# Why Matlab fdatool designed butterworth filter and butter command's co-efficients are different

I met an great confusion.

Using below command I have got co-efficents for a 300Hx cut-off high pass filter of order two.

[z p k] = butter(2,(300/44100),'high') [sos,g] = zp2sos(z,p,k)

Now this sos co-efficients are differ from what is obtained by fdatool.

As explained here https://dsp.stackexchange.com/a/19553/7868 , what i am wondering is why even for a 2nd order filter the co-efficients are different obtained by both methods.

• If your sampling frequency is 44100 Hz and your cut-off frequency is 300 Hz then the Matlab variable Wn must be Wn=300*2/44100, because Nyquist is 1 in Matlab. If this is the problem then I'll submit that as an answer. – Matt L. Jan 12 '15 at 8:40
• @MattL., You mean, i should use Nyquist limit instead of the sampling freaquency?, To say if the Fs is 48000, then butter(2,(300/24000),'high')  should be used? – nmxprime Jan 12 '15 at 8:44
• yes, that's what I mean. – Matt L. Jan 12 '15 at 8:44