Low Cost PCB's Low Cost Components

Author Topic: Possible to convert Commercial 900mhz Interdigital Duplexer to 1420mhz bandpass?  (Read 486 times)

0 Members and 1 Guest are viewing this topic.

Offline rwgast_lowlevellogicdesin

  • Frequent Contributor
  • **
  • Posts: 501
  • Country: us
    • LowLevel-LogicDesign
Ive had this duplexer laying around for a while it came out of a 1980s car phone setup I found at the thrift shop. I have no use for it as is but I need a good 1420mhz bandpass filter to stick in between amplifier stages. Im not the most talented person when it comes to building stuff like cavity filters nor do I want to spend a week building and mucking around with an interdigital filter. So I was hoping I could use this guy some how, maybe by removing the center sheild plate. Im not even sure where to connect the hardline inside. Here is a pic any info on how to turn this into a BP filter would be greatly appreciated!

Link to Picture

Im really not sure where the coax would couple to0, and usually a BP filter would have the post tapped in a different design..

Offline evb149

  • Super Contributor
  • ***
  • Posts: 1568
  • Country: us
That construction doesn't look very easily modifiable to me.  Maybe something to do with the tuning plunger lengths... Seems like it would be easier to start from scratch.  Maybe make a coaxial stub based filter or something if you don't want to build something interdigital.
Really you'll have to define your requirements for bandwidth and roll-off rate and so on to get a meaningful sense of what approaches will most easily implement your requirements.
You can just buy off the shelf resonators though if you need a much wider banddwidth then you might need offset tuned resonators or at least a way to "de-Q" a single one.

You could even get away with lumped element at 1420 MHz.. some good C0G caps / piston trimmers or gimmick capacitors, wirewound SMD or air wire inductors....


Offline rwgast_lowlevellogicdesin

  • Frequent Contributor
  • **
  • Posts: 501
  • Country: us
    • LowLevel-LogicDesign
So the plan is to have 1 .5NF 22db @1420 mhz lna on the back of each helical in a broadside array. Then use two PSA4 5043 mmic's .8-6NF and 18DB of gain each in series as a line driver and then maybe one more LNA right at the receiver. All amps and airspy will be temp stabilized using TECs, so gain and component drift wont really pose an issue.... in theory.

In order to do lumped element filters I think I can get away with using trimmers and 1pf caps in parallel and straight pieces of wire for the 10nH and below inductors. I feel like this should work out just fine manhattan style down a 50ohm microstrip on a piece of rogers. These filters combined with coax stubs should work just fine to keep the line drivers from causing IMD, but Im concerned about what to stick right behind the LNA's coming from the helix array, I need these to be narrow. Id really like to add some filtering in front of the first LNA but I have no idea what to put there as far loss and noise temp goes.

Offline ConKbot

  • Frequent Contributor
  • **
  • Posts: 974
What is your budget? In the terms of RF things, custom filters aren't that expensive, and some vendors have no problem doing a 1-off, as it's just retuning an existing design to a new center.  Once you figure out your bandwidth, minimum noise floor, and your first gain stage nf you can specify your maximum insertion loss and have a start at a decent spec.  In general though a sharper or deeper cutoff, or a more wideband filter will lead to a higher insertion loss in the passband, so don't over-specify

Offline rwgast_lowlevellogicdesin

  • Frequent Contributor
  • **
  • Posts: 501
  • Country: us
    • LowLevel-LogicDesign
Well I pretty much have everything I need already except the motors for an AZ/EL system so i'm not trying to spend a lot on filtering. I also live in an RF quiet area I only have two TV stations that broadcast strong signals near me at 525mhz and 6??Mhz, so primarily my only need for filtering is to get  hi amplification without the LNA's overloading.  I think my best bet may look something like two stubs to notch out the TV stations in front of the LNA, with a decent filter behind that LNA.

So here is where im getting a bit hung. First off when I add coax notch filters to get rid of the TV stations I get no losses at 1420mhz what so ever, as a matter of fact with enough notching I seem to receive an extra 3db of signal, although I have not done a through check to see if there are any SNR gains. What I don't understand is how to figure out if a homemade filter is adding to the noise temp, is there a way to easily figure out a filters NF. Instead of stubs an easily made coax cavity may offer better performance but I cant seem to figure out whether these filters also operate on harmonics like a coax stub, i.e a filter like this


would anyone happen to know if these notch/bp only the frequency of interest or the frequency + its harmonics?

