# Why signal coming out of a Constant K filter takes 0.5ms to stabilize?

I made a simulation of an audio amplifier with an input signal of 200 microVolts.

The question is why my filter does not deliver the desired output, which operates between 512 Hz and 2048 Hz.

From the Bode plot, I get a gain of -7dB which I assume it's from the matching impedance from input and output which is later corrected with an amplifying op-amp.

With the values from the schematic, running a transient analysis, I get the following data:

The larger signal is the input and the smaller one is the output on R24. We can see that at 0.5 ms the signal is not perfect, why does that happen and how can it be corrected?

• The settling time of the filter output is related to its bandwidth, which is only 1536 Hz in this case. – Dave Tweed May 21 '20 at 13:03
• If you look carefully at the positive and negative peaks of the output waveform you can see that the startup transient hasn't died away until the start of the fourth cycle at 3 ms (at least to within the width of the trace). The startup transient dies away exponentially so each output cycle is affected less and less until you reach the steady-state response. – Graham Nye May 21 '20 at 14:51

Why signal coming out of a Constant K filter takes 0.5ms to stabilize?

and

why does that happen

It's called the transient response - you can't expect a filter to immediately start producing the steady AC response from an initial application of a sinewave. At the point of initially applying a sinewave, the input signal goes from a steady value of zero to a half sine pulse. The filter does not know in those first few moments that what is being applied to it is a sinewave. It cannot predict the future hence, it does what it does and eventually settle to the steady-state AC response (as all filters do).

how can it be corrected?

It can't. It is what it is.

• Is it a viable solution in real word to use this type of filter in an audio amplifier? Wouldn't there be sound distortions or clipping? – tgarmp May 21 '20 at 11:57
• So you are saying that this is normal to happen. – tgarmp May 21 '20 at 11:58
• This is normal - sure, the waveform is distorted but so is the output compared to the input when you make a filter. – Andy aka May 21 '20 at 12:00
• Thank you for the answer and is my assumption correct regarding the -7dB? – tgarmp May 21 '20 at 12:01
• Yes, the -7 dB is coming from a 6 dB attenuation of the input formed by the potential divider R24 and R23. The extra 1 dB attenuation might be from the reactive components producing a slight attenuation. – Andy aka May 21 '20 at 12:04