Parallel Compression: Pitfalls & Possibilities

In posts one & two, we talked about the history of parallel compression, a few different ways to achieve it and how I’ve been using it. This post will close up the series with a few cautions and options for using it. It was pointed out in the comments of the first post that this is also called “upward compression” because it has the effect of bringing the softer sounds up, while keeping louder sounds under control. I call it “Z-Axis” control because just like in 3-D modeling, raising or lowering the parallel comp channel can bring a sound closer or put it back in the mix where it was. It’s a great tool to keep your worship leader’s vocal is right out in front and center and not buried somewhere behind the guitar.

Potential Problem

As Dave mentioned in his comments on the first post, one of the biggest issues faced when using parallel compression most digital boards is the latency issue. When you start bussing the same signal around a digital board through different paths, the variations of that signal will arrive at the L&R bus at a slightly different time. It might only be a few samples, but that can be enough to start causing phase issues. While researching this article, I found a YouTube video that demonstrated the technique. About a second into hearing the effect, I said to myself, “Hey, he’s got phase issues with those drums, they sound terrible.” Near the end of the video was a disclaimer slide saying he didn’t time align his tracks to account for the phase issue. Depending on your source, the effect of said time difference can range from almost inaudible to atrocious.

Avid is really the only digital console manufacturer to address this issue with delay compensation. Basically, delay compensation automatically makes sure all the output busses arrive at the main L&R bus at the same time regardless of the plug-in chain. With other digital consoles you have to line it up manually or make sure the processing path is the same (even if that means inserting comps that do nothing into your “normal” channels). Let’s take a look at a real-world example.

I mix on the DIGiCo SD8, which does not currently offer delay compensation. When I set up my parallel comps for drums, you’ll notice I have them sent to two groups, “normal” and “spanked.” The reason for that is latency. If I simply sent the drum mics to the L&R group and to the spank group (which would then go to the L&R group) that extra hop the spank group has to take causes the compressed group to arrive just a little later than the uncompressed ones. It may only be a matter of samples, but it’s very audible (and none too pleasant). When I tried it that way, snare immediately thinned out, and the cymbals became very harsh. Once I routed the drum channels to the normal group, and then to the L&R group, everything came right back. I’ve not had time to measure the timing to see if I need to add a dummy comp on the normal group, but by ear, it sounds good. At some point I’ll send a quick impulse down both groups (with one panned hard right and the other hard left), record the output and see where we are.

Whatever digital console you’re mixing on, chances are you’ll be needing to something similar to keep things in time. My hope is that one day all console manufacturers will be offering automatic delay compensation (it’s a simple matter of programming, after all), and that will open up all kinds of cool possibilities for processing our signals. Until then, plan on spending some time with it to make sure it sounds good.

The M7 Challenge

Since Yamaha M7 seems to be one of the more popular church audio desks, I decided to poke around and see if I could set up parallel compression on that one. It’s tricky because the M7 doesn’t offer groups. That means using mixes—and there are only 16, so they’ll go quick. If you want to do “group” parallel compression on the M7, say for the drum kit, what’d you’d do would be to route all your drum mics to two mixes (or 4 if you want to do stereo and have the mixes to spare); mix 1 will be our “normal” mix, mix 2 is our compressed mix. Make sure to un-assign all those drum channels from the main L&R bus; then assign mixes 1 & 2 to the L&R bus. Now you can use the comp that lives in the mix to create your smash, or insert one of the rack effect comps (like a 276). I haven’t actually tried it, but I’m betting if you do use a rack comp, you’d need to also patch one into the normal mix to keep things in time.

There are several disadvantages to this, not the least of which is that you’d have to run the mixes in post-fader mode in order to have any kind of mix control on the faders. That can set up some serious gain structure issues that you’ll have to manage. It also ties up 2-4 (of 16) mixes and 1-2 of your 4 rack effects. In a lot of ways, this falls into the, “Just because you can, doesn’t mean it’s a good idea,” realm. It wouldn’t be a tough thing to use two input channels to create a vocal smash effect, however. Personally, that’s where I’d draw the line with that console.

Most other digital consoles offer groups, which make it a little easier to set up the effect. Just be aware of the latency issue.

Like all advanced techniques, this one will take some experimentation to get right. You also have to be aware that you can do more damage to your mix than good if you over-use or mis-use the technique. Remember that no one comes to church to hear the cool parallel compression on the vocals or drums. Like sugar (or cayenne pepper) a little goes a long way. Don’t go crazy and try to parallel compress everything. On the other hand, when employed properly it can really add some extra sparkle to the mix.