Group Delay in Filter!

HI

I design the Active Filter.

My filter's group delay is 750n~1.5u.

I don't know this value is good or bad.

Who can estimate or determine the filter's group delay spec?

Pls let me know it.

Nobody will give good answer because it depends on your system requirements. Usually the S21GD number is obtained from RF System design and analyzes. Almost always the level of group delay is not important, instead level of group delay variation is what you need to check. Actually it shows what will be the difference in time delay for different parts of spectrum or how much will change timing into signal train after going through the filter. If timing changes a lot you will have ISI and BER will be reduced.

Thank you for your reply~!

I agree with your answer("...it depends on your system requirements. Usually the S21GD number is obtained from RF System design and analyzes. -RF-OM)

I design baseband filter in transmitter.
Unfortunately, I didn't find the group delay spec.

How can I check the Group delay spec in my system?

What counts is the difference in group delay across the frequency range your signal occupies. If the filter is tens of times wider than your signal, you usually will not have a problem unless your signal is near the edge of the frequency response.

As a rule of thumb, calculate what phase shift this differential delay will cause. Compare it with the modulation phase changes of your signal.

In your system design simulation and (or) analyzes there should be the BER analysis. It usually include group delay requirements. Check how much you will allocate for this baseband filter and what you will allow to contribute by RF front end filtering.

@RF-OM and others..
Hi ,
I think many of us know that excessive group delay can lead to BER reduction.But I could not locate any document which deduces the formal relationship between the two. If you have any material or suggestions pl let me know.

Additionally, I have a question:
We know that in the case of Gaussian noise, BER is directly related to the SNR through the Q function.
Now, the energy content of the signal is purely given by the magnitude of the Fourier Transform and hence with this line of argument group delay should not impact BER.

However for stochastic processes we don't have any thing like phase characteristics of the i/p and all we have is power spectral density(p.s.d).In that case again the output p.s.d of the system would be given by the i/p p.s.d * mag(F.T)^2 .

So though knowing intuitively that group delay should affect BER , I am not able to prove it mathematically.

Again, pl let me know any comments on this