Let’s look at the relationship of timbre to distortion, because the two are cousins if not siblings. We’ll also compare clipping to compression, differentiate the sound of the two, and get that firmly in your ear.

Four or Five Characteristics

Sounds have four main characteristics: pitch, timbre, loudness, and envelope. And duration, but we don't need that right now.

Pitch we know about. Loudness we know about.

Timbre is a mix of pitch and loudness. All instruments—indeed, all objects, have a timbre when you give them a good klonk or whatever it is that needs to be done to get the thing to make noise. Ping a glass with your fingernail and there is a distinctive "glass" sort of sound. Get two of the same glasses, ping them both and you'll notice that they don't sound exactly the same. This is because each glass has a slightly different overtone series—a slightly different set of harmonics, which are frequencies above the loudest frequency, the fundamental, that gives the note its pitch name.

The overtone series of every instrument is different. We call a sound with a lot of high overtones "bright," with lower overtones "warm." Depending on the math of it all, overtones can also make things sound harsh or smooth, or even in or out of tune. I wrote more in-depth on this stuff here.

Timbre is Overtones

So, you have a note at 329.63 Hz, and you'd like to make it brighter, so you put an EQ on it and turn it up, but you don't have the frequency of the EQ at 329.63 Hz, do you? You have it at like 8kHz or something. A shelf at 15kHz. That EQ is turning up the overtones of the note, right? It gets brighter because you're amplifying the overtones.

For some of you, this is "Duh." For others of you, this is, "Oh really...? Hmmmm..."

What if, instead of amplifying the overtones of an instrument, we added more overtones into the picture? We generated some additional overtones that are consonant and harmonious with the fundamental, and added them into the sound. It would be brighter, right? It would be subtle, not as noticeable as an EQ boost, but it would make an audible difference.

That, campers, is what happens when you drive a signal into tape or a circuit a little too hard and start clipping it. It generates additional overtones—harmonic distortion. This is what happens when you saturate tape, or saturate a transformer, or overload a circuit on a preamp. Heck, just passing a signal through a compressor with no compression happening causes some additional harmonic distortion to happen, which changes the timbre of the signal. This is what people are talking about when they say, "I'm adding this not for the compression, but for the color."

Instruments sound the way they sound in part because of their timbre. Equipment sounds the way it sounds in part because of its harmonic distortion. These are the same thing, really.

Timbre and Harmonic Distortion Fall in Love at an All-Inclusive Resort

So, some instruments naturally sound better with some pieces of gear because the timbre and the harmonic distortion are complementary. And things can also sound bad because of the relationship of these two things. I found out pretty early in my career when I was recording guitars through distorted amps, that sometimes, if I doubled a part with two different amps, it might actually sound a little thinner when mixed together, or buzzy and harsh, and in some cases, out of tune. It was overtones and the timbres not lining up.

It's dumb luck that the harmonic distortion caused by tape compression/saturation generally enhances the tonality of most instruments. Same thing for circuits using tubes. The same thing for transformers. Rather than EQing a vocal to get it to sit better, we can smush it a little bit into tape and it gains a bit of presence and "bite." We can get a bass or a kick to have more authority on small speakers by pushing it a little bit harder through some transformers, which tend to generate harmonic distortion that is lower in frequency than most of the stuff generated by compression/saturation/distortion.

Slamming cymbals through things often sounds like ass—really nippy and harsh. Too much harmonic activity. Higher voices and higher-pitched keyboard parts can get really nasty with too much extra harmonics up there. Danger Will Robinson!

Don't forget, ALL analog gear and all ANALOG MODELED digital audio adds some harmonic distortion, and things change timbre due to this as levels go up and down. At Korneff, we spend MONTHS on the modeling to get all the distortion and harmonics behaving in an authentic, analog way. It's easy to make a plug-in that does something. Relatively speaking. It isn't as easy as, oh, making toast. But it's much more difficult to make a plug-in that really captures the analog inspiration.

Harmonic distortion ain't the only thing that happens when you club a baby seal of an audio signal with a 600 pound tape deck. Or a feather-light Echoleffe Tape Delay. You also change the waveform.

Clipping Made Easy

Any sound has an envelope. This is how the sound varies in loudness or power on a micro level. My easy way to think of it: there's a distance between the loudest bit of the sound, usually the attack, but not always, and the quieter bits of the sound—the way the note dies off, the resonances of the body of the instrument (or of a speaker cabinet or a room). The little rattles and noises and squeaks things make.

Compression changes the distance between the loud bit and the quiet bit.

Tape compression and saturation squash the signal (compress it) with an immediate, instantaneous attack that definitely clips the transient a bit. This is true of ANY signal that you slam into clipping: you lose some attack. But saturation has a very very fast release. Like, slightly less than instantaneous. Actually, for our purposes, it's instantaneous.

So, when you squash something into tape, not only do you add harmonics, you lop off the loud bits and smush them down, and because you're increasing the level to do the smushing, you're also bringing up the quiet stuff.

Back to our snare. If we smash it into tape, it gets a little bit THICKER, because we're adding harmonics, and a little bit LONGER ACROSS TIME, because we're changing the relationship between the loud and the quiet. You understand that if we bring up the quiet stuff, the sound will appear to last longer, right?

And you realize that lengthening a snare will change the groove, right? It will sound more "behind the beat" if you squash it into tape.

Now, crushing guitars into tape adds a very nice set of overtones that give them a little more brightness, but the transients are getting slightly clipped, and they get a little bit less distinct and less punchy. You lose the "click" of things. Same thing with pianos or any sound with a fast attack. Slam it into tape or a tube or a preamp, clipping it, and you'll lose that transient a bit. Same with vocals. When I was recording rap stuff, I would cut the vocals a little bit lower so as to not lose articulation and wind up with it sounding mumbly. I would cut punk vocals clipping into tape to deliberately get them a little less articulate and at the same time bring up the spittiness and the mouth noises (that's quiet stuff) so the whole thing sounded more "in yer face."

If you think about it, if someone was in your face screaming at you, you'd hear all the mouth noises. You might even taste the mouth noises.

Homework

So, set up a mix. Route everything to a stereo subgroup. Label this CLEAN. Add two pre-fader sends from this and send each to a different stereo subgroup.

Put the Echoleffe on one subgroup and set it to Tape Emulation mode. This is the group that's going to clip everything using tape saturation. Label this one CLIP. Everything going through this will lose transients but gain harmonics and gain length (the quiet stuff will get louder).

Put the Pawn Shop Comp on the other subgroup. You can use the default setting. Set the ratio to like 6:1 and drop the threshold until the meter is bouncing musically. We want to compress, not limit. This compressor will bring out the transients and push down the quiet stuff a bit. Label this COMPRESS.

Route all three subgroups—the clean, the clip and the compress, to your stereo master.

How does it sound with all three up? How does it sound if you pull down CLEAN all the way? Did you set the sends to pre-fader? If you didn’t, you're about to find out why they need to be set that way.

Throw a LUFS meter on it and mess around with things. Can you get something like a -8 LUFS-S reading without it sounding like utter ass? And without driving things over like -1dB true peak?

Play some more. Maybe route some sounds just to the clip, and others just to the COMPRESS. What works best where?

By the way, if you don't know, what you're doing is parallel compression and parallel saturation. I know most of you know this, but there are a lot of beginners reading this, too.

You will learn tons if you do this.

We usually think of harmonics as being pleasant things to hear. They give an instrument its timbre, they provide brightness and clarity.

Don’t know what harmonics are? Go here and read.

Usually the harmonics that our ears like to hear are mathematically related to the fundamental based on whole numbers. Whole numbers: ones and twos and threes. Octaves are a multiple of 2, 4, 8, etc., things like that. Harmonics can be even numbers, but also odd numbers, and harmonics based on 3 or 5 or 7, while they might sound a little wooly, they don’t sound plain old bad. Also, keep in mind that sometimes the math on these things isn’t perfect. It might not be a perfect multiple of 3 but something close, like 2.98, but generally this is good enough.

Inharmonicity

However, there can also be harmonics generated that don’t have any whole number relationship to the fundamental, and these harmonics are usually unpleasant to hear. This is called Inharmonicity — when the harmonics don’t make whole number sense mathematically.

Strike Tones

On many instruments, inharmonicity happens in the strike or the initial attack of the note. Bowed and reed instruments—violins and flutes, as an example, don’t have inharmonicity because they don’t have a fast transient attack. Brass instruments typically have slower attacks as well.

Fast transient attacks, on the other hand, generate a lot of “inharmonic” stuff—lots of non-whole number overtones. On a piano, the initial strike of the hammer generates a lot of inharmonicity, and that strike is basically pitch-less for a split second. It’s only once the string resonates for a moment that we get a sense of the note. The same thing is true of guitars, bells, and especially drums. That initial strike is basically out of tune, and it is the resonance after the strike that conveys a solid sense of pitch.

The faster the attack, the more inharmonicity is generated in that moment. And, by the way, the transient is typically the brightest moment of a note, because it is so rich with harmonics both good and bad.

Actually, the strike of a note is usually very out of tune! Plug a bass into a tuner and watch how the tuner behaves when you slap a note versus using a softer attack with your finger.

Bells are a great example of the inharmonicity of a strike tone. Listen to Hells Bells by AC/DC and the opening bells are out of tune until they resonate. This has to do with their strike tone. I found a great video that explains this, and while most of you won’t ever record church bells, this is fascinating stuff and it will help get the concept of inharmonicity firmly in your mind.

SO.... instruments have inharmonicity in the attack, the strike. But what about gear? Compressors? Amps? Plug-ins?

Intermodulation Distortion

The way equipment and devices, whether analog or digital, create inharmonicity is through Intermodulation Distortion.

Intermodulation distortion is overtones that are way out mathematically from the fundamental. They typically occur when multiple fundamentals mix together in ways that generate, well... non-whole number math. Harmonics are generated that don’t have whole number relationships to the fundamental. Some of these new harmonics might be undertones that happen below the fundamental, and others above. In some cases the products of intermodulation distortion sound good, but the more complex the sounds get, things get really hairy quickly.

Remember that an instrument, unless it’s like a flute or something with a very simple timbre, already has a lot of overtones to it. A human voice has an incredibly complex series of overtones, so complex that virtually every person has a unique set, which is why we can recognize someone’s voice even if they just clear their throat. So there’s this ton of harmonic activity, then there’s harmonic distortion added to it, and all of those fundamentals AND harmonics have additional harmonics added to them, and then intermodulation distortion kicks in, and ALL those fundamentals AND harmonics AND additional harmonics start negatively reacting with each other adding in yet more harmonics that have bad math going on.

This is the distortion you hear when you crank up guitar amps, or slam things through the mix bus and drive it into clipping.

Here’s a nice, non-technical video on it that makes a lot of sense. You’ll hear why intermodulation distortion can be a huge issue.

Quick Takeaways

Some things to take away from all this.

1. Strike tones are out of tune and bright.
2. Intermodulation Distortion gets worse and more noticeable as the sounds interacting with each other become more complex. It’s hard to get a flute to exhibit any intermodulation distortion. It’s easy to get a full mix to sound awful with even a little intermodulation distortion.

 

 

Oh my, we're jumping back to DISTORTION, for a bit, and looking at what happens when you push a signal up, run it out of headroom, and generate harmonic distortion.

Isn't it cool that, if you've been following this series of posts, you can now understand everything I just wrote? It's also cool if you already knew all this stuff. Everything is cool. Even distortion is cool... if it sounds good.

You may have read, or heard, engineers say things like: "Compression is distortion, distortion is compression, saturation is distortion, saturation is compression yada yada yada" and now all of these terms are mixed in your head and it's confusing. So, let's straighten this out and give you some mental tools so you can get this crap under control.

DISTORTION and COMPRESSION

As you know (and if you don't, go here), as we crank up the signal through a piece of gear and run it out of headroom, the gear loses its ability to reproduce the signal and the wave clips. That is, the peaks of it - the waves that are very high in power - are rounded off a bit. And if you're knocking off the high peaks of a signal, you are compressing the dynamic range of the signal. So, a side product of pushing a signal into the distortion point is some compression.

You've probably heard this whenever someone overdrives up a guitar amp. You'll notice that there's not a lot of volume difference between the softly played parts and the loudly played parts. Contrast that to a guitar amp that isn't overdriven: the quiet parts can be very quiet, and the loud parts really loud. Try this with a Fender Twin - you'll hear the loudest, utterly painful clear guitar parts, and you'll have to squint for the quiet stuff.

Compression occurs early on, as you use up headroom, and it doesn't necessarily generate that much harmonic distortion. It will produce some, but it might be inaudible at first.

DISTORTION is NOT a COMPRESSOR

So, there is a compression of dynamic range when you have distortion, but it isn't the same type of compression that you typically get from a dedicated compressor.

Typically, a compressor has a bit of lag from when it senses a signal over threshold to when the gain reduction circuit kicks in. That lag is called "Attack Time", and sometimes it's fixed, sometimes it's adjustable, sometimes it's short, sometimes it's long, but in any event, that "lag" is pretty much the reason why a compressor sounds punchy: it lets the transient get through... the transient "punches" through - is a good way to remember this.

But when you slam a signal into a tube, or a FET, or into analog tape, and cause clipping, there is no lag. The transient doesn't get through, it is immediately squashed at the speed of not enough electrons. There's also a very fast release when you're getting this sort of effect.

