Filter design help

Discussion in 'Homebrew and Kit Projects' started by PY2RAF, Sep 1, 2019.

ad: L-HROutlet
ad: l-rl
ad: Subscribe
ad: Left-2
ad: L-MFJ
ad: Left-3
ad: MessiPaoloni-1
  1. PY2RAF

    PY2RAF XML Subscriber QRZ Page

    Hi there folks! o/

    I designed a 69.450 MHz 3 MHz wide Band-Pass filter, in order to use as a narrow band filter for a SDR.

    Schematic:
    [​IMG]


    In theory everything was great and perfect. See the LTSpice simulation:
    [​IMG]

    Exactly what I wanted. Yay, right.

    Not so fast. As a frequent QRZer here signs in his posts, theory only works in perfect vacuum. This is actually what I got in a 40-80 MHz sweep:
    [​IMG]

    Well, the filter actually eat 100% of the signal. Nothing comes out of it. Heck of a filter, right.

    Here is the filter PCB layout:

    [​IMG]

    The red line depicts the "passthrough" mode - The signal enters the first RF switch and moves straight to the 2nd RF switch, untouched. This pretty much rules out the RF switch issue.

    SO. My forum question.

    Can you actually envision what is wrong in this filter design? What am I missing?

    For completeness, these are the filter components:

    2400 pF cap - https://www.digikey.com/product-detail/en/kemet/C1206C242J5GAC7800/399-14673-1-ND/7103965

    2.2 nH ind - https://www.digikey.com/product-detail/en/abracon-llc/AIMC-0805-2N2S-T/535-11562-1-ND/2782791

    2.2 µH ind - https://www.digikey.com/product-detail/en/tdk-corporation/MLF2012A2R2KT000/445-1051-1-ND/504449

    2.4 pF cap - https://www.digikey.com/product-detail/en/kemet/C0603C249D5GAC7867/399-14972-1-ND/7382521

    Help? :)

    I appreciate the help o/

    73 de PY2RAF - Rodrigo.
     
  2. K7JEM

    K7JEM Ham Member QRZ Page

    Right off, without even looking further on the schematic, the 2400pF capacitors look way out of line. At 69MHz, a 2400pF capacitor has about 1 ohm of capacitive reactance, thus shorting out the 50 ohm impedance on both the input and output of the filter. The other components are probably wrong, too, but I didn't even need to look at them to see this drastic mistake.

    Edit: I see that the parallel inductor should resonate the capacitor, so ignore my comment! I will look some more at the circuit.
     
    Last edited: Sep 1, 2019
    PY2RAF likes this.
  3. K7JEM

    K7JEM Ham Member QRZ Page

    OK, another possible issue is low Q on the inductors. The 2.2nH are showing a Q of 10 at 100MHz, which is probably way low for a circuit of this type. You probably need a hand wound coil, and a Q of several hundred. These narrow filters require high Q components, and selection of coils and capacitors can be critical.

    Same thing goes for the capacitors. You may have to use caps designed for RF or microwave use to get a Q high enough to be useful. General purpose capacitors may not work very well at all.
     
    Last edited: Sep 1, 2019
    PY2RAF and SM0GLD like this.
  4. SM0AOM

    SM0AOM Ham Member QRZ Page

    Designing a narrow-band VHF filter at an impedance level of 50 ohms results in completely impractical component values.

    At a resonant impedance of around 1000 ohms, the components become more practical and realisable.

    You are probably using some filter synthesis software.
    Just set the design impedance to 1000 ohms and redo the design, and you will find that the values become much more realistic.

    Then use capacitive voltage dividers at the input and output to get back to 50 ohms.

    73/
    Karl-Arne
    SM0AOM
     
    PY2RAF, K7JEM and AF7XT like this.
  5. SM0GLD

    SM0GLD Ham Member QRZ Page

    I added the inductor resistance to the simulation and the gain dropped to -22dB.
    The Q must be higher and C3 must be trimmed to compensate for layout capacitance. 2.4pF is a very low value.
     
