how to design a dual band filter BPF?

[kamgan]

hi friends

how to design a dual band BPF?

 

[Mityan]

Split the signal into two channels, apply separate single band BPF, then combine back.

 

[kamgan]

hi mityan

thx.. is there any coupling device is needed for making dual band BPF? or a simple connection is enough to do this filter.

 

[Mityan]

I think no more is needed but summator

 

[BradtheRad]

A recent thread was about a poster making a bandpass filter with two peaks.

It was based on two series LC arrangements like this:



The peaks are about 500 and 1100 Hz.

Several could be ganged together at one time.

 

[biff44]

you need resonators that resonate at the 2 different frequencies.

 

[tony_lth]

If you can accept the big insertion loss, you can divide and then combine two branches.
If not, it is not right to just combine them.

 

[thylacine1975]

In order to avoid the insertion loss described by tony_lth, I've had moderate success by using a frequency selective splitter/combiner - a diplexer - to firstly split the signals into 2 BPF's, and then recombine them afterwards.

Beginning with the disclaimer that it's not a polished solution(!), I ported the concept of the "stub tuner diplexer" (described in **broken link removed**) to where I needed it at 2.4 GHz. I then used ADS to optimise my structure for the (excessively thick) substrate I had to hand and used a PCB mill to create the final diplexer:


It offers ~15 dB of isolation between my two frequencies (only 40 MHz apart) at either port, with an insertion loss of ~1 dB (mainly due to microstrip radiation loss).
Combined with (cheap) commercially available filters (https://www.l-com.com/familylist.aspx?id=3019), it allows for pretty vicious filtering of two ISM band frequencies!

 

[volker_muehlhaus]

If you can accept the big insertion loss, you can divide and then combine two branches.
If not, it is not right to just combine them.

This might be misleading.

If you just connect the two bandpass filters in parallel, the input matching will be bad because you will see the stopband input impedance of one filter in parallel to the passband input impedance of the other filter.

Depending on the frequency spacing between the two band, you can use a diplexer (frequency splitter) to separate and combine the two channels. This avoids the 3+3dB power splitter loss that Tony mentioned.