insert_coin
Newbie level 3
Hey guys,
It is my first time here and my first time working with SW filters or oversampling filters. In fact my problem is the following one.
I have designed a filter using the Fdatool that you can find in matlab. 250 kHz sampling freuqency some attenuation 5 kHz cut off elliptic and so on.
I am trying to implement that in the FPGA in a direct form one. The problem comes when i try to scale the coefficients from 16 (I think is how matlab gives the coeffs) to 18 bit fixed point.
At this stage i am lost. It seems that due to the high sampling with respect to the cut off, the numerator coefficients are to small in respect to the denominator ones and the quantization error becomes to big.
I do not want to put more details since i do not ask for a solution to my task, in fact i ask for a guidance or documentation where i can find how to scale the coeffs for a IIR filters. Because as i said is the first time and the time to learn!
Thanks a lot in advance
It is my first time here and my first time working with SW filters or oversampling filters. In fact my problem is the following one.
I have designed a filter using the Fdatool that you can find in matlab. 250 kHz sampling freuqency some attenuation 5 kHz cut off elliptic and so on.
I am trying to implement that in the FPGA in a direct form one. The problem comes when i try to scale the coefficients from 16 (I think is how matlab gives the coeffs) to 18 bit fixed point.
At this stage i am lost. It seems that due to the high sampling with respect to the cut off, the numerator coefficients are to small in respect to the denominator ones and the quantization error becomes to big.
I do not want to put more details since i do not ask for a solution to my task, in fact i ask for a guidance or documentation where i can find how to scale the coeffs for a IIR filters. Because as i said is the first time and the time to learn!
Thanks a lot in advance