For the filter behind the LNA I would like it to have losses under 6-8db @1420mhz, the lower the better obviously. Id like it to be sharp like a 7th order constant-k with a bandwidth 0f 300mhz to 50mhz. A saw filter would work just fine right here but they dont make a whole lot of 1.42ghz Fc filters from what I can find. About the only filter I can find to uy at 1420mhz is this which includes an LNA as is way out of budget!


Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 8687
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
All filters have higher modes, it's just a matter of shoving them to impedances or frequencies where they don't matter.  A cavity resonator is definitely no exception. :)

It seems you'd need a 3/2 mode relationship though, which isn't very likely to happen (a 1/4 wave resonator's next mode is 3/4 wave, give or take dispersion).

Noise factor relates to loss.  If there's no power loss, then there's no noise factor.  Note that loss is not insertion loss, because insertion loss is internal loss plus reflected power.

If you're getting negative insertion loss (i.e., gain instead), it's probably not very noisy.... :-DD

Seven Transistor Labs, LLC
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!

Online medical-nerd

  • Regular Contributor
  • *
  • Posts: 62
  • Country: gb

As I understand it the first LNA contributes the most to the noise figure of the whole receiving chain, why then are you putting amplifiers with NF 0.8 behind a LNA with NF 1.5 instead of the other way around?


Offline LaserSteve

  • Frequent Contributor
  • **
  • Posts: 566
  • Country: us
I find this software plus 3/4ths inch wide 1/4th inch thick brass strips from McMaster-Carr  make it easy...  Kick in 0.25 more dB of loss if you use brass instead of OFHC copper rods..  Its very error tolerant, especially when you add in a few adjustment screws.


From one of the on-line calculators, ie WA4DSY:

Design data for 3 section interdigital bandpass filter.
Bandpass ripple = 1.000 DB
Center frequency = 1420.000MHZ
Cutoff frequency = 1416.500 and 1423.500 MHZ
Ripple BW = 7.000 MHZ
3 DB BW = 7.664 MHZ
Fractional BW = 4.930 MHZ
Filter Q = 185.280
Est QU = 1966.202
Loss Based on this QU = 2.257 DB
Delay at band center = 114.624 NS
                                                                   Freq  DB loss
                                *                                  1410.0   32.7
                               *                                   1410.5   31.3
                             *                                     1411.0   29.7
                            *                                      1411.5   28.1
                          *                                        1412.0   26.4
                        *                                          1412.5   24.5
                      *                                            1413.0   22.5
                    *                                              1413.5   20.2
                 *                                                 1414.0   17.7
              *                                                    1414.5   14.9
           *                                                       1415.0   11.8
        *                                                          1415.5    8.2
    *                                                              1416.0    4.3
 *                                                                 1416.5    1.0
 *                                                                 1417.0    1.0
 *                                                                 1417.5    1.0
 *                                                                 1418.0    1.0
 *                                                                 1418.5    1.0
 *                                                                 1419.0    1.0
 *                                                                 1419.5    1.0
 *                                                                 1420.0    1.0
 *                                                                 1420.5    1.0
 *                                                                 1421.0    1.0
 *                                                                 1421.5    1.0
 *                                                                 1422.0    1.0
 *                                                                 1422.5    1.0
 *                                                                 1423.0    1.0
 *                                                                 1423.5    1.0
    *                                                              1424.0    4.3
        *                                                          1424.5    8.2
           *                                                       1425.0   11.8
              *                                                    1425.5   14.9
                 *                                                 1426.0   17.7
                    *                                              1426.5   20.2
                      *                                            1427.0   22.5
                        *                                          1427.5   24.5
                          *                                        1428.0   26.4
                            *                                      1428.5   28.1
                             *                                     1429.0   29.7
                               *                                   1429.5   31.3
         Quarter wavelength = 2.078 inches
Length of interior elements = 1.867 inches
     Length of end elements = 1.870 inches
         Ground plane space = 0.750 inches
               Rod diameter = 0.250 inches

End plates are 0.500 inches from center of end rod.
Tap external lines up 0.046 inches from shorted end.
Line impedances: End rod: 78.536, other: 80.355 Ext. lines: 50.000 Ohms
Dimensions (inches)
   El. No.  End to C    C to C      G[k]   Q/Coup
         1     0.500     1.498     2.024     0.644
         2     1.998     1.498     0.994     0.644
         3     3.496     0.000     2.024     2.216
         4     3.996

Box inside dimensions: 2.078H x 3.996W x 0.750T  inches

You have to, of course, optimize it before you build it, the above is an example of 3 iterations using hardware store rod sizes.

If you have nothing to align it with, I find High Sierra's harmonic generators kick ass, and have good value for the price,  especially the one with an external frequency input:



« Last Edit: August 09, 2017, 12:35:02 AM by LaserSteve »

Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo