(variable) low-pass filter

[xmat]

Hello.

I want to build a filter for audio signals, that does the following job:

It takes audio and filters high frequencies in a portion related to the audio's amplitude.

Think of it as a flexible/variable low pass filter; the louder the audio incoming, the more high frequencies are filtered out.

Do you have anything like that to suggest?
Can you direct me towards a specific (analog) signal processing technology?

Thanks in advance,

xmat.

 

[mstechca]

I want to build a filter for audio signals, that does the following job:

I think most filters today are for audio signals.

It takes audio and filters high frequencies in a portion related to the audio's amplitude.

I don't think such a filter is possible.

Think of it as a flexible/variable low pass filter; the louder the audio incoming, the more high frequencies are filtered out.
Do you have anything like that to suggest?
Can you direct me towards a specific (analog) signal processing technology?

I got a question. Why can't you just make a high-pass filter? This filter prevents frequencies above some number from passing through regardless of the volume. I have no clue as to why you want the high frequency signals to go through at low volume.

BUT, you can use the following equations to your advantage:

Filter frequency = 1 / (2 * pi * sqr(inductor value in henries * capacitor value in henries))

Once you get the frequency, use the following equation:

Filter circuit impedance = 2 * pi * Filter frequency * inductor value in henries

This impedance (or AC version of resistance) varies with frequency. A resistor in series with a speaker reduces volume. If you put these two factors together, you might be able to work something out.

Other than that, the only way you can pull this one off is with the use of a microcontroller (if even that).

 

[Russlk]

Do a Google search for "tunable low pass filter". Some digital filters can be "tuned" by the clock frequency. You would convert the amplitude to voltage and the voltage to frequency.

 

[audioguru]

National Semi used to have a Dynamic Noise Reduction (DNR) IC. It was a variable frequency lowpass filter than was wideband for loud signals and cut the highs (hiss) with low signals. The opposite to what you want.

They were trying to compete with Dolby and their encode-decode noise reduction for cassette tape, but National's method didn't require encoding.

 

[xmat]

thanks for help

Thanks for the answers.

All I want is to filter high frequencies during loud volumes,

to eliminate potentially troublesome high-amplitude resonance peaks in my application..

That DNR could actually help as a mindstep procedure; something that could lead towards a solution that has to be done from the beginning.

I hope someone comes up with a solution..maybe me.

Thank u guys,

xmat.

 

[mstechca]

It sounds like at loud volumes, the sound is "distorted" (or mixed with noise of any audio frequency).

I don't think what you are looking for is a strange filter.

Go to Harry's Homebrew website and go to the following:

Select Reference
Select Building Blocks
Select Transistor Biasing.

There are two Transistor Biasing options. Go through both, and you can get a detailed explanation on how an audio amplifier works with transistors.

At the end, he shows you how to improve the quality of sound at loud volumes.

It is well worth the look. After all, it is free material.

 

[audioguru]

Why not just damp the resonant peaks with a very low output impedance amplifier, or damp the resonances like quality speaker manufacturers do? :?:

 

[xmat]

thanx

Ok, I just saw the links you recommended.

Unfortunately there is no distortion in the signal, it is just the "physiology" of the sound source that gives more high frequencies and more peaks when the sound becomes louder and louder..
I want to eliminate those changes, but in a "dynamically changing" pattern;
not once-set-and-go, cause that will have a significant effect on low volume signals (i.e completely eliminate them)..

So what I need is a "variable/adjustable" filter that cuts-off high frequencies (i.e sudden changes) occuring at loud volumes only..Got it?

Thanks in advance,

xmat.

 

[audioguru]

Hi Xmat,
National Semi discontinued their LM1894 dynamic noise reduction IC last year, but some shop somewhere might have one.
Maybe you can invert its control signal and increase its low frequency bandwidth to make it work backwards like you want.
https://www.electro-tech-online.com/custompdfs/2005/06/AN-384.pdf >
https://www.electro-tech-online.com/custompdfs/2005/06/AN-386.pdf > application notes
https://www.electro-tech-online.com/custompdfs/2005/06/AN-390.pdf >
**broken link removed** product folder
**broken link removed** datasheet

 

[mstechca]

Ok, I just saw the links you recommended.

Unfortunately there is no distortion in the signal, it is just the "physiology" of the sound source that gives more high frequencies and more peaks when the sound becomes louder and louder..
I want to eliminate those changes, but in a "dynamically changing" pattern;
not once-set-and-go, cause that will have a significant effect on low volume signals (i.e completely eliminate them)..

So what I need is a "variable/adjustable" filter that cuts-off high frequencies (i.e sudden changes) occuring at loud volumes only..Got it?

Thanks in advance,

xmat.

Use two comparators. One will pass volume under the level you specify, and the other one will do the opposite. For the one that passes high volume, add a filter in series. I haven't done it before, but it is worth a try.

 

[audioguru]

Use two comparators. One will pass volume under the level you specify, and the other one will do the opposite. For the one that passes high volume, add a filter in series. I haven't done it before, but it is worth a try.

A comparator won't pass audio, its output is a square wave. :roll:

You need to rectify and filter the audio to obtain a DC voltage proportionate to the level and use it to control a voltage-controlled lowpass filter like the way that the LM1894 IC works. Except the control voltage must be inverted so that it filters-out high frequencies at high level and passes them at low level. :lol:

 

[mstechca]

A comparator won't pass audio, its output is a square wave. :roll:

You sure an analog one wont?

To me, I use logic gates to deal with square waves.

 

[audioguru]

A comparator won't pass audio, its output is a square wave. :roll:

You sure an analog one wont?

If I take a common LM393 analog comparator and feed it maybe only 100mV or more of audio, I am certain that its gain of 200,000 will cause its output to be a square wave. Try it before suggesting it. :roll:

An analog comparator is a switch. Its output switches either high or low to indicate that one input has a higher or lesser voltage than the other. :lol: