Originally Posted by
dennisa
considering the cateye uses a 4-bit (2^4 =16) microcontroller I'm surprised it's even that accurate.
So in theory over a mile you can only resolve 330ft increments (5280/16) , which is +/- 6.25% error. This also assumes there is no errors associated with the rims and tires.
I'm not privy to the inner thoughts of my cateye, but I would imagine it isn't actually counting in miles. My guess is that it's counting in revolutions and converting to total miles at the very end of the processs, without discarding it's wheel revolution count. So yes, 300ft increments in displaying how far you've gone at any given time, but not a cumulative error of 6.25%, i.e. the farther your trip distance the smaller the percent error.
Anyway, I cast my vote for it being an error in tire circumference calculation. That would scale linearly with trip distance.