6

When building filters for ham bands (direct sampling SDR frontend) - it is common to build filters using 2,6,10,12 material toroids. Are there any recent developments that offer lower losses / higher-Q in amateur bands? (6-160m)

Are there any other tricks to increase Q? For example, I guess for 160m band it is possible to wind coils with litz wire. Any other similar high-Q tricks, especially ones suitable for 20/30m bands?

If no better core materials are available, is there anything could be done with the wire itself? It is clear that it must be single layer. But beyond that - maximize diameter of the wire or there must be some air gaps left? Flatten wire in the inner part of the coil? What is maximum frequency where Litz wire can still help in increasing Q?

Mike Waters
  • 7,843
  • 4
  • 16
  • 50
BarsMonster - R2AYN
  • 376
  • 3
  • 14
  • That's an interesting question! Sadly, I don't even know what realistic Qs and inductances for these bands with the standard toroid cores are. – Marcus Müller Dec 17 '22 at 23:05
  • by the way, could someone comment what the common toroid materials *are*, that would be interesting! I'm guessing powdered carbonyl iron cores, but that's just a shot in the blue. – Marcus Müller Dec 17 '22 at 23:21
  • 1
    @MarcusMüller commonly the T50-6, T50-2, T37-6 and T37-2 cores are used in HF filters. Additionally mix 43 is a common choice for transformers in power amplifiers. When it comes to the filters in my humble experience mix 6 is a good choice for pretty much any HF bands. However mix 2 gives a bit more inductance for the same number of turns in expense for somewhat less temperature stability. – Aleksander Alekseev - R2AUK Dec 18 '22 at 11:33
  • @AleksanderAlekseev-R2AUK thanks! Interesting! looking up T50-6 cores, I find they have a reference L of say 4 to 12 nH per windings squared - that's a lot of windings if you want roughly 50 µH (?? Wild guess for 160m, based on 180 pF caps being pretty common and still far away enough from parasitic effects) of inductance. interesting, my feeling for orders of magnitude was way off here! – Marcus Müller Dec 18 '22 at 12:36
  • 1
    @MarcusMüller inductance is a square function of the number of turns: L [uH] = (Al * pow(Nturns, 2)) / 1000 where Al for T50-6 is about 4. So for 1 turns you get 0.004 uH but for 20 turns you get 1.6 uH. Practice shows that this is enough for most applications. For a 9-order 80 meters LPF I glued two T50-6 once in order to increase the inductance. Using T50-2 or T68-2 cores with Al = ~5 and ~6 is also an option. – Aleksander Alekseev - R2AUK Dec 18 '22 at 13:59
  • Great question! I wonder if the better and better bit depth, IP3 and SFDR of ADCs makes the analog filter less and less relevant. You can realistically sample the whole band with enough dynamic range for all but the most extreme nearby-transmitter conditions. Perhaps if the antenna is a short whip, some low pass would help for receiving at a few MHz – tomnexus Dec 19 '22 at 09:54
  • 1
    @tomnexus Due to poor noise figure of (best) ADC's you'd need to add at least ~10-20dB of gain before ADC to touch the fundamental limit. And this is where you'd get overloaded immediately without filters. – BarsMonster - R2AYN Dec 19 '22 at 11:30

1 Answers1

3

To my knowledge there were no recent discovers in the area of ferrite materials. Mix 6 is pretty much as good as you will get when it comes to building low-pass / band-pass filters for HF bands. I believe one of the reasons may be that even if you invent something better it will not present a valuable product for the market. Current low-pass and band-pass filters are good enough.

Lower losses and higher Q can be obtained with quartz crystals though. A Q of 100 000 is quite common and sometimes you can get crystals with Q of 150 000 and more. Quartz crystals are commonly used for building narrow band IF filters. For instance it's not difficult to build an band-pass filter with 2-3 kHz bandwidth and insertion losses of 1-2 dB. Also there are filters available from many manufacturers for common IF frequencies if you don't want to build one.

Aleksander Alekseev - R2AUK
  • 4,262
  • 12
  • 48
  • 1
    Excellent reference to other-than-LC-filtering :) Note that the fact that you [get](https://www.digikey.de/short/7wv2h8rm) a lot of quartz filters at a lot of frequencies, but not necessarily at the frequency your signal is at sends OP straight down the superheterodyne receiver rabbit hole – but having read his [blog](https://3.14.by/en/read/amateur-radio-call-sign-R2AYN-FT8) once in a while, I think they can deal with complex technical tradeoffs :) – Marcus Müller Dec 18 '22 at 12:50
  • 1
    Yes, quartz filters are awesome! In my case it's not IF filtering, so going with quartz is somewhat harder... I am doing bandpass filters for direct sampling SDR. – BarsMonster - R2AYN Dec 19 '22 at 06:04
  • @BarsMonster-R2AYN If the goal is to get a faster transfer between the pass band and the stop band you may be interested in trying elliptic filters. They use the same LC components but the transfer is much faster in expense of the higher ripple in the stop band. I don't see why else you may be interested in alternative ferrite materials. Building a band pass filter with insertion loss of 1-2 dB is quite a trivial task. – Aleksander Alekseev - R2AUK Dec 21 '22 at 10:28