Filtering Harmonics

Discussion in 'Wireless & RF Design' started by tenacity2986, Nov 1, 2013.

  1. tenacity2986

    Thread Starter New Member

    Oct 31, 2012
    When working with a 900MHz signal (902MHz to 928MHz passband) with allowances by the FCC for 1 W Output Power, I'm having trouble filtering the harmonics. What is the best way to filter harmonics to get minimum roll-off and maintaining bandwidth? Currently have a 5 element band pass filter before amplification and after. The first and second harmonic seemingly are very strong.
  2. Papabravo


    Feb 24, 2006
    When you say 'element', what are we talking about exactly? At those frequencies you can't be talking about actual inductors in your filters.

    Why are you using a bandpass filter? Harmonics are integer multiples of the fundamental so your second harmonic will be at 1.8 GHz. It doesn't take much to kill signals in that frequency range. Usually you get more bang for the buck with a low pass filter.
  3. Danm1


    Jul 19, 2010
    I think designing filters is a black art so I would use a design tool that does all the work. I've recently used the Filter Design Wizard on the Analog Devices web site, it's very nice.
  4. Papabravo


    Feb 24, 2006
    It is not a black art although it requires more mathematics background than the average design task.

    And you would never know if the tool was feeding you garbage. Stugeon's law "90% of every thing is crap" applies to the internet of things as well.
  5. LvW

    Active Member

    Jun 13, 2013
    Yes - it is very nice. And there are other filter design packages which also are very nice.
    However, if I remember well you intend to design a filter around 1GHz, don`t you? Don`t forget, that the AD program is for ACTIVE filters and you need an opamp that is able to cope with these frequencies.
    Perhaps you should consider also a PASSIVE filter structure.
  6. KL7AJ

    Senior Member

    Nov 4, 2008
    At that frequency, you'll have to use cavity resonators to get the kind of filter response you need.