    PY2RAF likes this.
  6. AF7XT

    AF7XT Ham Member QRZ Page

    You like this....

    Several times over! This was the "simple" fix to a problem that prevented one of my first HB projects from working at all.
    My Elmer at the time looked at me and " they don't teach anything at that school?"
    Seems this falls into the category of "if everyone knows about it , it's rarely spoken of "
    @SM0AOM Thanks for triggering that memory :)
     
    PY2RAF likes this.
  7. PY2RAF

    PY2RAF XML Subscriber QRZ Page

    Hey there Joseph / Karl-Arne, Mike, Dennis!!!! o/ o/ o/

    First of all THANK YOU VERY MUCH guys for your answer!!! Wow. I'm dazzled by your guys knowledge.

    Well, seems that this filter was a far long shot to me. I actually calculated the values using Bowick's RF Circuit Design book, and of course seem that I lacked a LOT more details to make this work properly.

    So seems that I need better (probably wirewound) inductors, as well Karl mentioned that it would be good to use some capacitive voltage divider on input and output.

    Thing is, I definitely TOTALLY lack the required skillset for it - Dennis mentioned that "if everyone knows about it, it's rarely spoken of", I beg to differ Dennis - DEFINITELY, in my opinion, this is far a LOT black magic, with fine and specialized required skill set. And I respect you guys a lot for it.

    Considering my despicable lack of knowledge on it, I was thinking as a B-plan to use a SAW filter - In a previous research (as suggested by my good friend @AD5GG ), I found the following compatible filters:

    GoLedge: https://www.golledge.com/media/1783/ma04600.pdf
    Vanlong: https://www.vanlong.com/download/BP60330.pdf
    Oscilent: http://www.oscilent.com/spec_pages/if_sawfilter_download/if_saw_filters_813-70-03.pdf

    Anything strong against this solution?
    If not, would you have any favourite along these three filters?

    P.s.: I'm very well aware of the high insertion loss inherent to the SAW filters. However, this can be tackled by the RTL embedded LNA.
    Besides, this is what I got from the previous panadapter version:

    [​IMG]

    13 dB loss at 69.450, which was worked nicely by the RTL's LNA.

    Besides, the filter is selectable. By default, it is in passthrough mode.

    Looking forward to hear from you guys - And THANK YOU VERY MUCH!!!!

    It will be great to sport your callsign in next board revision ;-D

    73 - Rodrigo.
     
  8. SM0AOM

    SM0AOM Ham Member QRZ Page

    These were the essences of the filter scaling exercises in tech college using filter tables normalised to 1 rad/sec and 1 ohm design impedance.

    Later, at University, Dr. Davidson lecturer in Circuit Theory, had us derive the equations behind pole-zero synthesis for a given pass-band and stop-band shape.

    Even later, the "School of hard-earned experience" taught us to make filter designs using practical and realisable components.

    Back to the topic at hand:

    What - 3 dB and -60 dB bandwidths do you require?

    73/
    Karl-Arne
    SM0AOM
     
    Last edited: Sep 1, 2019
    PY2RAF likes this.
  9. PY2RAF

    PY2RAF XML Subscriber QRZ Page

    Hi Karl!

    So, the FT-991A IF center frequency is 69.450 MHz.
    I'm using a RTL-SDR dongle, that has a 2.4 MHz total bandwidth.

    So the desired optimum characteristic is the -3 dB between 67.950 and 70950.

    For the -60 dB, a drop as steep as **possible ** :)

    For completeness: This is what I'm up to: https://github.com/rfrht/FT991A-PAT/wiki

    \o

    - Rodrigo.
     
  10. SM0AOM

    SM0AOM Ham Member QRZ Page

    This filter, scaled from a design out of memory,
    would give you about the right passband shapes.
    upload_2019-9-1_20-51-25.png


    upload_2019-9-1_20-53-55.png

    Some "tweaking" may be necessary to improve the shape, but it is a starting point.

    73/
    Karl-Arne
    SM0AOM
     
    AF7XT and PY2RAF like this.

Share This Page