So, this type of compression is very different from that caused by a compressor. It can be very useful, actually, and you're very used to hearing it, especially on records from the '50s, '60s and '70s.

SATURATION

Saturation is a term that describes a physical phenomena: if you record very hot to tape, the magnetic particles can't move any further, and that is called "tape saturation". Think back to 8th grade science class and making "saturated solutions" with that asshole Mr Frank, who always favored the lacrosse players over nerdy fucking musicians like me. Uh... I digress.

Saturation is also what happens to transformers, when a lot of signal is pushed through them and they become "saturated”. Here’s a topic for another post, I guess.

If compression is what happens as we start pushing a signal into clipping, saturation is what happens if we keep going: the signal gets squashed a bit more, and Harmonic Distortion starts to increase.

Increasing harmonic distortion adds upper harmonics, so, a signal moving into saturation tends to get brighter, and the more you push in, the brighter it gets. And this is the big use of saturation and "saturators" these days, to make things a bit more present by adding brightness and... COMPRESSION, right? Because using up headroom and generating harmonic distortion adds compression. But not "compressor compression", right? It adds compression that's not punchy.

DISTORTION

If you keep increasing the level, you'll keep increasing harmonic distortion, and eventually your ear will recognize things as sounding distorted. There isn't some spot where audio engineers agree: "Oh, that's gone from saturation to distortion." A classical engineer will hear ANY compression and saturation and call it distortion, whereas someone using saturator plug-ins might be drawing lines here or there. Someone like me, an old-school analog engineer, will probably just record stuff and get it to where they think it should be and not give a squirrel's ass about what it's called.

In other words, the words are arbitrary. What's happening is this: as you turn things up, you reduce the dynamic range and add upper harmonics. That's what it all is.

WHEN DO YOU USE THIS STUFF?

All the time, I guess. I usually pushed drums into analog tape, while recording, to tame the attacks a little bit and "lengthen" the hits (more on that later). I would, typically, cut the kick kinda on the lower side, because I wanted as much of the punch of that thing as possible, but snare I would usually smush in quite a bit, and cymbals too. Hi hats... if I wanted them crisp - meaning lots of nice ticky ticky transients - then I would cut them on the low side. If I wanted to make them more sloppy (squash the transient a bit) then I would:

a) cut them higher

b) cut them lower

If you answered a), you understand tape compression.

A basic way to think of using saturation/tape compression (or whatever this sort of thing might be called) is: Do I need this instrument to sound brighter? Do I need more punch out of it? Is it too punchy?

Realize that making it brighter, by generating more distortion, will typically nip off transients a bit. You're going to notice the loss of transients on faster things, not so much on slower things like vocals or guitars. As I wrote in last week's blog post, I used to always smush guitars into tape, and that was usually done to get rid of some of the transient activity, so things weren't so pingy and whistle-like (the Insufferable Midrange Filter on the AIP hadn't yet been invented).

And that is it for this week. I had hoped to make you all a video, but my tinnitus is bad this week so it wasn’t meant to be.

Yes, I have tinnitus. I got it years ago, from a week of sessions that was a little too long and a little too loud.

Tinnitus, if you’re in audio, is a bit like getting in a car accident while driving. You might be very careful, and take all precautions, and you can still get hit. If you’re on the road, you can get hit. Honestly, with tinnitus, you can be miles from the road up in the mountains and suddenly a car can drop out of the sky on your fucking head.

Someday I’ll write a bunch of things on tinnitus, but for now I’ll say this:

1) Wear hearing protection around drum sets, horn sections, PA systems and guitar stacks. And on subway trains.

2) Don’t go to ANY live gigs without hearing protection. It could be a concert of ants picking their noses. If it’s being mic’d, it’s too loud.

3) Get an SPL meter app for your phone and measure your environment. Note whenever you’re in a place that gets consistently above 80dB-SPL. Try to avoid those places, and if you’re stuck in one of them, leave as soon as you can - like within an hour. If it is louder, leave sooner. If it is above 100 dB-SPL, question why you are there in the first place.

4) Avoid earbuds like the plague. Never wear them on a train or in a car. This is like playing Russian roulette with a lawn mower.

If you have tinnitus... I feel ya. Most likely it isn’t your fault, and beating yourself up won’t help. Feel free to write me - Luke @ Korneff Audio dot com. Remove the spaces and make the dot a dot. You’re not alone and there are some things you can do so life doesn’t suck.

Here we are - the end of the line for this series of posts on levels, noise, distortion, etc.

Gain staging... from all the talk in online forums and people saying, “Well, you really need to watch your gain staging,” you’d think there's some sort of mystical science magic to it, but it’s really simple.

Gain staging is making sure that each piece of equipment in your signal chain has the best possible signal-to-noise ratio and enough headroom to prevent unintentional distortion.

We have to cover two concepts really quickly, then I’ll tell you how to gain stage things, and we’ll finish off with some tips (rules, suggestions) that make this even easier.

UNITY GAIN

What this means is that the level flowing into the piece of gear is the same as the level flowing out of the piece of gear. Think of a wire. If you feed a signal into a piece of wire, and the wire isn’t really tiny or tremendously long, the amount of power feeding in is the same as the amount of power feeding out.

If we stick a bunch of amplifier circuits and EQ circuits and processor circuits between the input and the output, unity gain is still what we want to have happening.

Now, there’s usually something to control Input Level, we sometimes call this a TRIM, and there’s usually something to control Output Level, and this can be called Output Trim, or Output, or it can be a fader, or, in the case of a compressor, it might be called Make-Up gain, or it can have a Make-Up Gain AND and Output level, but the basic idea is the same: There’s something to control the level of what feeds in, and the level of what feeds out.

Now, most equipment has some sort of meter - ranging from a couple of LEDs to a mechanical VU meter, and that meter is usually located after the Output level somewhere, but sometimes it is switchable, which is nice, because then you can see what your input level is before it processes things, compare it to the output level, etc.

SO, you’re always aiming for Unity Gain with each piece of gear, and what we want is the input level set so that the meter reads nominal, and the output level feeding out is at nominal. To do this, we set the level control knobs at the position that gives us Unity Gain, and that position is usually marked with a ZERO or some such.

The next step is to feed the Output of one piece of equipment into the Input of the next piece of equipment. NOW... this might get a bit tricky, so we have to cover Operating Level quickly.

OPERATING LEVEL

Simply put, Operating Level is the amount of power a piece of equipment wants to see at its input and output. This is what you’ll usually run into:

Mic Level is the level of power coming out of a microphone and it’s REALLY LOW. How low? Like -50dBu. What does that mean? It means really low. Don’t worry about it. Mic Level is so low that you can’t do anything with it until you bring it up to Line Level. That’s what a Mic Preamp does - it brings a Mic Level signal up to Live Level.

