andy pugh wrote: > > > I agree with your disagreement. The problem you describe is likely to > particularly acute in cases where someone has a 1:3 pulley ratio and > only bothers to type in the scale factor to as many significant > figures as they feel necessary. > > Nope, doesn't really make any difference. There are numbers that can be exactly represented in floating point. These are combinations of power of two fractions, such as 1/2, 1/4, 1/8, etc. But, there are many more values that have no EXACT representation, such a the 1/3 you mention. This is not a great problem for a one-time computation, the error is quite small, depending on the number of mantissa bits in that FP format. BUT, there will be error EVERY TIME the computation is performed. If you do this at the servo rate of 1 KHz, you are summing this tiny error a thousand times a second, it won't take long at all for this to accumulate to a large and totally unpredictable magnitude. In other words, one day the errors may nearly all cancel out, the next day, when the machine sits in a different position for a few minutes, the error could become vast.
Jon ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
