I am using a transfer function to describe filter system but it seems once the function has expression of delay components
(ex. the numerator expression is: a*s^N + b*s^(N-1) + ... + z*s^M, M>=0),
bode function will show every odd frequency response for numerator has larger than s^5 or above delays (ex. M>=5)
I suppose this is caused by positive feedback of S>4 (90 degrees X 4) and hope this problem can be confirmed by Matlab. I also hope this small flaw can be fixed in new bode analysis function.