help me Why my Digital filter doesn't operate

Status
Not open for further replies.
H

hbaocr

Full Member level 4
Joined
May 13, 2007
Messages
232
Helped
25
Reputation
48
Reaction score
4
Trophy points
1,298
Location
Hochiminh City university of technology (VietNam)
Activity points
2,765
here the LOW Pass filter code with NUM and DEN designed by matlab (sptool) that fsample=1KHZ;Fc=40Hz;ButterWorth IIR filter

#define pi acos(-1)
double Numerator[]={0.00018321602337,0.000732864093479,0.001099296140218,0.000732864093479,0.00018321602337};

double Denominator[]={1,-3.344067837711875,4.238863950884069,-2.409342856586321,0.517478199788041};

Code: 
void CVedothiDlg::Low_pass_filter(double *NUM,double *DEN,double *sample,double data_update,double *output,unsigned int length_output_data)
{

	unsigned char i;

	sample[4] =sample[3];
	sample[3] =sample[2];
	sample[2] =sample[1];
	sample[1] =sample[0];
	sample[0]=data_update;
	//buffer output
	for(i=length_output_data;i>0;i--)
	{
		output[i]=output[i-1];
	}
	
	//update output data
	output[0]=NUM[0]*sample[0]+NUM[1]*sample[1]+NUM[2]*sample[2]+NUM[3]*sample[3]+NUM[4]*sample[4]-DEN[1]*output[1]-DEN[2]*output[2]-DEN[3]*output[3]-DEN[4]*output[4];

}
and then In the main program,the input data is

Code: 
void CVedothiDlg::OnPlot() 
{
	// TODO: Add your control notification handler code here
	int i;
	DOUBLE y[1000];
	DOUBLE x[4];
	DOUBLE data;
	for(i=0;i<1000;i++)
	{
		data=sin(2*pi*i/1000)+sin(2*pi*500*i/1000);
		Low_pass_filter(Numerator,Denominator,x,data,y,1000);
		Scope.Channels[0].Data.SetYData(y,1000) ;//draw signal output
	}

	
}
why it doesn't work????
Do anyone help me to make the filter work
 

There are trivial C programming issues, I think. I would generally suggest to trace program execution with a debugger to check for plausible operation.

One point is that in an expression as 2*pi*500*i/1000 the compiler is free to define the order of execution to its needs, thus an integer expressions 500*i and 500*i/1000 may be executed first. One would possibly cause overflow, if integer is 16 (you didn't tell about the enviroement), the other truncation anyway. Writing (500.0f*i) or (float)i would guarantee correct execution.

The oher point is, that local variables in a function are generally uninitialized in C, so your filter storage elements contain arbitrary valuesa t the beginning.

Finally, I don't know how the Scope component works, if it's used correct with the y output vector.
 

There is obvious buffer overflow :
You use sample[4] but size of array x is four so allowed index is 0,1,2,3 but not 4.
Also I think Scope.Channels[0].Data.SetYData(y,1000) ;//draw signal output
should be after for loop to speed up execution.
 

FvM, the C expression 2*pi*500*i/1000 is equivalent to (((2*pi)*500)*i)/1000.
If 'pi' is a double, then everything gets promoted to double, so there's no risk of dividing two integers.
 

Kernighan/Ritchie says, that a compiler is free to reorder * or + operators in an expression. Also, * and / have equal priority. I don't see a rule that would forbid to evaluate i/1000 first. It could be, that VC6 has additional self-defined constraints to provide more clarity to programmers.
 

K&R2 section A.7.6 says, "The multiplicative operators *, /, and % group left-to-right."

You may be thinking of this statement from K&R2 section 2.12:
"C, like most languages, does not specify the order in which the operands of an operator are evaluated."
That's a different issue.
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…