VCA compressor - control voltage

[lefthandRED]

Hey there,

im designing a VCA compressor for a university project, I've got most of my stages done am now refining my control voltage signal path.

The questions i have are;

1. The signal has been rectified and then low passed. What is an appropriate frequency to low pass my audio to create an acceptable control voltage signal?
I've seen on a related post that 20Hz was used, this seems too aggressive to me and would ruin the dynamics of the control voltage, or am i way off?

I cant find any information on this anywhere else, it seems to be one of those questions no-one thinks to answer maybe?

2. I have been running some simulations using a low pass sallen and key filters with a bessel design, but i the low frequency cut off is giving me a really long rise time (~50ms), i need this right down as its ruining the point of trying to develop a fast response filter. Any suggestions of topolgies or anything that i could use to get a rise time of say 1ms?

any advice would be appreciated.


Nick
(I know that I should really have an RMS control signal but for the purposes of a learning experience i didnt want to just stick a chip in there and have it do it for me!)

[electrode10101]

Originally Posted by lefthandRED

Hey there,

im designing a VCA compressor for a university project, I've got most of my stages done am now refining my control voltage signal path.

The questions i have are;

1. The signal has been rectified and then low passed. What is an appropriate frequency to low pass my audio to create an acceptable control voltage signal?
I've seen on a related post that 20Hz was used, this seems too aggressive to me and would ruin the dynamics of the control voltage, or am i way off?

I cant find any information on this anywhere else, it seems to be one of those questions no-one thinks to answer maybe?

--When designing dynamic processors, timing circuits are usually designed/described in terms of rise time (attack time) and fall time (release time), found as the product of either resistance and capacitance, or current and capacitance. While timing circuits must be at least single pole, many dynamic processors use multiple pole timing circuits to achive desired results.

You didn't say if your compressor is to use feed forward or feedback detection for the timing circuits. This will determine the type of dectector required.

Is the VCA linear/db voltage change, or log/db voltage change?

What is the scaling of this change?

2. I have been running some simulations using a low pass sallen and key filters with a bessel design, but i the low frequency cut off is giving me a really long rise time (~50ms), i need this right down as its ruining the point of trying to develop a fast response filter. Any suggestions of topolgies or anything that i could use to get a rise time of say 1ms?

any advice would be appreciated.

We can discuss this after you answer the questions above.

BTW have you checked out the app notes at THAT Corporation Audio Technology ?


Nick
(I know that I should really have an RMS control signal but for the purposes of a learning experience i didnt want to just stick a chip in there and have it do it for me!)

[cbdb]

Yes you will need fast attack times if you want to catch fast rising waveforms, anything percussive. Your low pass filter limits the attack speed. For 1ms attack you need approx 1khz. Send a square wave into the filter, adjust the cutoff freq and see what comes out (rise time).

[seanvn]

There is an agc idea here:
www.edn.com/contents/images/080405di.pdf
The nice thing about it is that you can make it from junk box components. You don't even need a FET but it is certainly not hi-fi by any measure.