Increasing the Q of a home made band pass filter for LF

[Chris Wilson]

I am about to try and increase the Q of my home brew BPF for RX on
2200 meters. Are polypropylene capacitors good for this application? I
have been using film caps and wire ended miniature chokes,
(the ones that look like 1/4W resistors)so I have
ordered some #26 mix cores in 1.3" OD hoping I can get enough turns on
them. I have the required caps in polyester film but am wondering if
polypropylene are both different and better. I need to achieve circa
616nF so have been using 47nF and 560nF in parallel. As physical size
doesn't really matter I was also wondering if part of the paralleling
could practically be done with multi gang air spaced variable caps so
I can trim the thing a useful amount to centre it where I want it. The
schematic is at http://www.chriswilson.tv/filter.jpg I am aiming at a
pass band that is as tight on 135 to 139kHz as possible.

Thanks.

[G8HQP Dave]

Almost any capacitor will have higher Q than almost any inductor. Miniature chokes usually have quite low Q. As a general rule, the bigger the inductor the higher the Q - assuming that physical size remains well below wavelength.

[Chris Wilson]

OK, thanks a lot Dave, so these cores should be a big improvement on the leaded inductors hopefully?

http://micrometalsarnoldpowdercores....-DataSheet.pdf

[kalee20]

Iron powder toroidal cores are good, but not massively low-loss. They make excellent chokes.

Chokes exist to put inductive impedance in a circuit. Usually, losses are not important as long as they don't make it get too hot! Sometimes, losses are actually a bonus, as they can stop unwanted resonances (which is the opposite of what you want).

Personally, I'd look at a ferrite core, something like a gapped RM10 size. Almost all the losses will be in the winding resistance; with iron-powder, there will be core losses. The design is not trivial, because skin effect and proximity effect will kick in.

Yes, polypropylene capacitors are good - about 10 times better than polyester (look for the tan-delta in the spec). Though as Dave says, the losses in the inductor are likely to dominate.

[Chris Wilson]

Thanks again for another reply, are you able to recommend a specific core material?

[kalee20]

I've just looked at the schematic - and assimilated the capacitor value.

Wow.

I'd read 616pF - you are actually 616nF, so looking for only 2.2uH. That's low!

At 137kHz, you have a reactance of 1.89 ohms. So if you want a circuit Q of 100, then R must be 0.0189 milliohms. That's a challenge!

With a core having a fairly low Al of 160nH, you would still need just 4 turns (this will give 2.56uH). Advantage is that you can keep the turns well apart, minimising self-capacitance (which tends to be lossy capacitance). But you'll need to use thick wire.

What material? Well, 137kHz is right in the middle of typical SMPS frequencies, so anything optimised for that would be good (Q is not an issue, but temperature rise is - and they both correlate directly to a low-loss material). So Ferroxcube 3C90, 3C92, 3C94, 3F3, 3F4 would all be applicable.

[Chris Wilson]

Please bear in mind that temperature rise is no issue, this BPF is solely for reception purposes!

[Phil G4SPZ]

There was an old rule of thumb which suggested 1.5pF per metre of wavelength. This equates to 3,300pF with a consequent increase in inductor value and L/C ratio.

[G8HQP Dave]

You might be better running the filter at a higher impedance, and then matching to 50 ohms. Tuned circuit reactances in the region of 100-200 ohms (i.e. 1.3-2.6pF/m - so the same region as Phil suggests) may be easier to make at reasonable Q.

[Chris Wilson]

Some great advice, thanks! Would there be any mileage in looking at a resonant filter or are they not tight enough?

[G8HQP Dave]

Not sure what you mean by a 'resonant filter'. Your filter already uses resonance, but requires an inductor which is difficult to make with sufficiently high Q.

135 to 139kHz passband would need a loaded Q of 34 for a single resonator, assuming -3dB edges. Perhaps loaded Q around 50 for a critically coupled pair? That means an unloaded Q around 200-300 for lowish loss, which should be achievable - or maybe loss is not a problem?

[Phil G4SPZ]

Would an active preselector like this http://http://www.alg.myzen.co.uk/radio/136/pre_sel.htm possibly be easier to build?

[MrBungle]

Throwing a random idea on the table... can you use a low noise high slew opamp for this and lose the inductors in favour of capacitors in a gyrator arrangement? Solves a lot of problems. May create different ones.

http://sound.whsites.net/articles/gyrator-filters.htm

I've built a couple which are very sharp bandpass filters around the 100khz mark based on the above.

[G6Tanuki]

1.5pF per metre of wavelength stirs some ad-hoc memories too.

There are similar ad-hoc length:diameter 'rules' for winding coils to achieve the highest Q (basically getting the most inductance for the shortest length of wire). Depends on whether you're using a single winding or overlapped windings (overlapped windings add self-capacitance which complicates the resonance calculations).

Given that it's for reception only, have you thought of wrapping a "Q-multiplier" around your coil/capacitor, just as we did with elderly radios using 465KHz transformer-coupled IF-stages back in the 1960s to narrow the bandwidth for better SSB/CW reception?

[kalee20]

Originally Posted by Chris Wilson

Please bear in mind that temperature rise is no issue, this BPF is solely for reception purposes!

Yes realise that... The point I was as making was that ferrite optimised for a low temperature rise in a switch-mode power supply, will be fairly low-loss, and will therefore be capable of a high Q in a resonant circuit.

Completely different end result, but which puts similar requirements on the ferrite.

[Diabolical Artificer]

As Mr Bungle says, an active filter using op amp's would be easier probably especially as there is a really good filter design tool available on the Texas site.

Andy.

[MrBungle]

and smaller!

[Al (astral highway)]

Originally Posted by MrBungle

Throwing a random idea on the table... can you use a low noise high slew opamp for this and lose the inductors in favour of capacitors..,Solves a lot of problems. May create different ones...

A really smart idea and pretty much guaranteed to give far better results than with LC experimentation! Nice one!

[Chris Wilson]

Some excellent help and advice here thank you all very much indeed. I have made a good improvement by using proper toroids, home wound instead of the little mini chokes and by using polypropylene capacitors. When i am feeling more adventurous I may look at an active filter too. Thanks again