Fun With Filters 9Mhz IF

Discussion in 'Homebrew and Kit Projects' started by VK4HAT, Oct 21, 2016.

ad: L-HROutlet
ad: l-rl
ad: L-Geochron
ad: HRDLLC-2
ad: L-MFJ
ad: MessiPaoloni-1
ad: Left-2
ad: Left-3
ad: abrind-2
  1. VK4HAT

    VK4HAT Ham Member QRZ Page


    After having lots of success with DC receivers, i guess the next logical step is to move to a superhet design. So time to start playing with crystal IF filters.

    Everything starts with a schematic, no i did not design the filter using software, nor did i measure the emotional parameters of the xtals to match them up, just went hit and miss to see what happens. I made 2 identical filters, one with the 100pf caps and another with 50pf caps.


    Went with the super ugly design method on a scrap of board thats been used far to many times for other experiments. :)


    Using the lowest output setting on my signal generator, a series of measurements were made, converting the amplitude millivolts to Db and each filter was plotted and graphed in a spread sheet.

    1oopf cap version, data below -45db is not all that accurate as my scope was picking up general rf noise to the tune of 10mv, filter is about about 900hz wide.


    this was graphed prior to terminating the filter with 1K impedance, so it might be a bit wrong town, its something i will have to redo.

    Next job is to measure some parameters of the crystals and then use Dishal to give me better cap values and to built it again, to see if I can make it better with much less ripple and with more theoretical shapes.
    WA7PRC and AI3V like this.
  2. SV1IYF

    SV1IYF Ham Member QRZ Page

    Did what you are doing during the early 70s.
    It was the time the "ladder" network started appearing in radio magazines.
    Prior to this the "lattice" topology (half or full) was mainstream for diy receiver builders.
    I used CB crystals. They are 3dr overtone, hence their fundamental freq is a bit higher than 9 MHz.
    In the quest for a better 6:6O* db shape factor these things I remember:
    1) The shunt capacitors better have an increasing value from the ends to the center of symmetry.
    If you are patient enough, try adjustable trimmers in parallel with fixed capacitors.
    2) Matching all the crystals as per their resonance frequency is a good thing, but with crystals up to 200-300 Hz apart you can have a better shape with not significant increase in passband ripple.
    The trouble with this is that the position of each crystal in the ladder is important.
    Symmetry has to be maintained as well.
    So, in a 5 crystal ladder you should have 3 groups (1 for the centre +2x2 for the other 4 places).
    As you can not practically have crystals matched to the same Hz, there is a try and error job to find the best permutation (which crystal to what position).
    3) More important to 2) is to select crystals that have low series impedance at resonance.
    You measure this (along with the series resonant frequency) with a Pi network the legs of which are resistors .
    At one side of the Pi you connect the rf generator and at the other your measuring device (rf voltmeter or oscilloscope)
    4) For an SSB TX filter, a shunt crystal at the input of the filter, offset to the carrier frequency helps carrier rejection.
    Depending to whether this is a USB or LSB filter there is capacitance added to the input which should be compensated for at the output.
    5) Grounding all the crystal cans is a good thing, but then you will see another filter.

    (*) Unless you have very expensive test equipment, make your decisions on a more "modest" shape factor, for example 6 to 46 dB.
    WA7PRC likes this.
  3. KL7AJ

    KL7AJ Ham Member QRZ Page

    emotional parameters?
    WA7PRC likes this.
  4. VK4HAT

    VK4HAT Ham Member QRZ Page

    Thanks Nikos, there is a lot of good information in your post that I can apply to my next lot of builds.

    When you measure and get the wrong values, it will make you cry ;)
    N0SYA and WA7PRC like this.
  5. G4AON

    G4AON XML Subscriber QRZ Page

    When testing filters, be very careful with the in/out impedances. For example, if you look at a typical 500 Ohm 9 MHz commercial AM IF filter using 50 Ohm test equipment, the response curve in the passband will have significant peaks and troughs, more like the Himalayas than what you expect to see.

    73 Dave
  6. VK4HAT

    VK4HAT Ham Member QRZ Page

    Hi Dave, you are correct, I was listening to the SolderSmoke podcast the other day and they were talking about issues they were having with a filter of unknown impedance etc etc and how it was fixed by determining in and out impedances etc. Something that i did not do in my above experiments was use proper impedance terminations, and that is something that I will do when i revisit these learning adventures.

Share This Page

ad: k1jek