----- Original Message ----- From: "Mike Harrison" Subject: Re: [PIC] Floating Point on 16x > FP gives you RANGE, not Precision. Mike hit on something very important here. I think a lot of people are confused on this issue. The PIC is a microCONTROLLER. That means it is usually connected to some sort of real world device. In the physical world, rarely do things change over an order of magnitude or two, so in almost all cases 10 or 12 bits is adequate. Now, if you have to do some calculations, then it takes considerable thought on how to scale the intermediate values, but the situation where you can't do that scaling is extremely rare. Indeed, when doing control, going through that scaling exercise is important, because it helps cement in your mind the significance of various factors. You not only gain by avoiding big, cumbersome, floating point routines, but you gain by paying attention to the important variables and not getting hung up on insignificant variables. All that being said, floating point is available, and it isn't rocket science, either. Like most thing in the PIC, (or any embedded system for that matter), it pays to really understand what is happening, though. There are a number of floating point packages out there, each with very significant issues. I would suggest that one should roll one's own FP routines the first time out, in order to be able to make an informed judgement of the merits of the various alternatives. I'm a big fan of re-use, but unless that re-use is done in an informed manner, the results can be pretty unpleasant. So yes, if you need to solve a system of differential equations or something on that order, then go out and understand the floating point options. But if your application is remotely normal, why on earth would you subject yourself to the size, performance and grief of floating point? --McD -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist