I want to create a input energy dependent frequency shunt network, which goes in parallel with a loudspeaker,

At low volumes, a certain midrange frequency should be shunted off the loudspeaker. The shunt effect should go down as the power from the amp goes up, and more midrange signal reach the loudspeaker

Edit: I should have added that it's for guitar amps: Strum the guitar lightly and the filter eats some midrange and the sound from the speaker thins; hit it hard and the lightbulb's resistance goes up exponentially, so that more midrange gets to the speaker and fattens up the sound. That's the basic idea.

Series RLC variable resistance


At low voltage/current from the amplifier, the shunt network impedance should be low and reduce a certain audio band that reaches the speaker; let's say center f 500Hz, with a bandwidth of 300Hz , -25dB

As the energy coming from the amplifier goes up, the shunt network impedance should go up, therefore eliminating less signal and delivering more to the speaker.

The center f and bandwidth should stay the same as the energy goes up or down.

My idea is to use an LC band pass resonance network in series with a lightbulb. As the energy increases, the lightbulb heats up, resistance increases, which raises the impedance of the shunt network, and directs more energy to the loudspeaker.

A cold 100W bulb reads around 40 ohms, it goes up tenfold when hot. I imagine 3 bulbs in parallel would be 13 ohms when cold.

I realize the following issue:

The lightbulb's resistance apparently does not influences the center frequency but the bandwidth! When cold and measuring 10 ohms, the bandwidth is around 300Hz. When host and measuring 100 ohms, the bandwidth goes up to around 3000Hz (which I do not want).


The 16 ohm loudspeaker resistance will effectively be in parallel with the lightbulb resistance.

Would this deliver a stable low resistance, to keep the bandwidth in a narrow range?

Also the damping factor of the shunt network goes from 0.3 at 10 ohms, up to 3 at 100 ohms. Can someone explain the damping factor for me? Can this be translated to a total volume loss in dB? Or in voltage or current?

Is this whole thing even a possible approach or are there other issues with this idea?

Thank you for any help in advance :)

  • \$\begingroup\$ I think your question would actually benefit from describing why you want that – as a DSP person, I see a lot of questionable effects of what you want to do. \$\endgroup\$ Commented Jul 6, 2018 at 14:24
  • 2
    \$\begingroup\$ I don't think you have thought this through: In order to make a 100W light bulb "hot", you're going to have to waste 100W of your audio power in it! Do you really have that kind of power to spare? Not to mention that you're going to have on the order of 200V across your 16-ohm speaker -- can it handle 2500-3000 watts? It seems like you'd be much better off processing the line-level signal before you send it to the power amplifier. \$\endgroup\$
    – Dave Tweed
    Commented Jul 6, 2018 at 14:30
  • \$\begingroup\$ hey marcus thats propably a good idea, its going to be a novelty effect for guitar amps, strum the guitar lightly and the sound from the speaker thins, dig in hard and the midrange increases and the sound fattens up. \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 10:31
  • \$\begingroup\$ hey Dave, i look at it this way, when the bulb (or bulbs in parallel) is cold, it should have a low resistance similiar to the speaker, order to drain some of the midrange which its tuned to, when the strings are hit hard, it needs only a little power to raise the resistance in the bulb alot - compared to the speaker, in effect directing more sound to the speaker if this makes sense, a 100w bulbs resistance goes up to over 500ohm, just a fraction of that would be enough to show the desired effect i suppose \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 10:34

3 Answers 3


Here's where the problem starts: -

enter image description here

The output impedance of most audio amplifiers capable of delivering 10s to 100s of watts is going to be below 1 ohm. This means that if the L and C impedances perfectly balanced out (mid band) then you'd be loading the amplifier with a lightbulg and, guess watt (my little pun)... it won't make hardly one iota of difference - the amplifier will still deliver pretty much exactly the same amplitude to the speaker mid band as it would without the light bulb.

The light bulb will waste power of course.

If you want to cut midband you need a parallel tuned circuit wired in series with the speaker and what you then want is to lower the resistance across that parallel tuned circuit as power increases (the opposite to what a lamp does).

  • \$\begingroup\$ hey andy, its going to be for guitar tube amps, which have an output transformer to match the low speaker impedance to the high impedance the tubes like - which is in the thousands of ohms, does it qualify as a high imp output when there is a transformer involved? the lightbulbs role is supposed to eat power when the guitar is strummed lightly, when hit hard the power goes up, resistance in the bulb goes up and more midrange goes to the speaker and fatten up the sound thats the basic idea, a bandstop in series is interesting, how to do it? LEDs that open at a certain spot? \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 10:36
  • \$\begingroup\$ The geetar amp has probably got feedback internally to stabilise it and linearise it and that, irrespective of tech will make the output impedance low. If you have an amp schematic this can be judged of course. Yes LEDs could do the job but better still are series back to back zener diodes across the LC network. \$\endgroup\$
    – Andy aka
    Commented Jul 7, 2018 at 11:02
  • \$\begingroup\$ yes most guitar tube amps have NFB, but way less I imagine than HIFI. thanks for pointing out the zeners over LEDs, I have some questions but need to think it through first and come back, thanks!! \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 11:37
  • \$\begingroup\$ one more thing, does the speaker resistance influence the tuning frequency of the RLC network at all? there are several overlapping formulas I found, notch filter for a speaker which looks basically the same but takes the speaker in account, while other RLC filter calculators dont take this in account, also is it the same as a tank filter??? \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 12:15
  • \$\begingroup\$ It is a form of tank filter and yes, the speaker impedance does play a role in how much is notched out of the midband. \$\endgroup\$
    – Andy aka
    Commented Jul 7, 2018 at 12:56

You need to rethink your architecture. I assume what you are trying to achieve is a color organ (music to light) device.
The power supply frequencies used to drive the 'bulb' are mostly irrelevant, it's the frequency detection used to trigger the supply to the bulb that is important. This can be done at low signal levels using simple op-amp RC filters (no large inductors).

Be aware that the visual response of a conventional light bulb (whether 120 or 240 V) is very slow. These days you would get much better visual performance using power LED strip lights and the voltages used are much safer.

There are plenty of bouncing ball projects (here and here as examples) for light organs using op-amp based or chip filters (particularly clever use of the MSQEG7 based filter with an Arduino in the second). And if you are looking for high accuracy adjustable filtering you can use a DSP (FFT) to sample/analyze the audio signal.

  • \$\begingroup\$ hey Jack, its going to be a novelty sound effect for guitar tube amps, when the guitar is strum lightly the bulb eats some of the tuned frequencies, when hit hard the resistance in the bulb goes up exponentially directing more of the midrange to the speaker and fatten up the sound, I imagine it to be a stand alone unit to put ontop of your amp which kind of looks cool with the bulbs glowing :) \$\endgroup\$
    – maxbauer
    Commented Jul 7, 2018 at 10:41

ok here is the updated version:

the reactive mid shunt involves several lightbulbs in parallel (which will be switchable to adjust) for a total resistance of just a couple ohms - cold. energy from amp increases - bulb gets warm, resistance goes up, mid shunt becomes less efficient

the mid-expander network is supposed to work in conjunction with the mid-shunt, its basically just an open gate at this moment, might improve it with diodes maybe? energy from amp goes up and this route becomes less resistive letting more mids to the speaker

the compressor network also involves several switchable bulbs in parallel, to compress the highs and lows only which can be a desireable sound effect on a guitar amp,

I notice now there is a minor fault, the highs and lows in the compressor network should be -200Hz and +750Hz of course, and not 250/1500, or there would be a gap in the frequencies let through to the speaker.

the desired sound effect is, a more bass heavy and sparkling top end sound when played soft/clean, hit it hard and the mids shall be more prominent and highs and lows compressed, for a fat midrangy sound, very reactive and dynamic to the playing style.

here is a good tool I found : https://rf-tools.com/lc-filter/

the values of the components are just to fill in the blanks basically, I could not as of yet work out ways to combine these filters and calculate numbers that make sense so far... if someone could assist would be highly highly appreciated!!! ;):)

the big unknown is how much the lamps resistance will increase from lets say the amps output goes from 5w up to 30w or higher (these low wattages are perfectly normal in guitar tube amps even for big rock bands)

enter image description here

I dont have membership at circuit lab so I cannot run the circuit that I designed in that program, its rather expensive haha, or maybe it would be worth it?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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