I want to design active filter with minimal gain of 2 dB with center frequency of 420Khz and pass-bandwidth of 12Khz.
the order should be as minimum as possible to achieve 3db cutoff at 414k -426K.
I tried designing using filter pro software but Q in this case is very high so i am unable to get required results.
the frequency is in middle of High and low frequencies so i am confused whether to use capacitor or inductor.

which topology will be best for this kind of design Sallen Key or MFB?

 

Read this attachment

 

You can't really do a "narrow" filter without a high Q. You should dig deeper into the nature of filters to understand why this is true. I'm not familiar with the acronym MFB, what does it refer to?

 

Hello,

MFB stand fro Multiple Feed Back topology.
it is descibed in the PDF Joe posted from page 16-18 and on.

I would have a look at how the IF filters of AM radios are build.

Bertus

 

Q: Will it allow a narrow filter to be constructed with a low Q?
A: No it won't

The defense rests, you honor.

 

I think only possible way to achieve narrow bandwidth is to combine a separate low pass stage with a HPF..

but what is difference b/w Sallen Key and MFB that is still confusing to me..

i will post after having a look at @JoeJester `s file

 

Hello,

The required Q here is 35, which isnt exactly low. When there is a Q that high the sensitivity of drift in wo with component variation will be large, so it will be hard to keep tuned to the right frequency. If you could go lower on the Q that would work better.

The main thing though is to check the sensitivity with component variation when considering a particular design. The components can change from part to part and with temperature and possibly humidity.

 

Hello,

The required Q here is 35, which isnt exactly low. When there is a Q that high the sensitivity of drift in wo with component variation will be large, so it will be hard to keep tuned to the right frequency. If you could go lower on the Q that would work better.

The main thing though is to check the sensitivity with component variation when considering a particular design. The components can change from part to part and with temperature and possibly humidity.

the only way to lower Q is to increase Pass-bandwidth which is unacceptable to me..

is there any other way??

 

I think only possible way to achieve narrow bandwidth is to combine a separate low pass stage with a HPF..
...

I do not think that this is the case. You will always be better off with a bandpass filter because whichever order you place the LPF and HPF in, the second one will load the first one and alter its response. The only way to design a filter is to specify the required attenuation and ripple in the passband, the transition band, and the stopband.

 

the only way to lower Q is to increase Pass-bandwidth which is unacceptable to me..

is there any other way??

Then the filter you want is not realizable. Bandwidth and Q are inversely related by definition. Doing what you want to do is equivalent to building a perpetual motion machine.

BW * Q = f0

As you can see the bandwidth times the Q is equal to the center frequency. If the center frequency remains constant then lowering the Q requires increasing the bandwidth. Ergo you cannot have a narrow filter with a low Q, period -- full stop.

 

the only way to lower Q is to increase Pass-bandwidth which is unacceptable to me..

is there any other way??

Hi,

As others have pointed out, the frequency and bandwidth are related to the Q, so changing one changes the other:
Q=Freq/Bandwidth

For yours the Q is:
Q=420000/12000=35

One of the problems that comes up is the center frequency stability, which means how well does the filter stay tuned to the center frequency when one of the components changes slightly. If the center frequency shifts by only 1 percent with a Q of 35 that would mean it would shift by 4200Hz, which is almost all of one side of the bandwidth.

Did anyone ask yet, what do you need such a sharp tuning circuit for?

 

You need a q of 35....like Al stated and a gain of 2.

Yes it can be done. The math works great with an ideal op amp. Use the file I posted And read about MFB topology.

Use E24 capacitors and E96 tesistors.

 

The answers we are giving are technically correct, but unacceptable. I suspect the OP will be searching elsewhere for the answer he wants, since we are obviously not clever enough to give him what he wants.

 

The answers we are giving are technically correct, but unacceptable. I suspect the OP will be searching elsewhere for the answer he wants, since we are obviously not clever enough to give him what he wants.

True. Especially since I responded to him first both at ETO and here. At ETO, I don't think there are any further responses past mine. I wonder what other forum he's visited.

 

Hi Joe,

Did they do a sensitivity analysis in that op amp paper? I ask because i'd like to know myself too as well as for this thread.

I was playing around with an all passive dual RLC network, and got it down to 0.5 percent for a 1 percent component change.
Of course these circuits would probably need some sort of tuning mechanism too depending on the accuracy needed.

 

I don't know Al. That chapter was extracted from Op Amps for Everyone, SLOD0006A, where Ron Mancini is listed as the Chief Editor. I thought I've seen Mancini around here a few times, well, either here or ETO.

There are a couple chapters in that reference addressing stability and other design considerations. I've attached it here.

The TI website doesn't carry that document anymore, but it does carry a few selected chapters of it. It is, however, on a lot of university websites.

Another good TI Application Report is Handbook for Op Amp Applications, found at http://www.ti.com/lit/an/sboa092a/sboa092a.pdf and also attached.