High selectivity lumped element bpf

your requirements?

Center at Low frequency (hundreds of MHz), passband of 3-4 MHz and sharp roll-off (almost 40 DB attenuation at an offset of 3 MHz from center freq)

Yes you can do it this filter with lumped elements but the insertion loss will be very high due to high selectivity requirements because you will have to use many lossy components.
Instead, if you can design a helical/rod EM coupled filter in a good cavity, the insertion loss will be lower and the selectivity may be as good as wanted.

Then kindly provide me with some basics of such a filter(any design tutorial, theoretical background etc)



and what about Active filters? can i design an active filter having these specs?

Google for Helical filter design
Active filter? It might be possible as a mmic, I've seen some impressive active filters at high frequencies, but it would be a state of the art or beyond research project. Helical or tuned line cavity is the way to go for low loss, think about the Q you're asking of the filter components.

I explored helical filters a bit, but i found that they will be bulky ones. I have to design small size(at least reasonable) filter. Any other option keeping the Physical size constraint in view?

If you are going to use lumped elements then you will have to accept that high Q inductors are not going to be small, niether are the capacitors. The filter Q is going to be of the order of 100, so the components used need to be significantly better, I've not worked it out but you need to aim for around 1000 and even then the loss is likely to be quite high and the shape rather rounded.
There are plenty of free filter design and analysis packages around ( try QUCS ) you should be able to model your designs and see the effects of the losses.

I came across Dielectric Resonators filters and i have decided to proceed with this structure. kindly provide me with some easy to understand design material.

There is no 'easy to understand' design material for dielectric resonator filters. At the frequencies you are looking at you will need a coaxial resonator filter so search for coaxial dielectric resonator filter design', or just 'dielectric resonator filter design' . and see what turns up. The fiter will still be large and probably at the limit of what can be made. The lowest I have seen these filters used is around 400MHz. They are difficult to make unles you have the equipment to do it. The reosonators have to be cut to the correct length and that needs to be done carefully using a high speed saw or the material will develop micro cracks and the Q will be severly degraded. The coupling is often apperture coupling between the resonator cavities; this requires the removal of the outer plating from the resonator material, again not an easy task to get right without the right equipment.
This type of filter in my experience is best left to a specialist manufacturer. Specify the filter and fit it into your design.
Small size, high Q, low frequency and easy are not compatible parameters.

OK, Thanks for your reply.. i have found some design material.. but as you said, they are NOT easy to design and fabricate

what coupling technique(coupling between dielectric resonators) is best suited to get optimum design at such low frequency. I am unable to find this query on google. Kindly give some information regarding this aspect