If you have a problem or need to report a bug please email :

There are 3 sections to this support area:

DOWNLOADS: access to product manuals, support files and drivers

HELP & INFORMATION: tutorials and example files for learning or finding pre-made modules for your projects

USER FORUMS: meet with other users and exchange ideas, you can also get help and assistance here

Process nth order filter

DSP related issues, mathematics, processing and techniques

Process nth order filter

Postby juha_tp » Tue Apr 07, 2020 8:42 am

Lets say one have a nth order (n>2) filter ... how to process it without breaking it to SOS?

Are there any ready to use implementations available (I checked quite a many examples but didn't find one (bypassed some asm listings there cause of I'm not good in reading asm code))?
Posts: 21
Joined: Fri Nov 09, 2018 10:37 pm

Re: Process nth order filter

Postby martinvicanek » Tue Apr 07, 2020 5:36 pm

SOS means second order sections aka biquads, right?
You could use a direct form like

y = b0*x + b1*x1 + ... + bn*xn;
y = y - a1*y1 - a2*y2 - ... - an*yn;

xn = xn-1;
x1 = x;

yn = yn-1;
y1 = y;

However, in single precision rounding errors will quickly make it unstable even for moderate order n. That's why we moslty use biquads. ;)

A minor remark on why I broke the filter equation y = ... in two lines: Flowstone's code compiler has a limitation regarding the size of single line formulas it can handle. (It won`t throw an error if you exceed this limit).
User avatar
Posts: 1080
Joined: Sat Jun 22, 2013 8:28 pm

Re: Process nth order filter

Postby juha_tp » Wed Apr 08, 2020 6:26 am

OK, yes, second order sections.

I found my own SM example for biquad case:

Code: Select all
monoin inL;
monoin inR;
monoout outL;
monoout outR;

float a0, a1, a2;
float b0, b1, b2;
float inL1, inL2, inR1, inR2;
float outL1, outL2, outR1, outR2;

a0= ...
a1= ...
a2= ...
b1= ...
b2= ...

// Proces left channel
outL = a0*inL + a1*inL1 + a2*inL2 - b1*outL1 - b2*outL2;

// Process right channel
outR = a0*inR + a1*inR1 + a2*inR2 - b1*outR1 - b2*outR2;

// save previous I/Os
outL2 = outL1;
outL1 = outL;
inL2 = inL1;
inL1 = inL;

outR2 = outR1;
outR1 = outR;
inR2 = inR1;
inR1 = inR;

so, extending that would then be the way to go.
Posts: 21
Joined: Fri Nov 09, 2018 10:37 pm

Return to DSP

Who is online

Users browsing this forum: No registered users and 8 guests