Guest
|
 Posted:
Fri Dec 02, 2005 5:17 pm Post subject:
Filter design for asymetric frequency response of system? |
|
|
Hi,
I wonder if somebody would be able to point me in the right direction
about trying to design a filter to compensate for a frequency response?
I have a 1024 point image of the frequency response of the system. from
-44MHz to +44MHz (88MHz bandwidth)
This I invert, and put through a 1024 IFFT, and record the Real and
Imaginary output.
When I put this IQ signal through a signal generator and back into my
system I get a completly flat response - therefore showing my IQ signal
from the IFFT is the signal required to flaten the response.
Question: How do I now get a filter design from this information? - I'm
thinking that the output of the IFFT is the impulse response of the
system, which should then therefore be the coefficients of my filter.
However, I'm unsure how to use this as coefficients as my frequency
response is asymetric about DC.
When I just use the IFFT 1024 sample output for my I and Q filter
coeffs, the gain I get is both the same but the phase response of the
filters matches either the shape of the negative frequency part of my
response or the positive part.
The phase response of the system is flat, its just frequency gain I'm
interested in removing.
What I would like is two filters, one for my I samples and one for my Q
samples which correct for this asymetric freq response. I just haven't
clicked on how to use the information from the IFFT.
Is it as easy as saying the negative frequency response part is my Q
filter resopnse and the positive frequency response is the frequency
response for my I filter?
Best Regards,
Phil. |
|
Bhaskar Thiagarajan
Guest
|
| Posted:
Fri Dec 02, 2005 11:55 pm Post subject:
Re: Filter design for asymetric frequency response of system |
|
|
<phil.ware@ntlworld.com> wrote in message
news:1133534186.298399.191630@g44g2000cwa.googlegroups.com...
| Quote: | Hi,
I wonder if somebody would be able to point me in the right direction
about trying to design a filter to compensate for a frequency response?
I have a 1024 point image of the frequency response of the system. from
-44MHz to +44MHz (88MHz bandwidth)
This I invert, and put through a 1024 IFFT, and record the Real and
Imaginary output.
When I put this IQ signal through a signal generator and back into my
system I get a completly flat response - therefore showing my IQ signal
from the IFFT is the signal required to flaten the response.
Question: How do I now get a filter design from this information? - I'm
thinking that the output of the IFFT is the impulse response of the
system, which should then therefore be the coefficients of my filter.
However, I'm unsure how to use this as coefficients as my frequency
response is asymetric about DC.
When I just use the IFFT 1024 sample output for my I and Q filter
coeffs, the gain I get is both the same but the phase response of the
filters matches either the shape of the negative frequency part of my
response or the positive part.
The phase response of the system is flat, its just frequency gain I'm
interested in removing.
What I would like is two filters, one for my I samples and one for my Q
samples which correct for this asymetric freq response. I just haven't
clicked on how to use the information from the IFFT.
Is it as easy as saying the negative frequency response part is my Q
filter resopnse and the positive frequency response is the frequency
response for my I filter?
|
I'm not sure I can set you straight on your terminology but if you use the
real part of your IFFT output as coeffs for a filter that operates on your I
samples and the imag part of your IFFT output as coeffs for a filter that
operates on your Q samples, you should get what you want.
For correcting an asymmetric freq response, you need a complex filter which
is what you are doing with the above operation.
Cheers
Bhaskar
| Quote: | Best Regards,
Phil.
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in