Low pass filter, cut off 500 Hz, short response time...

What

You need to get hold of a decent text on filter design - I use the "Electronic Filter Design Handbook" by Arthur B. Williams and Fred J. Taylor - ISBN 0-07-070434-1. Since then there has been a third edition (ISBN 0-07-070430-9) which is supposed to be even better, but it seems to be out of print -

offers three second hand copies for about $130. You should be able to find it - or something very like it - in a university library.

The usual choice for what you seem to want to be doing is a "linear-phase" or "Bessel filter", though a "synchronously tuned" filter (consisting of identical multiple poles) actually gives fastest settling for a given number of poles, with the step getting to better than 99% after five time constants. Adding more poles makes the step steeper at the 50% point, and pushes the 50% point back from around one time constant at one pole to closer to three at ten poles.

A simple LCR filter is two poles ...

--------- Bill Sloman, Nijmegen

Do it digitally. Distortionless digital filters are superb because they can look backwards as well as forwards in time.

More on my website.

But if you must go analog, use my Active Filter Cookbook.

--
Many thanks,

Don Lancaster
Synergetics   3860 West First Street  Box 809  Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

My

J.

edition

seems

for

it -

fastest

better

step

one

Don Lancaster's Active Filter Cookbook has many fans. Williams and Taylor cover a wider range of filters and filter types.

------- Bill Sloman, Nijmegen

Unless you like inductors the size of tuna cans, use an active filter. (Except if you are building a speaker crossover filter).

Tam

No, the resonance peak is down to the style of filter. Generally a Chebyshev will be bad, a Butterworth somewhat better and a Gaussian has no peak. You pay a price for the lack of resonance in a droopier response around the corner frequency.

d

Pearce Consulting

Viram escribió:

When you design a LP filter, actually you are limiting the response time. For 500Hz the time constant is about 0.3 msec. If you want to limit the resonance peac, you can share the capacitor in some smaller parallel units and put one resistor in series with one of them.

Zverev is a great book, certainly, but it's really not a good 'introductory' tome on the subject. I think the poster would be better off with Don Lancaster's old book, or perhaps "Analog and Digital Filter Design" by Winder.

Noble Publishing is also the home of Randy Rhea, whose book "HF Filter Design and Computer Simulation" is quite good as well. Rhea's a very practical kind of guy... (although if you look at the price of his publications, he wants to make a pretty darned good living at it too!)

---Joel

If you want the bible of filters...:

The time plots--which is what the OP cares about--are all there.

I don't have the following, but it too has focus on the time domain and is less expensive:

Don's book works great where the needs are straightfoward and the frequency is not very high. It is a fast path to getting done, and that says something.

Hz.

peak.

the

Fred

it

copies

like

tuned"

around

Williams and Taylor cite Zverev very frequently, and include time plots for every filter type - both step response and impulse response. I imagine that Zverev presents the theory better - Williams and Taylor just direct you to the relevant bit of the literature - but Williams and Taylor is very much a book for working enegineers.

and is less

frequency is

something.

Williams and Taylor is handy when the situation gets more complicated - I once ran into a situation where we needed a fast-settling filter, but couldn't start the filter chain with a high-Q stage because of the risk of clipping, and was able to find a "linear phase with equiripple error" filter which did the job. The electron microscope involved ended up being loaded on the truck at eleven in the evening under the eye of the managing director, so I scored a few brownie points on that one.

------- Bill Sloman, Nijmegen

I'll second that one. If you head over to TI's site and check their analog knowledge base for a program called FilterPro it'll design RC active filters for you up to 10 poles with either MFB or Sallen-Key topologies. Though in my opinion they're far too conservative on the gain-bandwith requirements of the op-amps.