A few questions about Butterworth Filter

[jesterahs]

Hello everyone! Thank you for looking at this thread.

I am trying to design a 10th order Butterworth filter using passive elements (i am using the Insertion Loss Method, and using prototype numbers). So far, I have designed a 5th order Butterworth filter, and I am planning to cascade two of them together. My questions about this are:

1. Do I need a matched attenuator in between them? If so,
2. Should I design the matched attentuator to have a Zin = Zout = 50 Ohms (since the source and load resistance of my filters are 50 Ohms)
3. And if looking into the matched attenuator gives an impedance of 50 Ohms, can I remove the 50 Ohm load resistor from my first filter?

If this is confusing, and you are willing to help [ please! :) ], please let me know and I can email you a MS Word document of my work so far. Thanks to all for your time!

 

[Enthalpy]

Mamma mia!

Cascading two 5th Butterworth doesn't make a 10th Butterworth. It makes a poor filter. This could sometimes be advantageous with biquad active filters, not with LC filters.

You would need an active buffer, or a very efficient attenuator, so each 5th order section sees a pure resistive load or source. If not, the transfer function is flawed.

In passive filters, odd orders are preferred. Fewer inductors, unity gain at center frequency (what frequency is it? This changes the technology a lot! I guess VHF and above, as you choose 50 ohm), simpler calculations.

Butterworth is a bad transfer function. For an LC filter, use one with zeroes, especially an elliptic one. For the same selectivity, they have fewer components and a better time and phase response as well. Don't believe what people go on repeating, it's just false. Inverse Chebychev would also be good.

I think Insertion Loss Method dates back to hand calculations... For LC filters, the one very best method is not to compute by yourself. LC filters are tabulated in books. Take the one by Zverev (old but this didn't change). Elektor also published an excellent series of articles about filters will all the tables you need. And I suggest to get Filtercad, free here
http://www.linear.com/designtools/software
it has built-in transfer functions, computes step responses (compare them) and gives component values.

 

[oswaler]

Hello there!

It's great to see someone working on designing a Butterworth filter. To answer your questions:

1. It is not necessary to have a matched attenuator in between the two filters. However, it may improve the overall performance of the filter by reducing reflections and improving the overall impedance matching.

2. If you do decide to use a matched attenuator, it would be best to design it with a Zin = Zout = 50 Ohms to match the source and load resistance of your filters.

3. If the matched attenuator gives an impedance of 50 Ohms, then you can indeed remove the 50 Ohm load resistor from your first filter. This will simplify your circuit and improve the overall performance.

If you would like further assistance with your design, please feel free to email me your work. I would be happy to take a look and offer any suggestions or advice. Good luck with your project!