Hi,

Larry Bradley wrote:
<snip>
>For example, 728.25. I get the 728 part, and then I have the fractional
>part left:
>
>0.01  (binary).
>
>I know how to do it math wise:  0*2^-1+1*2^-2  =3D 1/4 =3D 0.25
<snip>

First, I have never used floats in a pic before so whatever I might
suggest below might have little vadility. Normally I get by by using
fixed point. Anyway, *if* you by normalised mean that the float it =
converted
into an fixed point format of sorts, then it's actually quite easy.

Let's take your example above as an start:

0.01 (binary) this is a 1q2 fixed point variable, if I were to convert =
this
into a int for ASCII conversion ( where I place the dp at any *fixed* =
point),
and use 2 digits after the decimal, I do something like this:

First we make it an 8q8 number by appending zero's (just to get to 8bit =
variables)

00000000.01000000 (8q8)

We want to convert into an int with the last two digits behind =
the(implied) dp, that gives
the following formula:

 Value =3D Value*100 / 256 =3D Value * 0.390625

i.e. first we scale the number (*100) to get the two digits after the =
dp, then
we 'fix' the implied decimalpoint by diving by number of bit after dp =
(in this case 8 bit
=3D 2^8 =3D256). If we take our 8q8 variable (1q2 in your original =
example) and
treat it an an 16 bit int, multiply with 0.390625 we will have an =
integer=20
with 2 digits after the dp.
The absolute easiest way to do this is to use Nicolias constant =
division/multiplication
on the piclist site, enter variable name/size and the constant =
(0.0390625) in my
example, desired accuracy, click generate code. Done !
Now you have an piece of code that will take your 8q8 input and generate =
an
scaled (x100) 16q0 output, which is perfect for normal bin to ascii =
conversion with 'fixed'
dp placement.

/Tony

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics