Cordic algorithm for atan in DSP processor

Hello,
I am trying to implement cordic algorithm for atan in ADSP 2191.
I have understood cordic algorithm. I wanted suggestion from you about
its implementations. what the speed achievable? Any source code???ASM
codong or c coding

Thank you,
with regards
praveen

"praveen" <praveenkumar1979@rediffmail.com> wrote in message
news:ff8a3afb.0312080201.1b0888f7@posting.google.com...
> Hello,
> I am trying to implement cordic algorithm for atan in ADSP 2191.
> I have understood cordic algorithm. I wanted suggestion from you about
> its implementations. what the speed achievable? Any source code???ASM
> codong or c coding

What is the reason you want to use the cordic algorithm?
From what I understand, the cordic provides advantages in hardware
implementations (asic/fpga). In a programmable DSP environment, it's still
certainly possible to implement, but you won't gain anything from it. You
are probably better off choosing some other method (look up tables,
approximations, etc) to compute atan.

Cheers
Bhaskar


>
> Thank you,
> with regards
> praveen

Hello,

what is the accuracy of estimation of atan using cordic algorithm?. 
If i use look up table it will be huge since my step size of LUT is 5
microradian (half octant ie pi/(8*5microradians) is huge).
Using approximate method what is accuracy obtainable???


I need your suggestion
waiting for reply
with regards
praveen

praveenkumar1979@rediffmail.com (praveen) wrote in message news:<ff8a3afb.0312082113.1eb22222@posting.google.com>...
> Hello,
> 
> what is the accuracy of estimation of atan using cordic algorithm?. 
> If i use look up table it will be huge since my step size of LUT is 5
> microradian (half octant ie pi/(8*5microradians) is huge).
> Using approximate method what is accuracy obtainable???
> 
> 
> I need your suggestion
> waiting for reply
> with regards
> praveen

Hi Praveen

The approximation error would depend on the order of the polynomial
chosen. This is an inherent error though. Horner's algorithm is a good
trick to use for this. I have used this before for computing atan
using Q-15 fixed point format and achieved average error of the order
of 10^-2. Of course actual error depends on the number of recursions
and the represenatbility of the coefficients in the  fixed point
format with min. possible error.

To evaluate the error Horner's algorithm may be treated as a first
order IIR filter. You may do away with the round off error if unsigned
multiplication is used and the coefficients being scaled by the
sampling interval.

Hope this helps
-Nithin

>>>>> "praveen" == praveen  <praveenkumar1979@rediffmail.com> writes:

    praveen> Hello,
    praveen> what is the accuracy of estimation of atan using cordic algorithm?. 
    praveen> If i use look up table it will be huge since my step size of LUT is 5
    praveen> microradian (half octant ie pi/(8*5microradians) is huge).
    praveen> Using approximate method what is accuracy obtainable???

At step n of the cordic algorithm the angle is about +/- atan(1/2^n),
so a few iterations gets you to a pretty small angle.  You can stop
there or, if necessary, use the atan series to finish it off:

      atan(x) = x - x^3/3 + x^5/5 + ...

Ray

In a DSP with a multiplier, a CORDIC is probably not the right solution.
It is a hardware solution to avoid doing multiplications.

praveen wrote:

> Hello,
> I am trying to implement cordic algorithm for atan in ADSP 2191.
> I have understood cordic algorithm. I wanted suggestion from you about
> its implementations. what the speed achievable? Any source code???ASM
> codong or c coding
>
> Thank you,
> with regards
> praveen

--
--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
401/884-7930     Fax 401/884-7950
email ray@andraka.com
http://www.andraka.com

 "They that give up essential liberty to obtain a little
  temporary safety deserve neither liberty nor safety."
                                          -Benjamin Franklin, 1759

praveenkumar1979@rediffmail.com (praveen) wrote in message news:<ff8a3afb.0312082113.1eb22222@posting.google.com>...
> Hello,
> 
> what is the accuracy of estimation of atan using cordic algorithm?. 
> If i use look up table it will be huge since my step size of LUT is 5
> microradian (half octant ie pi/(8*5microradians) is huge).
> Using approximate method what is accuracy obtainable???
> 
> 
> I need your suggestion
> waiting for reply
> with regards
> praveen


I'd suggest using Taylor's series. Look for the formula in book about
Engineering's Mathematics.

JaaC

praveenkumar1979@rediffmail.com (praveen) wrote in message news:<ff8a3afb.0312082113.1eb22222@posting.google.com>...
> Hello,
> 
> what is the accuracy of estimation of atan using cordic algorithm?. 
> If i use look up table it will be huge since my step size of LUT is 5
> microradian (half octant ie pi/(8*5microradians) is huge).
> Using approximate method what is accuracy obtainable???
> 
> 
> I need your suggestion
> waiting for reply
> with regards
> praveen

Also interesting could be to have a look at TI's DSP app notes and ADI's E-E notes.

JaaC

You could use a lookup table of modest size, and interpolate between table
entries to get more resolution.  Unfortunately you will have to ask others
how to calculate the required table size for a given accuracy, though.

Hello,

If you write the cordic in C for atan . How much is the usual number of cycle
required?????

waiting for reply
praveen