R
Rob
Hello,
I am wondering if there is a document which walks one through the
whole process of designing a FIR filter from getting the data to the
data final presentation?
I have a few problems;
-Creating a FIR filter which has a correct response. I have used some
filter programs but if I don't set the order to a high value it
doesn't look right. I want to keep the coefficients to a minimum to
save space. (For example a 3rd order filter)
-Selecting the cut off frequency. When I adjust this I seem to get
gain or I get a large attenuation.
-This may sound dumb but how do I calculate my sample frequency. I
seem to get different numbers when I calculate it. (if I sample 1280
times per cycle multiply that by 60 to get samples per second I get
76800Hz, but that seems to really impact the FIR filter response)
-How do I select the correct type of filter response to use and does
the FIR filter implement change with a different filter type or is it
just based on coefficients.
If I use;
-sample frequency of 1280Hz, cut off of 100Hz, order 10, I seem to
get a gain around the frequency of interest.
-sample frequency of 1280Hz, cut off of 400Hz, order 10, I seem to
get 0 gain around the frequency of interest.
-sample frequency of 1280Hz, cut off of 100Hz, order 5, I seem to get
a attenuation around the frequency of interest.
-sample frequency of 1280Hz, cut off of 400Hz, order 5, I seem to get
0 gain around the frequency of interest.
The coefficients which are calculated are all fractional values. What
is the correct way to scale them to whole numbers? I have read the
opencores Low Power FIR Filter and they suggest a few different ways.
When the coefficients have been converted does that same conversion
need to be applied to my incoming data and then reconverted on the out
coming data?
I find a lot of documents on FIR filters (more for DSP processors, but
not a lot for FPGAs) but they just talk about the FIR filter and not
the implementation.
So if there is a document which walks one through the whole design
process so I can create what I need.
Thank you
Rob
I am wondering if there is a document which walks one through the
whole process of designing a FIR filter from getting the data to the
data final presentation?
I have a few problems;
-Creating a FIR filter which has a correct response. I have used some
filter programs but if I don't set the order to a high value it
doesn't look right. I want to keep the coefficients to a minimum to
save space. (For example a 3rd order filter)
-Selecting the cut off frequency. When I adjust this I seem to get
gain or I get a large attenuation.
-This may sound dumb but how do I calculate my sample frequency. I
seem to get different numbers when I calculate it. (if I sample 1280
times per cycle multiply that by 60 to get samples per second I get
76800Hz, but that seems to really impact the FIR filter response)
-How do I select the correct type of filter response to use and does
the FIR filter implement change with a different filter type or is it
just based on coefficients.
If I use;
-sample frequency of 1280Hz, cut off of 100Hz, order 10, I seem to
get a gain around the frequency of interest.
-sample frequency of 1280Hz, cut off of 400Hz, order 10, I seem to
get 0 gain around the frequency of interest.
-sample frequency of 1280Hz, cut off of 100Hz, order 5, I seem to get
a attenuation around the frequency of interest.
-sample frequency of 1280Hz, cut off of 400Hz, order 5, I seem to get
0 gain around the frequency of interest.
The coefficients which are calculated are all fractional values. What
is the correct way to scale them to whole numbers? I have read the
opencores Low Power FIR Filter and they suggest a few different ways.
When the coefficients have been converted does that same conversion
need to be applied to my incoming data and then reconverted on the out
coming data?
I find a lot of documents on FIR filters (more for DSP processors, but
not a lot for FPGAs) but they just talk about the FIR filter and not
the implementation.
So if there is a document which walks one through the whole design
process so I can create what I need.
Thank you
Rob