Sunday, March 1, 2015

Building and Characterizing RF filters with simple and cheap tools

(UPDATED 2015-03-06)

SEE ALSO: 100 MHz upconverter, and Double Conversion upconverter

I have a need to build a steep skirted RF High Pass Filter.  The filter I'm going to try to build isn't maybe the best/perfect option, but I'm looking at it more as a test and learning experience.  So I can't claim thiis to be the perfect option.  

With that in mind, and wanting to do this as cheaply as possible here are some of the details about this little mini-project.  The FTDX-5000 has a fixed 9 MHz IF output.  
It's pretty difficult to obtain or build a HP filter with a sharp enough skirt where we don't have a high insertion loss, or not enough reduction in gain in the unwanted Lower portion of the filter when I'm  using a 9 MHz input to the single-stage upconverter.

In order to attempt push the boundaries a bit, I'm thinking I will try to use a frequency doubler/multiplier on the 9 MHz IF OUTPUT from the FTDX-5000.  This will give me 18 MHz separation between the LO and the Pass Band, instead of only 9 MHz.

This comes at a cost of -11db on the resulting 18 MHz signal from the original 9 MHz.
  •  9MHz x2 = 18 MHz using MK-3 $60.00 shipped. I don't think this will work!  I'm pretty sure this requires a very high input level that the 9 MHz IF won't have.  
    • It's probably better to think about doing a double up-conversion using a second OCXO like a 12.8 MHz = 9 MHz = 21.8 MHz instead!  These 12.8 MHz OCXO's are 2x the cost of the 100 MHz OCXO I already have.  But building the filters is much simpler since we have a larger band spread and things become a lot easier from a filter sharpness perspective.  
      • The down side is that if we're not really careful with the filters we end up using we could actually be introducing MORE spurs, noise etc...
This also means I'll have to swap out the HF Band Pass filter I've *been using* (a Minicircuits ZX75-12+) - Probably another mini-project filter to do there too I guess.  By the time this whole thing is done I'll probably had to build all of my own filters instead of getting off the shelf stuff.  The only issue with doing that is that manufacturers and re-sellers these days are really pushing SMD/SMT parts.  The last batch of parts I got were no larger than a spec of pepper.  Literally, and I had to abandon that one and reorder larger inductor coils and ceramic disc caps.

The filter I'm wanting to build is intended to reduce the 100 MHz OCXO oscillator main carrier by 55-60 db below the 118 MHz+ Pass Band I want to make use of.  Since I don't need this 100 MHz carrier in the pass band that flows into the Nooelec 0.5ppm RTL SDR Dongle I'm trying to eliminate it from being passed into the Dongle as much as possible.  The idea being that I just don't want it to de-sense the receiver in the RTL Dongle.

Honestly I'll be pleasantly surprised if this has a noticeable positive impact on the 118 MHz+ Pass band desired.

The filter I've come up with reduces <= 104 MHz by at least -65db to -69db while only reducing 118 MHz+ by about -5.53db effectively 'insertion loss' at my desired frequency.  So the net effect should be (-65 - -5.53) = -59.47db.  That seems like a pretty good reduction of the LO showing up in the input to the RTL Dongle.  Currently it's about +5db above the 109 MHz+ desired pass band where I'm not using a frequency doubler.

The new HP filter I'm planning to build will have a -5.53db insertion loss at 118 MHz.  That's
the 100 MHz LO + the 18 MHz (9 MHz X2).  For a total loss of -16.63db.  I'm not too concerned about this at this point as I have an +22db LNA available for HF (LNA4HF).

(using Iowa Hills RF Filter Designer v2.2)

So what I'm hoping this might accomplish for the upconvertion process is to reduce spur images showing up in the Pass Band.  (not that there are many at all, in fact at this point I can't really see much of that going on)  So this is just a sort of research/test and development/learning process for me.  I have an idea, and I want to test if it helps or not, or doesn't do anything all, or makes things worse.

I've already added a 41db voltage variable attenuator.  So there should be plenty of options for optimizing the RX chain.

I'll post everything about this as begin this mini-project on THIS page.  I've already ordered the parts and I'll post the parts list here asap with design details on the filter etc.

This is a work in progress updates will be coming as I get them.


  • Iron Hills Filter Designer (Freeware) & Notes [Download]
  • Touchstone RF Analyzer (Free Version) [Download]
  • Nooelec Ham-It-Up v1.2 w/Noise Source option installed [Buy Ham-it-up] & [Buy Noise Source Kit]
  • Filter Parts List
    • All Parts Numbers are from except those notes as
    • ----------------------------------------------------------
    • SMA EDGE MOUNT JACK (Female): 538-73251-1150 (INPUT)
    • SMA EDGE MOUNT (Male) 712-CONSMA013.062 (OUTPUT)
    • ----------------------------------------------------------
    • Ceramic Disc Capacitors 5%
    • Coil craft
      • 1812SMS-56NJLB Air core RF inductor 56 nH
      • 1812SMS-R12JLB Air core RF inductor 120 nH
      • 1812SMS-R15JLB Air core RF inductor 150 nH
      • 1812SMS-82NJLB Air core RF inductor 82 nH

  • Construction Photos & Notes
  • Test Results (Touchstone RF Analyzer, and resulting spectrum differences at the RTL Dongle)

Related Testing (NW0W)

TEST SETUP WITHOUT LNA(s) - Using "HAM IT UP" Noise Source
SBP21.4+ (on the HF input from NS)
ZX75HP-44-S+ and Par Electronics VHF-FM Notch Filter on the output

TEST SETUP WITHOUT LNA(s) - Using "HAM IT UP" Noise Source
SBP21.4+ (on the HF input from NS)
ZX75HP-44-S+ and Strosberg Eng. FTL201A FM-Bcast Notch Filter on the output

Related Results using Touchstone RF Analyzer + RTL 

W2AEW use of the Ham-It-Up Noise Source

Inspired by (Adam Alicajic ~ 9A4QV)

No comments:

Post a Comment