I am calculating force from acceleration which is output as a variable PWM.
More acceleration the longer the T1 is high.  The documentation for the sensor
says that fixed point math will be fine.  Per your description, I can just
multiply by 2 to essentially shift the decimal point over.  and then shift it
back when the calculation has concluded.

Do I do any thing with the remainder, or just don't worry about it since it
will be in the 1000ths position and I really don't need to worry about that for
my application?

Sean Breheny wrote:

> Hi John,
>
> Please give some more info. Are you using this in a high-level language
> (such as C)?
>
> I would suggest that you do one of two things:
>
> #1) Multiply all numbers by some constant (best for it to be a power of 2,
> like 1024) to begin with,do your math,then divide the result by the same
> constant. The quotient is the part to the left of the decimal,the remainder
> is the part to the right of it. I believe this is called fixed-point math.
>
> #2) Use floating point math routines.
>
> If you are using numbers with a small number of digits and the range
> doesn't vary much (i.e., you work with numbers from 0 to 200 and only need
> xxx.xx precision), use method #1. It is MUCH faster. You can use shifts to
> do the multiply and divide by the power of two at the beginning and end.
>
> If your number range is variable (i.e., you work from .0001 to 1 billion,
> and need x.xxxxx * 10 ^ e type numbers), use #2
>
> Sean
>
> At 04:10 PM 11/7/99 -0500, you wrote:
> >I have a formula K*(T1-Zacutal)/T2actual
> >
> >where K is a constant = 4*Bit Scale Factor (like 255 bits from 50% duty
> >cycle to 100%)
> >T1 is the high time of the PWM during normal operation
> >Zactual= Zcal*T2actual/T2cal.
> >T2actual is the PWM on time during normal operation.
> >Zcal is T1 during calibration.
> >
> >When I do the multiplication, division and subtraction of the above
> >numbers I begin to loose the fractional numbers, like 28.456 turns into
> >28.  This can be a problem especially at values near 0.  I always get 0
> >for my calculation when I should have something like .5.
> >
> >Does anyone know or have references to multiplication, division and
> >subtraction routines that do not loose the decimal.  I have seen
> >routines that give a remainder but then I don't know how to use the
> >remainder in the arithmetic operation.
> >
> >Thanks
> >John
> >
> |
> | Sean Breheny
> | Amateur Radio Callsign: KA3YXM
> | Electrical Engineering Student
> \--------------=----------------
> Save lives, please look at http://www.all.org
> Personal page: http://www.people.cornell.edu/pages/shb7
> mailto:shb7@cornell.edu ICQ #: 3329174