A Couple Of Things That Puzzle Me About Compressors

[Mark]

1: Why does the Threshold knob have positive, as well as negative, values? Mine goes up to +20dB.

2: If I set the Threshold to, say, -10dB adding +10dBs of Make- up Gain should bring the overall level (-peaks) back up to 0, right?

I'm pretty sure that's how Make-up Gain actually works. I just need confirmation.

[drumsound]

1: Why does the Threshold knob have positive, as well as negative, values? Mine goes up to +20dB..

Because signal can be above or below 0dB VU

2: If I set the Threshold to, say, -10dB adding +10dBs of Make- up Gain should bring the overall level (-peaks) back up to 0, right?

I'm pretty sure that's how Make-up Gain actually works. I just need confirmation.

Make up gain it to get the signal back to generally where the peaks were after your have reduced the gain via compression.

[Professor]

First off, there are 3 key controls on the compressor. You already have a fairly good handle on the threshold and the make up gain controls, but you seem to be forgeting about the ratio.
If you set the threshold to -10dB and the makeup to +10dB, that really doesn't mean anything if your ratio is set at 1:1. Indeed, a peak that would have hit 0dB would now hit +10dB. But if your ratio is set at say 2:1, then a peak that would have hit 0 would now hit -5dB before the makeup gain, and +5dB after. If your ratio is set at 10:1, then a peak that would have hit 0dB would hit -9dB before the makeup and +1dB after.
Remember that your compressor is really just an automatic volume control.
It allows you to turn up a signal that is too low in the mix, but it automatically watches to make sure that occasional peaks in the level won't exceed a maximum level. In that way it increases the 'average level' without increasing the 'peak level', or more simply, it makes the soft stuff louder without making the loud stuff too loud.

Now as for the threshold being adjustable to levels above 0dB, well that has to do with special circumstances and the use of the compressor in its more extreme settings where it becomes a limiter.
If you have signals in the analog domain which are bouncing up above 0dBv to maybe +6 or +10, then those signals would be pushing near -6dBFS on a digital scale and you might want to limit them before hitting perhaps a DAW or other digital device. In those cases, you could set the compressor to a threshold of say +6dB with a ratio of 10:1 with no makeup gain. At that setting, the signal would have to peak to +16dB to move the output to +7, and +26 to move it to +8, and that kind of level is highly unlikely without lots of distortion at the source, and so the compressor is effectively a limiter keeping the maximum peak level between +6 and +8dB.
Also, if you were getting really fancy and using the sidechain input to use your compressor as a ducker then you might want the threshold set rather high before one signal takes control of the dynamics of another.

Either way, if you are having trouble getting used to compressors, I usually suggest going with lower threshold and ratio settings as this sort of 'light compression' is often much more transparent and natural while still giving a pretty substantial boost to quiet tracks. For example, setting the threshold to -30dB and the ratio to 1.5:1 (which I think of as 3:2 because whole numbers are easier) means that a peak which would have hit 0dB will now only reach -10dB. From there, a makeup gain of +10dB means that the peak is right back at 0dB, but the rest of the track is pulled up +10dB higher than it had been, so the soft stuff is louder while the loudest stuff is still where it was. Those sorts of lower ratios, 1.2:1 (6:5) or 1.25:1 (5:4) combined with lower threshold points can be pretty powerful in a mix while staying pretty subtle too.

-Jeremy

[DJ RELAX]

sorry to hijack but does anyone wanna go more in-depth about sidechains? what do they do, and what do you feed the sidechain with? a mult? a tone? thanks

[Professor]

Amazing, it feels like I just spent a healthy portion of Wednesday's class explaining this as part of the compressor talk for the "outboard equipment" chapter.
But at least it's pretty fresh.

Think of a side chain as a way of using a different signal to control or 'activate' the compressor than the signal which is actually being compressed.
Normally we think of sending a signal into a compressor and when that signal crosses the threshold level, it activates the compressor which applies some sort of gain control to that signal before it is expelled out the other end. Most of the time, that's all we do is simply control the level of one particular signal all by itself, or maybe a stereo pair that are controlled together.
But if you have the ability to activate a compressor with an external source, you open up some interesting possibilities.

