TMS320F280049: Error in trigonometric functions

Pranav Ananthakumar

Part Number: TMS320F280049
Other Parts Discussed in Thread: C2000WARE

Hi all,

I am using the TMS320F280049 series for my application. I have used the sin function by assigning it to a double and provided it parameter also as a double variable. When I pass PI i.e 180 degrees in the sine function, it returns a negative value like -4.37139533e-08. Ideally it should return 0. I have included math.h as well. How do I rectify this issue?

Code:

double number;

double sine_of_number;

number = M_PI;

sine_of_number = sin(number);

Regards,

Pranav

over 2 years ago

0 Siddharth Deshpande over 2 years ago

TI__Mastermind 34765 points

Pranav,

Which library are you using for the sin function? Also , let us know the C2000Ware version that you are using.

Best Regards

Siddharth

0 Pranav Ananthakumar over 2 years ago in reply to Siddharth Deshpande

Prodigy 170 points

I am using C200Ware_3_03_00_00. Using math.h header file for sine function.

0 Siddharth Deshpande over 2 years ago in reply to Pranav Ananthakumar

TI__Mastermind 34765 points

Pranav,

So, you are not using any libraries within C2000Ware for computing the sin. You are using the compiler provided header file "math.h".

Is this correct? If so, let us know the compiler version that you are using,

Best Regards

Siddharth

0 Pranav Ananthakumar over 2 years ago in reply to Siddharth Deshpande

Prodigy 170 points

Yes, I am not using any library within C2000Ware for computing sine. I am using the compiler provided header file. The compiler version I am using is TI v21.6.0.LTS Also what library with C2000Ware do you suggest to use for sine computation?

0 Kier over 2 years ago

Mastermind 8075 points

Pranav Ananthakumar said:
Ideally it should return 0

Ideally it should never be zero because $π$ is irrational. Having said that, the value of M_PI is smaller than $π$ because it is truncated so sin(M_PI) should be positive at least where you are seeing a negative value. The conclusion is that there must be some rounding up of M_PI somewhere to approximately 3.1415927.

What are your compiler settings for --float_support and --tmu_support?