Instrument Level is the amount of power that comes out of a bass or a guitar with a passive pickup. It’s also really low, and in my mind it's basically the same as a mic level signal. For those advanced campers, I’m ignoring impedance today. If you don’t understand that previous sentence, that’s fine. You'll get there eventually.

Line Level is the level of power flowing through gear - consoles, tape decks, compressors, coming out of synths and keyboards, etc. There are three possible line levels: Consumer, Pro Audio and Broadcast.

Consumer line level is -10dBV. This is the line level of home stereo equipment and also output level of a lot of synths and keyboards. What does -10dBv mean? Well, it means if the thing is set to unity you have -10dBV feeding in and -10dBV feeding out and that’s all need to know. -10dbV is a LOT more powerful that -50dBu. Ignore all the V’s and u’s for now. -50dB is less than -10dB, right? Close enough for rock and roll today.

Pro Line Level is +4dBu. This is hopefully what the majority of equipment is at in your studio. Can’t tell? Pro equipment uses bigger, heavier, tougher connectors. Consumer stuff uses shitty little connectors. With pro line level stuff, if the meter is at 0 and gain is at unity, you have +4dBu feeding in and out. And it’s got a lot more power than -10dBV consumer stuff. Again, ignore the V’s and u’s and just look at the numbers for now. +4 is more than -10 and a lot more than -50.

Broadcast Level is +8dBu. I don’t even know how common this is anymore as I don’t do work in radio stations or TV, but it is 4dB hotter than Pro Line Level. You can probably ignore this.

Speaker Level is what comes out of a power amp and plugs into a speaker. It’s like a SUPER BOOSTED line level. Line level is too weak to move the diaphragm of a speaker, so a power amp is needed to crank shit up. A dumb idea is to plug the output of a power amplifier into anything other than a speaker. Poofsky.

Again, hopefully your equipment is all +4. It won’t be - you’ll have some guitars and keyboards and, of course, mics, and they won’t be at +4, but that’s why you have preamps. Plug the mic level and instrument level and consumer level stuff into a preamp, and add gain to get it to read 0 on the meter. Now, going out of your DAW or mixer, you might be feeding into a pair of “consumer level” active monitors. Usually there’s a switch so you can match the Pro Level output gain to the consumer level input gain. If you’re thinking the switch knocks off at about 14dB of gain, you’re right.

GAIN STAGING WHEN TRACKING

Ok, here we go.

Starting with a mic preamp: Turn the input gain all the way down. Set the output gain to Unity. Plug in the mic. Have the singer or musician play, and turn up the Input Gain until the meter is reading 0, or nominal. Done. If the meter has slow ballistics and you’ve got drum fast transients, run the meter a little lower, like -10 or -15. Slow transients? You can run it a little hotter and increase your S/N ratio. But really, unless you’ve got slow meters and fast transients, park it around 0 on the meter and move on.

Plug the output of the mic preamp into whatever is next - a compressor, an EQ, etc. Set the EQ flat, set the compressor threshold all the way up, etc. If there’s an output level control or makeup gain set that to Unity, that is, to 0 or to 8 or whatever. Watch the meter. If there’s no input to adjust it should hang out around 0. If there’s input gain then set that to Unity or play with it until the meter is at 0. Now, as you adjust the EQ or the compressor to change the signal, the gain will change, so you’ll have to adjust the Output level perhaps, or the input level - it depends on how crazy the gain change might be.

You keep going until you reach what your final stage is, either an analog tape deck or a digital tape deck, or a DAW, or perhaps a live mix console... whatever.

Hit analog tape at 0 on the meters, unless it is drums, in which case hit it a little lower unless you want distortion. Hit digital tape decks, like ADATS and DATS and Sony DASH machines as hard as you can without going over.

Hit DAWs at around -18 to -12dBFS. Yes, you can hit it harder, but for now, you want things bouncing around in that -18 to -12 area.

ADJUSTING LEVELS

Now, where do you adjust the level if things are hot at the tape deck or the DAW? Well, the best place is the Mic Preamp input. Yes, it will screw up your compressor settings a bit, but that’s life and engineering and you’re paid to tweak things. The mic preamp is doing almost all of the work here, so that is where you adjust it. When tracking, get in the habit of setting levels at the earliest spot in the signal chain, at the preamp. And NEVER (and I mean this almost absolutely) use a fader to fine tune your gain. The exception: if you’re riding levels while tracking, then use the fader. Other than that. Leave it at Unity. Have I made this clear?

MISMATCHED OPERATING LEVELS

When you’re feeding something low level into something higher level, you want to adjust things at the INPUT STAGE of the higher level piece of gear. So, with a low level mic going into a preamp, you tweak the gain of the preamp. If you’re plugging some strange shitty consumer -10 compressor you bought into a +4 thing, add gain using the +4 device’s input trim.

What if you feed +4 into -10? Well, turn DOWN the output of the +4 device by about 14dB so you don’t overload the -10 device.

FINITO

AND... there you have it. Gain Staging. It’s easy. This blog post is done. What follows below is a bunch of common sense hints that are worth following.

See ya next week.

COMMON SENSE HINTS

1) Nominal is nominal is nominal. If the operating level of each piece of gear is the same, then setting everything to nominal will work. When in doubt... NOMINAL.

2) Set levels as hot as possible without getting distortion. You’re always trying to maximize the s/n ratio.

3) Use the hottest mic possible. It’s really hard to overload a modern condenser, let alone blow it out.

4) Preamps generally have a lot of headroom, so they can usually be run pretty hot. But LISTEN. Some preamps overload in a nice way, others crack and snap. And this sounds like shit. When in doubt, back it down a bit. You can always add distortion later, but you’ll never get rid of it once you have it.

5) Most mechanical (dial) meters are VU and have slow ballistics. Run your level lower on these when it’s percussive stuff, and at nominal for everything else, including entire songs. You can run your level higher on VU meters when the transients of the input signal are slow.

6) LED meters might appear to be fast peak type meters, but in my experience they usually have similar ballistics to a VU meter. Run some drums through it, run some vocals through it, watch how the meter responds. Or look in the goddamn manual.

7) Want to calibrate everything in your signal chain? Stick a guitar amp in the room without a guitar plugged into it, crank it up so it hisses (white noise). Throw a mic in front of it. Plug the mic into a preamp with the output at Unity and adjust the input gain to get it to 0 on the meter. Feed that through each piece of gear in your signal chain until you get to tape or DAW. No guitar amp? Mic the fridge. Or water running in the sink. Don’t get the mic wet.