Let's say that I'm compressing a vocal track and I want to compress it more (turn it down lower) when an annoying sound like a sibilant "S" or "T" sound comes along. Well I could split the signal going into the compressor so that one part goes into the compressor, through, and out the other end and on into my final mix. But I can take the other half of the split though an equalizer and into the sidechain to control the compressor. If I find the particular frequency band where the S & T sounds are and crank up that band pretty loud in the EQ, then every time those sounds pop up they will hit the compressor controls pretty hard and activate the compressor more often, so those sounds are turned down while the rest of the vocal sails through with just little or no compression.
We call that device a "de-esser" because it helps to remove "S" sounds from the recordings. But when you see what's going on, you realize it can control any pesky sound in any track.
If you have a bass that tends to "woof" a bit on certain low notes, you could set a side-chain EQ that boosts those frequencies so they are hit harder by the compressor as the bass goes through. The result will be a smoother and more controlled bass track. Or maybe it's the opposite and your bass is a little weak in the low mids. Then you could dip the low mids a bit in the side-chain and the comrpessor will let more LF signal through before the comp becomes active.
It's weird that the EQ applied can give a reverse effect on the signal coming through, but once you get used to it, it's not so bad. And of course, if your compressor is set to boost the hell out of a signal when it is activated, then it will work in a similar ratio rather than reversed. Say for example if you boost the low mids between maybe 100-225 on a snare drum sidechain, and have the compressor set to boost the hell out of the signal. Then the comp will take off every time the snare is hit (or at least hit hard) but will stay quiet when the hit hat is trying to steal the show.

OK, that's one possibile use.

Another cool one is to have the level of one instrument controlled by a completely different instrument or sound source.
Let's say I have a singer/songwriter playing acoustic guitar and singing at a live gig, and rather than chasing levels, I'd just like the compressors to help me mix the gig automatically. So I split the voice and send one half in as a side-chain control on the compressor for the guitar, maybe with just some light settings so it only shaves off a few dB, like -18 threshold and 1.2:1 ratio so we're talking a -6dB reduction for a peak that would have hit 0dB. When s/he is just playing guitar then the instrument can go through at a nice, full volume. But when the voice comes in and crosses that threshold, the guitar gracefully steps only slightly out of the way, and then with maybe a 1-second release time, it gradually comes back as the singing stops.
This one is called a "ducker" and isn't used much in recording, though it's used all over the place in broadcast, tele-conferencing systems, classrooms, and court recording systems. In broadcast, it allows the host to 'talk over' the guests or the music without a shouting match. In tele-conferencing, gates might be a little more common, but a ducker can make sure that the CEO can talk over everyone else. In a classroom, a teacher can show a film and talk right over the top of it without having to mess with the volume controls. In a court-recording system, the plaintiff & defendent may start arguing, but when the judges speaks, the other mics are squashed down so the judge comes through on the recording.
Of course, it could work in reverse too, so that perhaps whenever the bass is playing the piano gets a little louder, but when the bass drops down the piano follows. Or exactly the opposite.

Those are probably the two biggest applications for the side-chain, and you can see they are really handy in live sound while they may be kind of unecessary in recording, especially in an automated or DAW-based mix environment.

-Jeremy

[Mark]

That's concise and useful. Thanks Jez

[Dave-H]

Thanks Prof. ! Good stuff in your post!!

[Fletcher]

Prof... it's nice to see someone be able to explain the theoretical crap in such an easy to understand manner!!

I will mention that not all compressors have "fixed knee" ratio selections... some [like the Crane Song LTD. STC-8] have a variable knee which allows you to tailor the amount of compression to things like 2.785:1 [or "I dunno":1, which is the one I use most]... and then there are compressors like the Drawmer 1968 where it's "the more you bend the GR meter to the left the steeper the knee" kind of a thing [but I reckon those are more the exception than the rule].

[mjau]

or "I dunno":1, which is the one I use most

Yep, the "that's too much, and that's too little, so let's split the difference" : 1 ratio"

[Kelly]

The sidechain ducking does get used for dance/electronic music though.

