Hi Charlie, thanks for sharing your experiments with filters. I only use C0G or NP0 caps in my own filters. They cost more but as you have clearly demonstrated the performance gains for higher quality caps are well worth it. Interesting that your NanoVNA does not behave as mine does. Mine is an old one with antiquated firmware but whenever I change the frequency of the sweep then I have to recalibrate for the new span. If I don't then I get erroneous results and one consequence is that the insertion loss always appears greater than it actually is. Once I have recalibrated for the new span then the IL improves dramatically. Maybe that was a glitch which was fixed by subsequent models and updated firmware. Anyway, thanks very much as always. Keep up the great work. 73 Nick M0NTV
@@CharlieMorrisZL2CTM When you change the span you change the sample points. I believe that the nanoVNA compares the response (the "S2" part) to the pre-sampled output (stimulus). If you have not recalibrated, its stimulus samples don't coincide in frequency with the response samples, and, I believe, the frequency response calculations at those response frequencies will be less accurate. But I'm no expert!
Hi Charlie, interesting how the two filters have different characteristics, not sure how much polystyrene caps will improve it but it’s worth a try. Have you tried inductively coupling the filters? I found that I could get much better roll off with around 2.5/3 dB insertion loss which works very well at 80 mtrs. The circuit is essentially the same Butterworth 2 order filter but you use the toroids as transformers as well as inductors to up the impedance into the filter and back down again out the other end. To do this for 80 meters going from your drawing component numbers 2x C = 470pf with a 30p trimmers in parallel for tuning. C12 use 2 caps paralleled in its place 39p and 10p. The toriods in the circuit the same as you have them but with 30 turns on a T37. Don’t use the in/out caps just put an extra winding of 6 turns on each toriod one leg to ground and the other is the filter in/out. This will give you 50ohms in/out. I got the circuit from QRP Labs and it works a treat. BTW I’m using multilayer ceramic caps nothing special. Just a thought. Best Regards.
RF circuits need as low of a ground impedance as possible. I believe the extra insertion loss is due to filtering on the ground signal. You can keep everything simple and low-loss by converting from single ended to differential at the antenna. In designing differential filters, i have only used them between differential inputs to differential outputs and they have always been very simple and included an impedance connection between + and - to help eliminate common mode
Hi Charlie , thanks for sharing , Are the phase points of the toroids the same or opposite to each other, are the input and output phases of the filter different , and should the toroids be wound in opposite directions?
Great work. Q: doesn't the VNA lose its calibration settings when you change the frequency span? From my experience, for best results, VNAs require recalibration, unless someone is willing to give up on frequency resolution upon zoom-in.
Did you find that the Mix 6 had lower loss than the Mix 2 on 40 meters? BTW, polystyrene are very low temperature caps. They melt with even modest power through them. I've never seen them used in even QRP level RF output circuits. I'd use them for an LC oscillator, a sample and hold, or a timing capacitor application, but that's about it.
Typically above 100pF, the commonly available capacitors are not NP0 of C0G, so losses increase. I have seen mylar capacitors being used at 3.5 MHz. Naaah.. You have to specifically look for NP0/ C0G above 100pF..
@@CharlieMorrisZL2CTM The mylar ones are no good for RF, but polystyrene or NPO ceramic ok. I found some ones from China which were NPO but values all around the place (+/- 20%)