Questions for Boz about modelling plugins

Sent @bozmillar PM asking if it was ok to make this a thread and he said yes…

…I’m a big fan of his +10 DB Compressor, modeled on the ADR Compex. It’s cool on lots of things but I particularly like it on drum and mix bus, vocals and bass guitar. It’s a bit of a conundrum because it somehow manages to be punchy but presents the resulting sound in quite a smooth way; I’ve seen it described as creamy, which is a terribly ambiguous word in the describe-sound-in-metaphors genre… but I think it kinda fits. To put it in more technical language, you can put it on drums and get 4dB+ gain reduction with the associated punch on the transients and release-time-breathing but without the negative side effects you normally get from compressing so hard.

So, that got me thinking. How does a plugin developer actually go about turning a real analogue piece of gear into plugin form, and capture the sound/ character? What’s the actual process?

Here are the questions as I put them to Boz;

"How much technical measuring goes on, how much listening you do etc. Also, what are the principles behind modelling a real piece of gear - for example, do you treat the hardware like a “black box”, measure inputs and outputs, then devise and algorithm that behaves in the same way? Or do you deconstruct how it works into components - detector circuit, op amps, whatever - and model them?

Also, we often talk when modelling the more well-known compressors about idiosyncrasies in their operation; for example, the fast attack of the 1176 not preventing it from being punchy thanks to the ratio overshoot over time, the “all buttons in mode”, the way an La2a has a program dependent release, frequency dependent behavior etc. Are there any such behaviors that the Compex has that you had to model?

Two things about its operation I think are interesting from my use of it; It’s clearly designed to work around a sensible signal level - if you’re the kind of mixer who cranks everything up to 0dbfs+, you might struggle to use this comp effectively. Was that level calibration a deliberate choice? And the GR meter - it seems to respond much more quickly than most other plugin comp VU meters I’ve seen. On some signals I can get 4dB or more GR without really hearing many signs of compression. Is this a consequence of the behavior of the compressor, or was it a choice about how the meter would actually operate?"


Here’s my copy-paste answer:

When I made this one, I had the schematic of the thing along with the hardware so that I could see what it was doing, so I used that as a starting point. I would run all sorts of test signals through it to get it to where it would measure the same, then once I thought that was good, then I’d run music through it to make sure it sounded the same. In the end, I was really interested in making sure it sounded like the hardware on music and not just on measurements.

Also, yes it was a choice to make it operate at lower levels. I’m still not convinced that was the correct choice, but in the end, we just went with choosing to make the plugin act like the hardware in every way possible. I’m sure it throws a lot of people off, especially since a lot of people tend to mix with higher levels these days than they used to when this unit was designed.

For the meter, I would put the plugin on the screen right next to the hardware unit and adjust the ballistics in the software until they matched as closely as possible. I was probably a bit heavy handed about this part, so it doesn’t match the hardware in all cases, but it was pretty close.


I love the +10dB stuff. It’s my (not so) secret sauce for punchiness on drums. Of all the plugins I own, +10dB really seems to have a unique sound and response.


It’s definitely an interesting plugin and it works really well on drums. I made the blasted thing, and I still can’t quite put my finger on why it works so well for adding punch to drums. It’s a little bit frustrating actually.


Haha, that’s wild! I remember when I bought it, I took one look at the interface and went straight to Groove 3 and bought a video explaining it’s operation - It’s one quirky little monster, for sure!

1 Like

It really is. When I first got the hardware, I started messing with it and I thought “Holy crap, how do I even work this thing?” Half the knobs and switches hardly worked because they hadn’t been changed in years.

A very large part of me wants to take the algorithm and package it in a more standard UI.


It’s funny, the psychology of that. From a purist’s point of view, it shouldn’t make a difference whether it looks like a Compex or the Airwindows approach of having NO UI besides sliders which may or may not show numeric values.

But, it’s a cool looking plug which I think adds to its charm - the way it makes you approach setting levels. For me, the first thing I do is knock it up to 3:1, use the input control to get the meter swinging, the output meter to restore unity gain. Then I dial in the release, attack and decide whether to change the ratio, in pretty much that order. I almost never use the threshold knob.

I’ve never quite been able to work out the auto release. I use it often, but with something like an SSL comp it’s pretty clear what it’s doing - short signals over the threshold get a short release, the longer and further over the threshold the longer the release. Does the Compex work on the same principle but much faster, and if so where is the threshold that controls that? Is it controlled by the threshold knob, the input volume, something else…?

Another question, and maybe a cheeky one; now it’s been out for a couple of years, is there anything you wish you had done differently?