8) The above is too much work? Turn your mic preamp input all the way down. Set the output of everything to Unity. Set the input gain of the rest of the signal chain to Unity. Provided everything is at +4 operating level you’re done.

9) You’ll make WAY less mistakes when patching things if you always think OUTPUT feeds the INPUT, and always plug stuff in that way - the patch cable goes into the OUTPUT first then you plug it into the INPUT. If I’m doing a complex patch or I’m using a strange patchbay (and I am old and my brain is turning to shit so most patchbays are strange to me these days), I say in my head or even out loud, “The Mic pre output goes into the TLA-50 input. Then I grab the next patch cable and “The TLA-50 output goes into the Pultec input...” I have always done this, even when I was young and smart and fast. It reinforces the signal flow in your head, it eliminates almost all patching errors, and it keeps you from looking like a fucking moron during a recording session because you patched something wrong.

10) When patching in STEREO, put the patch cord for the LEFT signal in your LEFT hand and the RIGHT signal in your RIGHT hand, and then do the above: “The preamp outputs feed the compressor inputs..” Always put left in left and right and right and you’ll reduce the chances of cross-patching something to like 0. I don’t know why schools don’t teach this shit. It will save your ass.

11) Another stereo hint. I always put Left side signals on Odd numbers and Right side signals on Even numbers. And I always put them beside each other. SO, if I have a stereo pair of mics as overheads, the left is plugged into 9 and the right into 10, as an example. I NEVER break this rule. live mixing too. If something's on the Left side of the stage I want it on the Left side of the console so I can grab it with my Left hand. It keeps everything straight in my head. Of course, if something happens to one of my hands, like it gets bitten off by a pony, then it’s mono for me.

12) Tape stuff down if you don’t want it bumped.

When last our heroes met they were discussing Dynamic Range and Nominal Level.

Dynamic Range is the space from the Noise Floor - the spot where the signal is covered up by noise (very very quiet) to the Distortion Point, which is the spot where harmonic distortion becomes very noticeable.

Nominal Level is a semi-arbitrary spot within the dynamic range of a piece of equipment that the manufacturer has decided gives you a high Signal to Noise ratio and enough Headroom. It’s based on their knowledge of the design of their equipment and conventions in the audio industry. What’s problematic about Nominal Level is that what is actually being measured can be different on each piece of equipment.

Manufacturers put meters on their products, and now is the time to understand how meters relate to nominal level.

SPEEDOMETERS and the VU METER

If you’re in a car in New York in the US, and you’re driving at the speed limit, your speedometer would look like this:

If you’re out west in Wyoming, the speed limit is higher, and the speedometer might look like this:

If you’re in Germany on the Autobahn, the speedometer might look like this:

But you! You’re a crazy traveling’ bastard! Bit confusing if you’re driving in Wyoming at 70 mph, then NY at 55 mph, then you go to Germany and it’s not even miles per hour, it’s now kilometers per hour, then suddenly you’re in NY, speeding around on the highway and a Scorpions tune comes on and you have a flashback to Germany and suddenly you’re going 130 mph. A cop pulls you over, tasers your stumpy ass, etc. True story.

So, let’s say we invent the UNIVERSAL SPEEDOMETER. And it looks like this:

It doesn’t show how fast you’re going in some specific unit, it just shows you how far you are away from the speed limit, and it’s calibrated to wherever you’re driving. In NY,  we set the Universal Speedometer to “55” and if we put the needle at 0 we’re at the speed limit. In Wyoming, 0 means we’re going 70mph, in Germany 0 means we’re going 130kph. So, now, no matter where we go to drive, with the Universal Speedometer, we get the car up to the 0 and we’re fine. Who cares about the exact number in mph or kph because we know WE ARE AT NOMINAL.

If we need to pass someone, we can speed up, the meter goes up, and we use up some of our HEADROOM to get better performance and speed and get around some car in the way. And if the meter is really low, we know we’re close to the NOISE FLOOR and driving too damn slow.

The Universal Speedometer is a VU meter. It doesn’t tell you what the nominal level is, it tells you something much more important, which is: Are you at nominal level or not. And as long as we slap a VU meter on all of our gear, we now know exactly where to park the level at: 0. 0 is nominal level.

And it really doesn’t matter what the meter looks like, if it’s LEDs or LCD or mechanical or virtual, nominal is nominal.

VU Meters and Average vs. Peak

Now, even though I sort of implied that all meters are the same they aren’t. It’s audio. There’s always something to fuck up the simplicity.

Meters have a speed of response, and it can be different from meter to meter. Some meters are fast and others are slow. Some meters measure peak energy, some meters measure average energy.

So, let’s say I’m in a room with the lights off. The room is dark. Let’s say I flip the light on for a split second and then click them off again. The room is bright for a moment, but then it’s dark again. So, it’s dark on average, but there was a “peak” moment of light. If I start flashing the lights on and off quickly, you might start perceiving the room as being “bright” rather than “dark,” because the AVERAGE light in the room across time is higher.

SO, if you are “set” to notice the average brightness of the room, you’ll respond one way, and if you’re set to notice “peaks” you’ll noticing something else.

So, how a meter responds depends on if it’s “noticing” the average or the peaks, or some sort of in-between.

VU meters are set to notice the average power of a signal. So, if you run something with a slow attack through it, like a violin or a guitar or a voice, or an entire finished song, the meter gives you a good idea as to where your signal level is at. But an instrument with fast transients, like a drum, the attack happens too quickly to be noticed by the meter. And by the time the meter responds the transient has already gotten through and it could be WAY above nominal and actually causing distortion, but a VU meter doesn’t tell you that.

Meter Ballistics

Fast responding peak meters are tracking the transients of signals and are basically telling you how much headroom you’re using up. This is really useful information, but it’s different than what a slower meter is telling you. Both types of meters are really useful, especially together, which is why so often a VU meter has a peak light on it.

Meter Ballistics is how fast the meter responds. You can get an idea of this by looking at the meter. If it’s a mechanical meter and it has to swing a little needle around, even if it’s really fast it’s never as fast as an LED meter can be. But an LED meter might be electronically slowed down to respond like a slower mechanical meter - virtual meters on your DAW might be set to respond to average rather than peak, too. You can also look in the manual and find out if the meter is measuring peak or something more average (look for the letters RMS, which basically means "average.”).

Often, there are two meters within a meter set, and one is measuring the average power, and the other is measuring peak. On a VU meter, the needle is indicating average while an LED might be indicating peak. On an LED ladder meter, you might see both things happening—the LEDs lighting up from top to bottom, but an occasional single LED lit up and hanging around at the top. Again, we have average vs. peak. When you're metering violins, which have slow attacks, you’ll notice that the two meters read very close to each other, whereas if you’re metering drums, there will be a much bigger difference between the two.

