Using an A to D Converter in a Compressor Sidechain

I'd like to explore the idea of using an A to D converter in a compressor sidechain. It seems to me if the converter was infact part of an embedded microprocessor several interesting variations could be programmed that could be selected dependent upon your needs at the time. Please share your thoughts ....

RNC..

It was the RNC that caused me to think about this. However it seems to me that the RNC is a 'simplistic' or at least first generation device.

I don't think it's simple.

After all, an A/D converter dosen't do the trick alone. You'll need a DSP as well - and you'd need to come up with the associated DSP programming. Not an every-day task, if you ask me..

Jakob E.

I've done embedded programming before. There are some pretty fast FFT algorithms out there in the public domain. What are the DSP elements that need to be considered?

In my experience mixing digital with analog on the same PCB is mostly about proper layout.

The calculation for a simple attack/release/threshold/ratio digital sidechain are pretty basic. A few multiplications, even fewer additions and one or the other logarithm. Nothing fancy like FFT. I would be surprised if you had latency issues with a more or less suitable DSP.

AD/DA have some latency, so a limiter is hard to get. The A*M*E*K 9098 comp has digital sidechain as well, IIRC.

Samuel

When I mentioned FFT as I was thinking of processing to be dependent upon the spectrum of dominant passbands passing thru the micro. For example when compressing a drum kit the high-hat cymbals have a unique signature that could be handled differently from that of the kick drum. Keep in mind that I am talking out loud and speculating here. A little experimentation may prove my premise to be a worthless idea.

Hmm...

Would an MSP430 from Texas Instruments do the job? Some models have AD/DA capability built right in.
They have development kits that come with C compilers for $99.

Tamas

If you need low latency FFT would not be such a good idea. Ideally you'd want an algorythm that processes individual samples not huge buffers like FFT cause that would cause a latency of at least the number of samples in the buffer. You'd need buffers of at least 512 or 1024 samples to have some control over the low freq's.
Try some biquad (or other) IIR filters. They're faster and have less latency. And there is lots of sample C code for stuff like that floating around on the net. (http://www.musicdsp.org)

carl wrote:

When I mentioned FFT as I was thinking of processing to be dependent upon the spectrum of dominant passbands passing thru the micro. For example when compressing a drum kit the high-hat cymbals have a unique signature that could be handled differently from that of the kick drum.

Click to expand...

couldnt you do that with gates and eq (analog)?. i do what your talking about everytime i compress. thats what the sidechain is for.

couldnt you do that with gates and eq (analog)?. i do what your talking about everytime i compress. thats what the sidechain is for.

Click to expand...

If you do it digitally you can switch algorythms to change the character of your compressor without changing the circuits.

I'm not really sure about this but i think most ad converters nowadays do oversampling so they can have simpler analog antialiassing filters before the ad stage. The problem here is that they will do digital downsampling and this will introduce some latency too. So it might be a good idea to check if there are any special low latency AD converters you can use. You probably won't need high quality audio converters in a sidechain anyway.

I've got a VCA developers board from That Corp, as well as developers gear for Motorola and Rabbit embedded controllers. Perhaps I should set aside some time and do some exploring. I'll will at least take a look at those filter routines on the MusicDSP board that was suggested and go from there. Thank you for your suggestions and help.

Isn't there an Alesis chip that can do this? AL3101 or something like that... you could use it as the sidechain to drive the THAT.

regards, Jack

I didn't know that the AL3101 existed. Oh well ...

More to research and learn.

Hi:

Now Alesis chips are made under another brand name (Wavefront)http://www.wavefrontsemi.com/.
They have A/d,D/a and Dsp products for your application. :grin:

Available from ProfusionPLC in England - where we also get our "That" stuff..