch (spelling?) by name, praised the value
of a crystal oscillator to ensure accuracy.
Is the third digit at random?
McCulloch, a founding partner in a Silicon Valley electronic biz,
20 years ago started his own business manufacturing hard drives capable of
storing massive amounts of memory for the time. He used a system called
"Run Length Limited" which was originally developed by IBM for main frames.
About a year later Apple and others routinely installed hard drives with
even more memory.
Somebody has a McCulloch timer in the corner of his garage. He
built several units in conjunction with Terry McHenry who is now racing a
Huffaker-built Fiero in SFR races.
--John Kelly
|