The bass and kick drum in dance music will often conflict with one another. What they do is send the bass through a compressor and the kick in the side chain. What happens is that when the kick sounds, the bass will drop out, making room for the low end. Since in electronic music, the bass and kick can be very similar in sound, you often dont even notice the bass dropping.

Not sure, but I believe this is used on entire mixes too. Split the kick to two busses (or mult), and the rest of the mix another bus. Send the non-kick-mix bus to the compressor input and one of the kick busses to the side chain. This will cause the most of the mix *but* the kick to drop in and out when ever the kick sounds. This will catch the ear quicker than just doing it on the bass though, but it's a neat way to make room for the kick when all else fails.

[soundguy]

1: Why does the Threshold knob have positive, as well as negative, values? Mine goes up to +20dB.

2: If I set the Threshold to, say, -10dB adding +10dBs of Make- up Gain should bring the overall level (-peaks) back up to 0, right?

I'm pretty sure that's how Make-up Gain actually works. I just need confirmation.

Mark-

here's a real easy way to think about threshold controls if it hasnt clicked with you yet. Imagine that every limiter out there had a threshold that was fixed at 0 dB. When you input a signal higher than 0dB, then the unit would start compressing. Depending upon how the ratio is set combined with the level you feed the limiter will determine how much compression is happening. If you are using your compressor in series with a unit that has a variable gain control for its output, then its no biggie. If you are plugging your limiter into a mic pre and want more compression, you just turn the gain up on the mic pre or if you have it on your mix buss you'd just raise your master fader.

there are other circumstances where you are feeding the limiter from a source with a fixed level output. In this instance you have no way to control how much signal is going to the limiter, so a smart guy put the threshold control on there so you can adjust the input gain to the circuit from within the compressor itself. Examples of this could be using a mic pre that does not have a gain control (there are a few like this out there) so no matter what the pre is always sending out 46dB or whatever it is calibrated to. Another would be feeding a compressor from a buss to run parallel with its source, not all consoles are designed with active bussing with their own gain controls, several old consoles were designed with passive bussing where the level of the channel determines the level that feeds the buss. So if you are mixing and have your source set to -12dB in the mix, that is the level that is going to come off the buss and into the compressor. You have the negative threshold control to lower it down so the compressor will compress that -12dB signal (you cant move that fader otherwise your mix will change, right?) and you have the positive threshold the limiter can accomodate the +14dB signal coming off your fixed gain pre if you only want to compress that signal 2dB or so.

On many compressors the ratio control is just an approximation which is why others dont have ratios on them, depending upon the element that is used for compression in the sidechain, the ratio changes as it workes harder. This is true for nearly all optical limiters I have ever heard be it an la2 with its unique optical element or others that use vactrols. Because of the changing ratio, it makes a lot more sense sometimes when they do have ratio controls to have them on a variable pot instead of a switch. There's two schools of thought on that one for sure. Certain mu tubes have different ratio characteristics as well which is why you rarely see ratio controls on tube limiters, same idea, different manifestation. On the other side of the spectrum, fets and vca's can have very predictable and repeatable ratios which is why you typically see ratio controls on nearly all of those types (unless they are fixed limiters) and why they are almost always switches and not pots.

dave

[workshed]

A quick note on sidechains that I failed to realize until it was too late: if you ever wire your sidechain into your patchbay, make sure you normal the sidechain input to the sidechain output (or however that goes), or your compressors won't do their thing when you're not using the sidechain. Now I have to pull my patchbay out and flip the cards on a couple slots when I can find the time. Whee!

[KennyLusk]

Professor AND Dave giving a class on compressors! I printed this one babe! Thanks guys.

[majortom]

Thanks you guys for the great info,

I do have what may be a dumb question though...when you a saying to "split" the signal when using the sidechain could you be more concise on what/how you are actually doing this. Thanks again. Todd.

[scott anthony]

...when you a saying to "split" the signal when using the sidechain could you be more concise on what/how you are actually doing this. Thanks again. Todd.

Splits are easily done with mults... typically 4 or more jacks on a patchbay wired together so when you plug a patch in, you get 3 out. Feed one back to the channel and one to the sidechain.

No mults? Use a buss or send on your board.