Displays on the DAW and Nominal Level

On your DAW, you're probably seeing an LED bargraph display by each fader, and you can also add virtual meters of any type to channels, buses, etc.

Typically on your DAW, you're going to be letting those channel meters bounce around on average between -18 to -12dbFS. Oh my... yet more dB stuff to talk about. Eventually.

SO... now you know pretty much exactly what that meter is doing, and you know what nominal level is, and how this all fits together. Next week, we’ll talk gain staging and setting levels.

Thanks for all the good feedback. Much appreciated.

Let’s put the whole thing together today. How Noise, Distortion and Signal Level all fit together. How it all works.

DYNAMIC RANGE

All devices in audio - from a human voice to a mic to a preamp to a converter to a console to a power amp to a speaker to a human ear, all have a lower limit and an upper limit.

The lower limit is self-noise, the noise floor.

The upper limit is the distortion point, which is the spot that harmonic distortion becomes a big problem. By the way, the manufacturer decides what is unacceptable harmonic distortion.

So, that is the playing field in audio - from the Noise Floor to the Distortion Point. And we call that area the DYNAMIC RANGE.

Dynamic range can be huge. Your ear has a dynamic range of around 180 dB. You can hear from an ant picking its nose to something as loud as a gunshot about a foot from your head. But don't shoot a gun off near your head. If you’re listing to things ON PURPOSE and without HEARING PROTECTION louder than 112 dB you’re crazy. We will, of course, talk about dB later... much later...

Mics have dynamic ranges around 120 dB, which is comparable to a human ear under normal circumstances. Mic preamps have dynamic ranges all over the place, from as high as 130 dB down to 90 dB or even less. Digital audio recordings can have dynamic ranges well over 100 dB, depends on how they’re designed. Analog tape sorta sucks - lotssss of hisssss - dynamic range can be in the 70’s down to the 60’s even with noise reduction. Radio stations barely hit 50 dB of dynamic range.

HEADROOM

We need to find a place within the dynamic range to set our levels so we avoid being a shitty engineer. Let’s reason this out.

Ok, we do want levels as high as possible, because noise sucks. But what if something unexpected happens? If we set a mic preamp level to right under distortion, and then the vocalist moves in a little closer to the mic, or sings a tiny bit louder, the increase in power can clip the mic preamp, and you’ll hear distortion. So, we need a little bit of safety margin up there so we have some room in case something gets unexpectedly loud. That’s HEADROOM.

What are typical headroom figures? It’s all over the place and it's anpthor thing that is rather arbitrary. On analog tape decks we were usually recording to give ourselves about 9 dB of headroom on the tape. Mic preamps usually have very good headroom - from 18 dB to 26 dB or even higher. Like dynamic range, it’s variable and depends on the type of gear and the manufacturer, and the engineer.

NOMINAL LEVEL

We want to set our levels as high as possible to keep our S/N (Signal to Noise) ratio as high as possible. And we don’t want to clip, so we’re going to give ourselves a little room on top - headroom. That level, which represents a compromise decision, is the NOMINAL LEVEL.

What usually happens is we have the musician play or sing, and we watch the meters and listen, and we set the level so that we have some headroom just in case. Nominal is generally a pretty high average high level. We set nominal level when we're engineering, but we're setting it in accordance with how manufacturers designed their gear.

Remember that meters jump and move, so visually, nominal level won't stay exactly in any one place, but will jump around a bit.

So, here is what it all looks like:

Dynamic Range is from Noise Floor to Distortion Point.

Nominal Level is a High Average level setting.

Signal to Noise Ratio is from Noise floor to Nominal.

Headroom is from Nominal to Distortion Point.

Signal to Noise Ratio + Headroom = Dynamic Range.

What are typical nominal level figures? It depends. It depends on the type of gear you’re working with. The nominal level of an analog tape deck is measured one way, while the nominal level of a mic preamp is measured another way, while the levels on your DAW are measured yet a different way.

If you are thinking, “Wait. The nominal level is basically different all over the signal chain. Manufacturers decide where it is, engineers decide where it is, the type of gear affects it. Jeez Louise, how do I set levels so everything sounds rocking’ good?"

You use meters and common sense. And experience.

OK! It’s been a pretty long slog through this stuff, but hopefully you’re a bit clearer on it all. I can be confusing, and usually when I explain it I can wave my arms around and demonstrate stuff and it makes more sense and I look like a nut.

I can’t emphasize how important knowing that diagram - Dynamic Range with Nominal in the Middle, is. If you can hold that diagram in your head while you’re setting up gear and getting your levels, your recordings will improve immensely. I want you all to be great engineers.

 

Previous posts have talked about what happens when audio signals get too powerful, too loud. Distortion is what happens. That ain’t the same pork chop is what happens. For a refresher go here.

This week, let us look at kinda the opposite. If distortion is what we hear when things are too much, what is at the other end, the quiet, weak side of things?

Noise is at the other end.

NOISE and SIGNAL

Noise is anything that you’d rather not hear, basically. And Signal is the thing that you actually do want to hear.

• Watching sports on TV and hearing the announcer clearly = signal
• Spouse/Significant Other/Toddler w/Poopie Diaper/Pet Cat in Heat = noise

When we like how noise sounds, it isn’t noise anymore. It becomes signal. What is noise and what is signal is rather arbitrary.

Example: you’re recording drums. The snare is leaking into the tom mics. The snare leakage is noise. So, you put a bunch of gates on the tom mic and spend 45 minutes getting rid of all the snare leaking into the toms.

Cue the band, cue the drummer. Do the count in 1 2 3 4...

And it sounds like shit. The snare sounds like you mic’d up this monkey:

Because the leakage into the tom mics was actually HELPING the snare and the whole drum set. So, you pull off the gates, and now that leakage, which you previously thought of as noise, became part of the signal.

At a live show, the audience is noise, the sound of the band is signal. In your car, the radio is signal and the sound of the engine, the wheels on the road, the wind rushing past the car is the noise. And suddenly, you hear a “pop” and then a flapping sound outside the car, and now the radio becomes the noise, so you turn it down to hear if you have a flat tire, because the road sounds outside the car are now the signal.

Please note that when Noise gets in the way of hearing the Signal there is a Problem.

SELF-NOISE and the NOISE FLOOR

Self-noise is the noise that a device makes when it’s turned on and power is running through it. If you aren’t running a signal through your console or your interface, and you turn up the speakers, you’ll hear hiss. Hopefully, the hiss will be very quiet, and you won’t hear hum along with it.

Hiss is the sound of the device working, the sound of electrons running around the circuit. This hiss is self-noise. All devices that have power flowing through them make noise. Your body generates self-noise, unless you’re dead.

