1
\$\begingroup\$

I would like to build a passive band-stop filter for my earphones. I recently had some help from the user "Andy aka" with designing that type of filter and he also pointed me to the LTSpice application for running simulations.

After playing with some values, I think I have what I want, but i have a few questions before I go and purchase the necessary components.

Here is the circuit:

filter circuit

And here is the list of components from mouser:

mouser list

I would just like to know if I'm missing anything or if these components and graph look correct?

I want to cut the 5800-6000hz region by roughly 7.8-8db with a Q between say 14 and 17. The resistor R2 in the schematic is there in place of an adapter I will connect at the earphone jack (R3). This is because my audio interface is 32 ohms. If I use the filter on that interface I remove R2 and the values of the graph remain the same. When using a 1ohm device like my portable amp, the adapter brings the graph values back in place.

I just want to verify that this should work and I have all the components I need. I added an extra components as a fail safe. :-P

Also, are any of the other values critical with the components? I tried to get the best tolerances I could for each and the highest reasonable power handling...

\$\endgroup\$
1
\$\begingroup\$

The 6.8uf electrolytic capacitor is not a suitable choice for an accurate band pass filter like this, note that the tolerance (spec sheet) is 20%. Most electrolytic caps are also polarized parts (needing a DC voltage) and are not normally a good choice for use in a series connected filter like this (even though you show a polarized voltage source as the audio driver). A much better choice for accuracy would be to use a film type capacitor (polycarbonate, polyester, polyethylene, etc). These can be found with better tolerances and are not polarized.

The wire wound resistors are not the best choice either, even though they can be tight tolerance wire wound resistors have some inductance (some types try to limit this) this would throw off your calculations a bit. A carbon film resistor with a tight tolerance is a better choice. Also using a 3w power rating for the resistor is very much over-kill for an earphone circuit. A 1/4w 1% resistor is likely fine here.

Also note that the accuracy of your stop band center is affected by the tolerance build up of all the components. Your design requirement is asking for a stop band in the range 5800-6000hz, this is only a 1.7% range at the center value of 5900hz. Note that your inductor is already at a 5% tolerance, the other component tolerances could further increase the offset error to the center frequency .

\$\endgroup\$
1
\$\begingroup\$

The way to design precise filters in the absence of parts of sufficient tolerance is to use adjustable components. For example, if you need a 47+/-1% pf cap, you would use a smaller cap and put an adjustable capacitor in parallel to make up the difference.

Same with inductors, only the adjustable inductor would go in series w/ the original.

Then, of course, you need a way to tune your circuit to the right value. The easiest way would be an RLC meter, but you may need to get more creative than that if you don't have one. Maybe make some sort of resonant circuit with a tight Q, and look at the resonant frequency, tweaking until its at the right place.

So, buy the resistor at the tolerance you need. If you can't get tight enough caps and inductors, use trimmer caps and adjustable inductors

\$\endgroup\$
  • \$\begingroup\$ thanks guys. these are exactly the answers i was looking for. it seems i am not knowldgeable enough yet for this task perhaps :-/ \$\endgroup\$ – luisdent Jan 21 '15 at 18:56
  • \$\begingroup\$ Trimmer caps and adjustable inductors are great at RF frequencies, however at audio frequencies (as being used here) the size of such components would be quite impractical. The parallel and series placement of extra components is the best idea here. (A multi-tap inductor might have some merit, but likely still at a high cost.) \$\endgroup\$ – Nedd Jan 21 '15 at 19:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.