Group Delay in Active filter

[frankqt]

I am a newbie to the analog world. I am building an OOK system and I don't quite understand something.

As a background, My carrier frequency is 5MHz, the OOK system works such that the presence of 5MHz carrier would indicate 1 and lack of it would indicate 0.

I have designed an active filter using TI's filter Pro application and simulate it. It works however the problem is that the filter takes too long to stabilize due to group delay. So, if my data rate is high (say 200KHz), filter cannot catch up and it blurs between 1 and 0. At least on simulation.

How can I solve this problem?

As a side note, if I use two filters (low pass, <5.2MHz) and high pass (>4.8MHz), I do not see such delay as opposed to BP filter, I also don't get this.

 

[keith1200rs]

What Q did you choose for your bandpass filter? Try reducing Q.

Keith

 

[LvW]

.........
I have designed an active filter using TI's filter Pro application and simulate it. It works however the problem is that the filter takes too long to stabilize due to group delay.
.....
.

I doubt that this is a problem of group delay. I think it's a question of settling time. Therefore, Keith's mentioning of the filter Q.

 

[frankqt]

Got it.. When I reduce the Q, the filter become much faster, now I get it. What is the relationship (some sort of rule of thumb) between settling time and Q?

Also, what is the relationship between settling time and Carrier frequency? (In other words, say for 5Mhz carrier, if I want to have a 500Khz bit rate, what is the maximum settling time is recommended?)

 

[keith1200rs]

I cannot remember the exact formula, but the number of cycles a high Q filter takes to settle is proportional to Q (there is probably a pi somewhere in the formula). As LvW says, it is the settling time due to Q not group delay that is the problem. Group delay in itself just creates a delay.

Keith.

 

[frankqt]

Thanks Keith. I have done some research and you are right, the sharper the filter the longer it takes to settle.