I realize that this project will NOT be to the liking of many. Some will argue that it is too simple and that I should get on the SDR bandwagon. Others will say that it is too complex, and that I should limit myself to simple thermatron-style CW rigs. To all of these critics, I say "HA!" This is a hobby and I will build what I want to build! "Ha!"
It all started with the VFO from a Kenwood TS-820S. I bought this on e-bay, hoping just for the anti-backlash, reduction drive mechanism. But as with the Yaesu FT-101, the sellers -- perhaps unknowingly -- sent along the entire VFO. I described all of this back in October: https://soldersmoke.blogspot.com/2026/10/a-tale-of-two-old-analog-vfos.html
One big advantage of the Kenwood VFO was its lower frequency range: 5-5.5 MHz. I looked around for a suitable IF frequency for 17 meters. 13 MHz seemed good. AI promissed no birdies.
So I started fooling around with crystal filters. My adventures in October appear here:
https://soldersmoke.blogspot.com/2026/10/about-limits-and-problems-and.html and
https://soldersmoke.blogspot.com/2026/10/using-ai-to-design-crystal-filters.html
This week, while rummaging around in my junk box, I found the Mostly DIY RF board for an 8 pole QER filter. (Thanks Todd) So I ordered some more of the same small 13 MHz crystals from Mouser:
I should point out that the Mostly DIY RF QER filter would work fine if the signal reaching the filter was LSB (either LSB alone, or USB after sideband inversion).
I aready had a 10 pole Cohn filter built with these crystals. I looked at the bandpass with my Antuino:



You did an excellent job matching the input/output impedances to get those nice flat passbands. I've heard some concern over the use of the short HC49US crystals, but as long as their ESR is low (certainly under 100 and preferably around half that) they work the same as the taller HC49U type *for filters*. There's reports they can't be pulled as far as the taller type in VXOs, though. It could be that there's been manufacturing improvements in the short xtals as it's easier in the last few years to find low-ESR types.
ReplyDeleteBTW, that's one of the original run of prototype PCBs you have there. I called them "bandaid" boards because the clear soldermask over copper made them tan colored. Thanks for the MDRF link.
I think that’s a cool project.
ReplyDeleteChuck, W7VX
Fancy test instruments are interesting, but to me the most important criteria for SSB is how does your radio sound to other hams? Certainly the BITX20 and BITX40 produce legible audio using only 4-crystal filters for either LSB or USB, although with a narrower than ideal bandwidth. However, the more complex filters can achieve the target 2.7 KHz bandwidth which some younger hams with better hearing can appreciate. Certainly I have used rigs with 4, 7, and 8 crystals for my 2000 POTA contacts and nobody has ever complained about my signal, but I have received several compliments. All of my crystal filters are homebrew and were tested using only a frequency counter and DMM readouts, so although crystal filter building can be a bit tedious I consider it to be one of the least difficult aspects of homebrewing. --Walter KA4KXX
ReplyDeleteThe Cohn filter doesn't look particularly good. The beginning of the passband is quite flat but at the end large ripple is visible and insertion loss increases by up to 8 dB just because two 22 pF caps were omitted. Port impedance looks off too considering filter bandwidth.
ReplyDeleteThe crystals in the QER filter have a larger electrostatic capacitance, hence the more shallow lower filter slope. Significant passband ripple is visible, which suggests the filter may not be terminated correctly.
I think -- for my purposes -- the Cohn filter looks quite good. Sure it is a bit wide and as you point out the flatness deteriorates at the high end. But I will not be using this high end -- I will not be using it on transmit or receive. I suppose the matching could be a bit better, but I think it is good enough. In homebrewing, we can't let the perfect be the enemy of the good. You talk about "significant passband ripple" and say that the matching in both the Cohn and QER filters is less than perfect. How many filters have you homebrewed? (You are anonymous in the comment.)
ReplyDelete