Babar,

Answer: No, you don't need to take every measurement... it would be too hard to get precise measurements anyway because you'd have to rely upon someone else's speedometer and there is no telling how well they did their design. :)

What you do is open a spreadsheet and model the values in a table for 0kmph to 99kmph. Its well worth the time to reason that formula out even in trial and error compared to the cost and time to sample live data.

I did this a few days ago in about 15 minutes for my 27 inch wheel bicycle in non-metric measurements. The advantage of a spreadsheet for doing tables is that you can always improve it before production. For example:

My bicycle has 27 inch diameter tires on it. I don't know the specifics of bicycles now, but during the course of such a project, I'm have to find out wether that is the outside rim measurement of the wheel, or just wht it relates to. To increase accuracy, I may need to boost the 27 inch figure to match the average properly inflated tread contact diameter.

I could do this observationally by havings someone get on the bicycle and do a few measurements and average them. I'd check the results against a second technique of measuring the length of one or several revolutions of the loaded bicycle tire across a pavement (chalk line mark and ten revolutions in a straight line would be good). The latter technique would actually be preferred.

If it turned out that 28.435 inches were the actual tread to tread diameter, I'd just replace it into the spreadsheet cell that held the 27 inch value. Immediately my entire table is recomputed at floating point accuracy and then rounded into my 8bit or 16bit form of the actual table in assembly code. I'd do a cut and paste of the table into my assembly code, compile and test the improved version. :)

I'll write some later on the topic of the spreadsheet development. E-mail me what spreadsheet you can read.

-aka Jay C. Box