I am looking for a PIC (16F84 or similar) assembler routine to compute
arcsin with 32 bit floating point numbers.  Routines using lookup
tables and interpolation need not apply, this needs to be accurate.
Alternately, a good arctan routine would be acceptable, but I do need
arcsin for this application and so I would have to compute an extra
square root, which I'd like to avoid.  I thought I found an arctan
routine on the web, but now I can't locate the link.

Failing an implemented routine, a well described algorithm would be
acceptable.  I already have the power (McLauren, Taylor) series
expansion algorithms for both asin and atan.  In my application, the
angles will be between 0-30 degrees; a range where the power series
converge relatively quickly.  If I'm going to implement something from
scratch though, I'd like to use an algorithm that converges well over
the complete range of the function, so it is more general. So, any
math majors out there?

Thanks,

Gregg Lobdell
Seattle, WA