2

I'm building a filter from a kit with some given (microscopic pre-soldered SMT) capacitors and 2 toroids. It looks like the notch frequency is a bit too low with 10 wire turns around a T37-2 iron powder toroid and a bit too high with 9 turns. The windings have all been wound tightly and evenly spaced.

Is there some variation in toroid winding method that could get the inductance closer to halfway between 9 and 10 turns?

Possible alterations are winding the toroid with uneven turn spacing (turns scrunched to one side or the other, or to the top away from the PCB or to the bottom closer to the PCB. Winding with fatter wire or thinner wire. Winding with a twisted wire pair soldered together at the ends. Winding all the turns looser (by using a spacer during winding). Winding some of the turns loose and some turns tight (all on one side or alternating?).

Which of those strategies might I try to tune my filter? (without throwing away the parts and PCB and starting over with a different design...)

hotpaw2
  • 4,731
  • 4
  • 29
  • 44
  • Either change the capacitors, or change the inductor's core. – Hearth Jun 23 '22 at 04:58
  • Are there toroids that come with a range of such small differences in permeability? Or can a toroid be machined to change its wound inductance by a small amount? (a fraction of one turn delta) The tiny SMT capacitors are too small for me to even see, much less change. – hotpaw2 Jun 23 '22 at 07:29
  • You can try perhaps coupling" a new "winding" with "some" load (capacitor or resistor). Simulate first to see what happens. It is better if you know the schematic of the filter. – Antonio51 Jun 23 '22 at 08:56
  • 1
    @hotpaw2 I think you are missing hearth's point. Use a core with less change in inductance per turn. Why are you stuck on 9-10 turns? Or put another way, why don't you disclose what you are doing and why and what your design process looks like? The answer will fall out from those details soon enough. – jonk Jun 23 '22 at 08:57
  • The answer may depend on toroid core *permeability*. Low-permeability (\$2 < u_{rel} < 20\$) cores like iron-powder toroids might accept adjustment of scrunching turns together to increase inductance *better* than high-permeability cores. – glen_geek Jun 23 '22 at 11:16
  • @hotpaw2 I'm not talking about using a core that gives you the right inductance with 9 or 10 turns. I'm talking about using a core that gives you the right inductance with 90 or 100 turns. – Hearth Jun 23 '22 at 13:17
  • It's not my design. It's a kit with provided components. The instructions are to wind a T37-2 iron powder toroid with 10 turns. I checked the resulting filter response with a NanoVNA, and it's a bit off. I'm trying to see if I can improve the build results of the provided kit. Not create a new design or kit. – hotpaw2 Jun 23 '22 at 17:14
  • 1
    What is preventing you from doing 9.5 turns? – evildemonic Jun 23 '22 at 17:28
  • I've never seen a description of how both ends of half a turn through a toroid could get connected to a flat PCB. Maybe in a 4 or more dimensional or sufficiently non-Euclidian space? – hotpaw2 Jun 23 '22 at 19:09
  • Make a u-turn right before the last pass-through. Both leads end up on the same side of the toroid. You could also drill a hole through the toroid and pass the lead through that, giving only half its flux to the core. – evildemonic Jun 23 '22 at 19:28
  • This question shows a 2.5 turn coil (not exactly a toroid, but shows the right idea): https://electronics.stackexchange.com/questions/431238/ferrite-beads-with-turns – evildemonic Jun 23 '22 at 19:32

2 Answers2

2

It looks like the notch frequency is a bit too low with 10 wire turns around the toroid and a bit too high with 9 turns.

Without a schematic and some pieces of information, I can just guess this.
Here is an example of what can be done in my particular circuit.

Just add a coupled inductor (1 turn or 2) with a capacitor and adjust the value of this capacitor until the notch slip at the right place ...
NB: the right curve is for the "lowest" value of the capacitor.
Increasing the value "slip" the notch to "lower" frequency.

enter image description here

Antonio51
  • 11,004
  • 1
  • 7
  • 20
  • I like this idea because one could glue a tiny trimmer capacitor nearby with a one turn coupling, and (non-conductive) screwdriver tune out some of the component variation without desoldering anything. – hotpaw2 Jun 23 '22 at 22:33
2

The position of turns matters. How much, depends on the mu_r of the core.

Micrometals has some reference material on the matter:
https://s3.amazonaws.com/micrometals-production/filer_public/7e/d0/7ed096a0-fe6e-4df1-9da9-e129c1ee73d2/q_curve_catalog_issue_h.pdf

[this is a current link from their website, but it doesn't look too permanent; anyone got a better one?]

Specifically the diagram on page 4, but you may find the surrounding information helpful as well.

If your toroids are bi-colored, you may be able to identify them -- most manufacturers seem to follow Micrometals' color scheme, or indeed are produced by/for them -- and then find specs. Typical performance can then be found here, or in related documents on their website.

Tim Williams
  • 22,874
  • 1
  • 20
  • 71
  • 1
    Moving the turns might be a good solution. In figure D on page 4 of the Micrometals documentation, http://www.iec-international.com/micrometals/micrometals/downloads/Q%20Curve%20Catalog%20Issue%20H.pdf , it shows that, for a small number of turns, scrunching the turns together to use only one third of the toroid can increase the inductance by over 50%, with lesser increases of inductance for lesser scrunching of the turns, compared against the inductance from evenly spaced turns. – hotpaw2 Jun 23 '22 at 17:20