At night, if it's really quiet, you might hear a whooshing in your ears and perhaps a very very quiet whining sound. If you put a cup or a shell to your ear, you’ll easily hear the whooshing — remember as a kid when you put a shell to your ear and could hear the ocean? It wasn't the ocean. It was blood flowing through your ear, reflected back into it by the shell. You’ll hear the same whooshing if you put a coffee cup up to your head, rather than a barista named Amanda yelling or a tractor on a coffee plantation in Guatemala.

The whoosh is your blood flowing. The whine is your nervous system working. This is really quiet stuff, about the quietest things you can hear. We call this the Threshold of Hearing. This is like the sound of an ant picking its nose.

Now, you don’t normally hear this stuff in your day-to-day life because everything around you is noisier. Noise causes masking when the signal gets too quiet and falls below the noise. The limit to how quiet a signal you can have is how low the noise is. You can’t really go below the noise, so that bottom limit is called the Noise Floor. You can’t get lower than the floor, right?

The noise floor of a piece of audio equipment is typically really low. Guitar amps have more noise — how often have you heard a sustained guitar note decay away into the hiss of a guitar amp? It goes below the noise floor and then you can’t hear it anymore.

The noise floor is a shifting, somewhat arbitrary thing. When you’re mixing live, is the hiss through a PA system really an issue? It might be during the sound check when the venue is empty. But once it fills up with people, the noise floor caused by the audience is considerably higher than the hiss of the PA and effectively masks it. And if your PA hiss is heard above the audience... jeez, you suck, you stumpy bastard.

SIGNAL to NOISE RATIO

You’re in the coffee shop talking to a friend. The friend who is talking is the SIGNAL — the thing you want to hear, and the background chatter, espresso machine sounds, etc., are the NOISE — the things you don’t want to hear. The louder the coffee shop gets, the louder your friend will have to be such that you can hear their signal over the noise.

Signal over Noise... let’s call this the Signal to Noise Ratio. S/N ratio. If this is a low number, the noise is loud and it's intruding on the signal. If this number is high, the noise is quiet compared to the signal. So, now you understand this bit more:

First distortion, now noise... soon you’ll be able to just read this stuff.

Greater than 81dB of signal compared to noise isn't brilliant, but it's ok.

The S/N ratio is different for different types of equipment. It’s comparatively huge for microphones and really good preamps, and much less so for cheaper equipment, guitar amps, PA systems, etc.

Noise builds up. When recording, the ambient sound of the studio feeds into, oh, say a condenser mic, which adds some hiss, and then into a preamp, which adds a little more hiss, and then into various converters and devices, all of which add hiss. And all of this noise adds up, and that’s the noise floor. Then someone wacks a snare out in the studio, and that goes slamming through everything and it’s much louder than the noise. High S/N ratio. The snare rings out for a moment, then decays into the ambience of the room. And once it decays to a certain level, we’ll notice the noise again. S/N ratio is a fluid thing.

CAPTAIN OBVIOUS

This is frickin’ obvious but it must be said: you usually hear noise when things are quiet, when the signal is low and the S/N ratio is small.

Another frickin’ obvious thing that must be said: analog recording techniques were mostly developed to compensate for noise, especially tape hiss.

Tape hiss... the sound a piece of magnetic tape makes as it slithers over the heads of a tape recorder. The more tracks you have, the more tape hiss you get. Dolby, DBX noise reduction, noise gates, etc., were all developed to control tape hiss.

Digital recording was developed to totally get rid of tape hiss.

I can’t tell ya how much time I spent in my engineering career trying to get rid of noise. Automating mutes. Gates. Notching out things in the high end. Yada yada yada. I never really used DBX systems because I thought they sounded terrible, and if I was recording at a nice high level to really good tape, and was careful with muting, I could make a virtually hiss free record.

You can’t hear hiss when the band is cranking.

I cannot understand why anyone would make a plug-in that adds “authentic analog noise” to the signal chain. Restaurants are allowed to have a very low percentage of cockroach bits and rat crap in the food. Would you add cockroach bits and rat shit when you cook at home to get that “authentic restaurant taste?” Fuck no.

Next week we’ll put all of this together and figure out dynamic range and metering.

Be well. Stay safe.

Distortion, in the simplest sense, is when what comes out is different than what goes in. Think about eating dinner and what happens six hours later.... that ain’t the same pork chop, is it?

Something in the process, in the piece of equipment, is changing the signal.

Usually, what happens is that the piece of equipment runs out of ability to accurately reproduce the input signal. But what the heck does this mean, actually? Let me give you a few examples. If you get this clear in your head, so many things will suddenly make sense.

Let’s Look at a Speaker

A simple speaker is a cone of paper that’s being pushed forward and backward by an electromagnet (the coil). There’s a flexible springy area around the cone of paper called the surround, and the base of the cone is attached to another springy thing called a spider. The surround and the spider are attached to a frame called the basket. The spider and the surround allow the cone to move forward and back while supporting it in the basket. When the cone moves forward and back it pushes air forward and back. The coil is what causes the cone to move - pushing it forward and back, depending on the signal that’s fed into it. Like this diagram:

A simplified speaker

If you feed in a low frequency signal, the cone moves back and forth slowly, and as the pitch goes up, the cone moves back and forth faster and faster. If you feed a weak signal in, the cone moves back and forth over a small distance.

Linear reproduction of an input signal

If you crank the power up (the volume) the cone moves back and forth and covers a longer distance.

Louder but still linear...

However, the cone can’t move an infinite distance back and forth. There will come a point when the surround and the spider are completely stretched and the cone can’t move any further. The speaker has run out of ability. Does that make sense?

Speaker can’t move enough and the output is clipped

When the cone has ability to move, it does so, and it can accurately track the up and down of the waveform. When the surround and spider run out of stretch, however, the cone can’t track the waveform. It moves as far as it can, can’t go any further, so it essentially jams - it stays still. And the waveform that comes out of it is now different from the waveform that went into it. And if you look at the waveform the speaker is emitting out, it’s clipped — it’s squared.

Remember last week, when we mixed odd order harmonics in with the fundamental and caused a square wave? This is exactly what’s happening with the speaker, but in reverse. When its movement is “jammed" it generates a bunch of harmonic distortion. And if we looked at that waveform, we'd see a square wave. That doesn’t look like the original pork chop, does it?

So, a speaker has a certain amount of ability to move and reproduce a waveform in a linear (linear means what comes out is the same as what went in) manner. If we put in too much power, we run the speaker out of ability, and the result is distortion.

How much ability does a speaker have?

It depends on things, but to look at it very simply, if a speaker is rated to 150 watts, it has 150 watts worth of ability.

Let’s Look at an Amplifier

Ok, so a speaker is rated to 150 watts, so that means an amplifier which is rated to 150 watts... hmmm... that means the amp has 150 watts worth of ability to reproduce the signal, right?

