Monday, July 10, 2023

Martein's Bandpass Filters - PA3AKE

Above you can see the really nice 15 meter filter that I built using data provided by Martein PA3AKE. 

https://martein.home.xs4all.nl/pa3ake/hmode/bpf_all.html 

Before I built Martein's filter, my bandpass had been inadequate.  Looking at the signals coming out of the diode ring mixer in my 15-10 rig, I realized that when I was on 15, there would also be an output on 10.  And vice-versa.  These outputs would have to be knocked down by the bandpass filters.  I had been using simple dual tuned circuit filters. But when I looked at the filter shapes of these filters in NanoVNA, I could see that On 15 the 10 meter signal was only down about 20 db.  And on 10 the 15 MHz output was also down only by about 20 db.  That's not enough.  Take a look: 

Before,  with the dual tuned circuit filter

After with Martein's Filter 21.5 Mhz

AFTER with Martien's filter 21.1 MHz

While the earlier filter had provided only about 20 db of attenuation at 28 MHz, Martein's filter provided at least 68 db of attenuation.  That is really nice.  And the passband is nearly flat at 1 db attenuation. 

I built mine using some of the guidance provided on Martiens site.  I did use T80-10 toroids (I got them from kitsandparts.com).  And I did not use copper clad boards.  

One of the charming features if Martein's filters is the total lack of trimmer caps:  Martein recommends tuning the filters by simple squeezing the coils (to increase inductance and decrease frequency) or by spreading out the turns a bit (to decease inductance and increase frequency).  I did the later when NanoVNA showed that I didn't quite have all of the 15 meter phone band on the flat portion of the curve. 

I also like the way Martein provides the values for BP filters for all of the HF ham bands.  Very useful. 

Next I will build one for 10 meters.  And I will probably go back to my Mythbuster and 17-12 rigs and build Martein's filters for these rigs.  
Thanks Martein! 

9 comments:

  1. I just took a look at Martein's very-impressive page on these BPFs. His attention to the mechanical layout of the filters is especially intriguing. I don't think I've seen in the mainstream amateur literature any mention of breaking ground planes to minimize inter-resonator leakage. In fact, continuous and solid ground planes seem a sine qua non in all Ham lore. There's plenty on walling off resonators with shields (though very few actually do that), but Martein's approach seems as good or better. I guess raising the toroids off the PCB with spacers lowers leakage to ground. He also seems to prefer polystyrene caps to ceramic. Higher Q?

    ReplyDelete
  2. I agree, and I did pay attention to the mechanical construction. I used perf board with no copper cladding. I put a strip of copper clad adhesive along the bottom. I used 5% ceramic (bead) caps from Mouser and carefully measured the inductors. The resulting filter was spot-on. Very FB indeed! -68db! Woo-hoo! 73 Bill

    ReplyDelete
  3. Just a reminder that this is the same type of filter used in the BITX 40, and anyone can quickly and easily go to the rf-tools.com/LC-filter website, select Chebyshev 3rd order Bandpass Direct-coupled, shunt capacitor filter, input whatever frequency cutoffs they want, an inductor value, and the schematic and a plot of the performance immediately appears.

    ReplyDelete
  4. Good catch Walter: I hadn't noticed that a very similar filter is in the BTX40 module. But what was the URL of the web site that you referred to? It sounds very useful. 73 Bill

    ReplyDelete
  5. Hey Bill - https://rf-tools.com/lc-filter/ - same one I mentioned the other day,

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. Another good source of bandpass filter info is the Soldersmoke posting of July 16, 2021 about a Mythbuster rig. .... ..

    ReplyDelete
  8. It would be instructive to examine the performance of the BITX40 BPF as implemented on its PCB: small vertically-positioned toroids with fine-wire windings, surface-mount caps, and positioned above an unbroken ground plane. Oh, and only 1.25" long. I suppose it would be going too far to wonder whether or not there's any mutual coupling with the winding of the relay that looms hard by the filter's toroids.

    Such an examination would not find fault with the BITX40 design: it's a $50 transceiver on a 5x5" board, and it works well enough for its intended purpose. But I think in many ways there's a miniaturization bias that sits--like the Id--at the back of our minds that assumes that smaller is better (or at least more impressive) than the clunky electronics of old. Filter topologies aside, resonators made with small, low-Q components and with scant attention to undesirable coupling won't perform very well.

    ReplyDelete
  9. Todd: Great question. I pulled out a BITX40 board and a NanoVNA and did some measurements. I didn't remove any SMD parts from the board and instead just tack-soldered some test leads in. The passband looks fairly flat from 6.5 to 7.3 MHz. It shows about 4.73 db loss in this passband when I have the BITX40 board powered. Unpowered, the shape is the same, but the loss drops to 2.6 db. I then when to RF tools and asked for a third order filter of this type with a similar passband -- it came up with values very similar to those of the BITX 40. I can send you pictures. So it does not seem that miniaturization, or board placement close to the relay effected filter performance all that much. I was a bit surprised to be reminded that the filter was not followed by a TIA. 73 Bill

    ReplyDelete