new shiny hammer- or Universal comp

[mediatechnology]

Power switch?

[JR.]

Who would ever want to turn something that pretty off?

Actually this layout is more a pro forma mental exercise to look for a logical arrangement and how few controls I can get away with. I could do it all with a LCD display and a few up/dn buttons but this light show would be easier to use.

As much as I like "my" ideas I don't feel this is commercial enough to pursue so this is just some mental masturbation. I may figure something that makes sense... The $5 comp with USB dongle seems interesting.

JR

[mediatechnology]

Well it looks pretty cool to me John. I probably wouldn't want to turn it off either.

[JR.]

I have dumped the output volume control and added side chain and pk/ave/rms switches...

This is fun and in the digital domain it's no big deal to accumulate 3 different versions of signal after A/D similar to meter, peak, ave, and rms. The peak is obvious FW rectify, AVE is a little harder with content averaged between zero crossings before applying att/rel. The RMS would square the peak signal before applying att/rel time constants, then take square root before crunching result, so it's not you father's RMS...

If I did this in DSP I could inject parametric eq into side chain, perhaps multi band dynamics... tooooo many options, but easy to blue sky as long as I don't have to finish.

JR

BTW, I could make the unit go to sleep if no signal present for x hours and wake up when it detects signal at input.

[emrr]

John, this is very intriguing. Please keep it up.

[JR.]

Despite, the apparent full house of controls, I am missing at least one more. The gate-downward expander needs a max attenuation limit to prevent unnatural sounding noise floor. Perhaps a max gain change limit could be used on each section, but gate seems to need it most. As I recall, I used 3 pos switches for gate slope and attenuation, and slopes were very modest.. 1.1:1 and such.

IMO if you can hear these working they aren't right, and noise floors are never dead silent. The main effort is to get the comp to comp just the signal but not the noise. I guess the intent of the soft knee approach is to get some compression happening before the actual threshold is reached. So another way to skin the same cat.

JR

[mediatechnology]

JR - Have you taken a look WaveFront?

http://www.wavefrontsemi.com/index.php?products

I know most of your DSP is being done in PIC but I just stumbled upon these guys.

[JR.]

Funny, their example of some assembly language code for a comp/expander says Alesis on top.

I'm looking at DSP right now for measuring drums, I plan to use FFT to parse out the sundry non harmonic resonance modes.

For the relatively simple number crunching to make a digital controlled analog comp I don't need the fancy DSP hardware.

I did notice a LOG instruction, that I don't recall from my early fiddling with the Microchip DSP. I assume at worst I'd just roll my own log conversion.

JR

[mediatechnology]

I think they have close ties to Alesis. I did look at the log-antilog and it's a piecewise table lookup accurate to 12 bits I think.

The off-center 45 rpm record code in the DSP example looks very compact.

[JR.]

I haven't though about it (yet), it seems in converting between PCM and decibels crudely could be a pretty straightforward 6.02dB per Significant bit, plus a calc on the remainder for fraction of 6 dB.. Of course the resolution you can get depends where you are in the range relative to your zero dB.

JR