January 29, 2015
i am reading the FX book and came to filters, and i wanted to try out the FIR Designer.
I watched the video and tried to do same things, until i had an issue.
With the optimal method, when m_nIRLength (Order) reaches 350 the m_h_Left[i] produced are all = 0.000000000... ( 0<=i<=1023)
From that value of m_nIRLength, the array is filled only with 0.0000000 always. The only difference i can see, is that the graphic of the frequency response (with filled flag enabled) is empty for values 350<= x = 356.
The audio in real time, totally blocks, and also the impulse response graphic.
Also for some other values of m_nIRLength, i get the audio signal blocked.
I checked with midimapper and avid mbox 3 asio drivers, and got same result. Tried to resize the audio buffer, but nothing.
The CPU load is almost at minimum, i am on a Windows 7, i5 / 8gb ram.
There is something i am missing ?
January 29, 2017
As stated in the book, the Remez Exchange algorithm may not converge for a given set of filter specs with the Optimal method. When it does not converge, the coefficients are zero (or sometimes garbage). So, that is what is happening. For the record, I am using the Parks-McClellan C-code (it's been around for a long long time).
The Frequency Sampling Method filters do not have this issue as there is no search algorithm in that design.
Most Users Ever Online: 36
Currently Browsing this Page:
Guest Posters: 1
Newest Members:Matt, dspstudent, strings4v, TheSmile, semihyavuzz, alfredLue, danioc, midnightskate3, Alia5, Shamal Sundar
Moderators: W Pirkle: 209
Administrators: Tom: 67, JD Young: 80, Will Pirkle: 0, W Pirkle: 209