0
\$\begingroup\$

I doing some amateur radio experimentation and want to create a bandpass filter. I believe that a Butterworth filter would be good because it has a flat frequency response in the pass band.

My center frequency is 1000Mhz and the bandwidth is 75 Mhz.

Since the the freq. is relatively high, I think the proper thing to do is use passive components in a ladder-type topology.

Does this seem like the right general path to follow for my design parameters?

Also, I'm going to build a circuit to do some testing. What should I on the lookout for in terms of my passive components (i.e. low ESR caps, etc.)

Thanks.

\$\endgroup\$
5
  • \$\begingroup\$ 1. You need to decide how fast you want the skirts to roll off to choose the order of your filter 2. You'll want to model you filter with variation in the component values according to the tolerances of the parts you're considering using. \$\endgroup\$
    – The Photon
    May 20 '18 at 5:32
  • \$\begingroup\$ I decided to use a 4th order Butterworth filter. Seemed like an it gives a reasonable trade-off in terms of the roll off in the stop band. As for components I was thinking of using "standard off the shelf tolerances" like 1% for my resistors. Is there a trade-off for tolerances? For example, like "one really wants a tight tolerance on the cap" for this or this reason. \$\endgroup\$
    – Brad Walker
    May 20 '18 at 5:49
  • \$\begingroup\$ 0.1% resistors can be bought for reasonable prices. Better than 1% capacitors and inductors, not so much. Even 1% may be hard to find depending on the value you need. \$\endgroup\$
    – The Photon
    May 20 '18 at 5:56
  • \$\begingroup\$ I'm finding, in my amateur radio hobby, that passive components aren't really so passive. \$\endgroup\$
    – Brad Walker
    May 20 '18 at 6:01
  • 2
    \$\begingroup\$ Resistors for 1 GHz? You may want to rethink about it. Inductors should be used instead. Just make sure that the SRF of your capacitors and inductors is higher than your max operating frequency by some margin. \$\endgroup\$
    – Mike
    May 20 '18 at 6:26
1
\$\begingroup\$

The wavelength at 1GHz is approximately 20cm (\$v\approx\frac{2}{3}c\$) for electrical signals. In order to be able to use a lumped system, you'll need the physical dimensions of the filter to be (preferably much) smaller than that in order to avoid reflections.

As you'll probably don't want to have any power losses in the passband, I guess you'll be going for an LC ladder filter. Design of such filters is quite heavy on the maths unless you use a tool like (eg. Tony Fisher's LC filter calculator). The inductor and capacitor components (as @Mike mentioned) should be able to handle these high frequencies, meaning that at least the Self-Resonant Frequency (SRF) should be high enough. You will find that this also limits the number of component values that you can use. You can avoid this constraint somewhat by connecting components in parallel/series where applicable.

In general you'll want to go for components with a high Q-factor. There are techniques to compensate for lower Q-factors (predistortion), but I could not find an online tool that implements this.

Design of the filter is usually followed by a sensitivity analysis to find out the allowed tolerance on the component values.

\$\endgroup\$

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.