0
\$\begingroup\$

I am a newbie to filter design. I would like to know: is it common to use a first step active filter followed by five or six passive LC-filters for very high orders? Or do people usually don't combine those two?

In my case I'd have:

Active filter with some gain → five LC-circuits → 2nd order active filter (without any gain)?

\$\endgroup\$
1
  • \$\begingroup\$ It's OK. Depending on the overall system architecture I'd consider making the first stage(s) passive. I've done this with a low pass filter, to reduce nonlinearity (intermodulation distortion) from high level out-of band signals mixing in the active stage. But if you're fighting the noise floor this may be counter-productive. \$\endgroup\$ – Brian Drummond May 7 '16 at 14:33
1
\$\begingroup\$

People don't "usually" combine active and passive filters in the way you're suggesting, but it's certainly possible. It's not generally a good idea, but it's certainly possible.

Fairly common is a stage of passive filtering followed by high-order active filters. In this case, the first stage filters out-of-band noise without worrying about the active element being driven non-linear. The relatively clean output is then handled by the active elements.

Except in unusual cases, a 5-stage LC filter can usually be done better by an active equivalent, particularly if all 5 stages are trying for the same frequency. For such applications, the ability to tune each section, the better component sensitivity, and the reduced loading effects between stages all make active filters a better choice. This assumes that the frequencies involved are well within the capabilities of the active elements, of course.

\$\endgroup\$
1
  • \$\begingroup\$ I'd agree that a 5-stage LC filter would seem odd, but if the active filter may behave wonky when given signals that are way out of band, or its output may contain such signals (both behaviors are common with switched-capacitor filters) even crude RC or double-RC filters may clean things up quite cheaply. \$\endgroup\$ – supercat May 7 '16 at 15:51
0
\$\begingroup\$

There's nothing inherently wrong with doing that. Active filters have the possible advantage of having high input impedance which doesn't load the input so much and low output impedance which allows filter stages to be joined together without creating complicated transfer functions.

A 5 LC circuit will result in a pretty complex transfer function on its own that may make it difficult to design the cutoff frequency to reside at a single place. You'll also want to ensure the cutoff of the 5 LC-circuits matches up with the 2nd order active filter if you're doing it just to increase the rolloff.

To summarize, active filters can generally be added together and their frequency response won't change, but passive filter stages all impact each other so you can't just take a single passive, add it to another passive and assume that the transfer function will double the rolloff at the cutoff frequency of each.

\$\endgroup\$
1
  • \$\begingroup\$ @user337720 You have those spikes because those are the LC resonant frequencies most likely. If you drew your circuit and calculated your transfer function, you would likely see where the issue is. It would be poles at those frequencies that aren't getting cancelled out by zeros. If the last stage is active, you should be able to just add it in since it should be high input impedance. If you need a flat frequency response before the cutoff, there's probably a better filter design that you should use instead. Lookup the butterworth filter if you're interested in maximally flat in the passband. \$\endgroup\$ – horta May 7 '16 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.