EXACTLY!!! That is exactly right. Amps - and not just power amps or guitar amps, but the little tiny amplifiers stuffed into the circuit boards of your recording console, also have only so much ability. When they run out of ability to reproduce a signal, and when that happens, the result is distorted output, non-linear (what comes out isn't the same as what went in) output.

As a signal feeds in, the amplifier uses power to reproduce it. As we turn up the input signal, the amp needs more power to track the waveform in a linear manner. But there isn’t infinite power. The amp isn't connected directly to the sun. Eventually, the amplifier cannot draw anymore power, and it loses its ability to track the waveform, and it squares the wave, just like a speaker that runs out of springiness.

Amplifiers use power to reproduce signal, and if they don’t have enough power, they generate harmonic distortion. A simple way to look at, but a very useful way to look at it.

Everything Runs Out of Ability

A singer can only get so loud before their vocal cords can no longer move — they physically slam into each other in the voice box. The vocal cords run out of ability. The resulting vocal has a growl to it — distortion. Harmonic distortion. And if the singer keeps doing this, they start losing their voice, and if they do it enough, they can do permanent damage, just like you can blow a speaker out, or blow up an amplifier.

Your ears. Your eardrum can only move so far. The little bones in your ear (there are three little bones in each) can only move so far. The little hairs in your cochlea, which turn sound waves into nerve impulses, can only move so far. They run out of ability to move, to track the waveform as it gets loud, and the result is distortion. And you can hear this distortion, and you can feel it. And if you consistently run your ears out of ability, you’ll get tinnitus. Or, if the waveform is loud enough, you can blow your eardrum out — literally tear it apart.

Stuff certain mics into a kick drum and one good hit can break the diaphragm in a split second, and if it doesn’t break it, the mic will clip the waveform as it runs out of ability to move and starts generating harmonic distortion.

Do digital processors run out of ability? Yes. Digital processors do math, and you can basically use up all of the processor’s ability to perform mathematical calculations. The result however, isn’t harmonic distortion. It’s a loud click or static "scratching" sound, and if you feed that through a speaker, the speaker runs out of ability to reproduce it almost immediately, which is why it sounds awful and is really bad for your speakers. And your ears.

Everything runs out of ability, and when it does, you get that unrecognizable pork chop.

A short post this week, but an important one if this is stuff you’re trying to wrap your head around. Hit us up on Facebook or Discord if you’ve got a question.

 

Last week I wrote about Bias, and how if an amplifier or an audio component isn’t biased correctly it might not work or it might cause a lot of harmonic distortion.

This week: what the heck is harmonic distortion, and what the heck is a harmonic?

What’s a Harmonic?

So, first of all, what is a harmonic.

If you take a note, like a C, and play it on a guitar or a piano, because of the physics involved, not only do you hear the note C, you also hear, very quietly, other notes that are mathematically related to the C you’re playing. Like, you might hear a C an octave higher, and then another octave above that, and you might hear an E and G mixed in there as well. It’s actually quite a bit more complex than that, but the point is that if you play a note on virtually any instrument, you get more than the single note that defines the perceived pitch. That other stuff are the harmonics.

I found this video, which is an ok explanation - it could be clearer, but if you want to take a moment, a quick watch might help you understand some of the physics involved.

The harmonics of a note are caused by the physics of vibrations, and by the construction of an instrument, or of a person’s face if we’re talking about the harmonics of a sung note. And, in fact, the harmonics of an instrument are a huge factor in why an instrument sounds the way it does. A guitar with steel strings has a different set of harmonics than a guitar with nylon strings. The two types of guitars have a lot in common in terms of harmonics — you can tell they’re both guitars — but the steel string is typically brighter and more metallic, and that’s because of its harmonics.

The harmonics have a mathematical relationship to that C you played (we can call that the fundamental), and the particular pattern of harmonics is what makes an instrument recognizable as an instrument. And some patterns of harmonics sound better to our ears than other patterns of harmonics.

In fact, harmonics do tend to be high frequency information, and we will see why that is important in a bit.

What’s Harmonic Distortion?

Harmonic distortion is when harmonics are added to a sound, a signal, that aren’t there in the original signal.

Back to playing a C. If we played a C on a very simple instrument, like a flute, you would get a very pure sounding C — it wouldn’t have a lot of extra harmonics happening, unlike a guitar, for instance. The complex body shape and construction, the physics of a guitar, actually add harmonics to the C. It’s a bizarre way to think of it, but you can consider a guitar a generator of harmonic distortion. So is a piano, a trombone, a human head, throat and body, etc. These all are sort of "harmonic distortion generators."

Electronic components (amplifiers, etc.) also add harmonics to a signal. Usually a well-designed circuit adds a very, very tiny amount of harmonics, and we can’t hear it because it's such a small amount. That is also harmonic distortion. A badly designed circuit can add enough harmonic distortion that one can clearly hear it. A circuit can also be adjusted in such a way that it generates a lot of harmonic distortion.

There are amounts of harmonic distortion that can be very noticeable, and certain patterns of harmonics are more noticeable, and some patterns sound good, and some sound like shit.

Harmonic Distortion = Sonic Finger Print

All the elements in an audio recording signal chain add some amount of harmonic distortion. Microphones, speakers, preamps, compressors, power amps, guitar amps, effects pedals — all of these things add harmonic distortion. Some are designed to add as little as possible, and others are designed to add huge amounts. Microphones sound different from each other, in part, due to the harmonic distortion they add, as do speakers, mic preamps, etc.

As mentioned earlier, some patterns of harmonics our ears like better than others. Tubes, whether in compressors or guitar amps, tend to have harmonic distortion that our ears like. Tubes are often described as sounding “warm.” That’s the mathematical relationship of the harmonic distortion (the harmonics added) of a tube circuit.

Solid state equipment also has distinctive harmonics patterns that it adds to a signal. That’s part of the reason Neve sounds like a Neve, and a Mackie sounds like a Mackie.

THD

THD stands for Total Harmonic Distortion, and it’s a measurement of the amount of harmonics a piece of equipment adds to a signal passing through it. The manufacturer of the equipment will usually specify this as a percentage at certain frequencies, something like, “Less than 0.5% percent THD from 20 Hz to 20 kHz at full rated power.” Some manufacturers specify it in much looser terms: “Less than 1% THD.” Generally, the better the gear, the lower the % of harmonic distortion, and the more specific the manufacturer will be about it.

What’s a lot of harmonic distortion, and what’s a little? Depends. 0.5% is pretty good for a tube component, but pretty awful for a solid state component. A really high end solid state device can have incredibly low harmonic distortions - like 0.002%.

Tube mics are typically in the 1% THD neck of the woods. 1176 Limiters have around 0.5%. A Neve 5211 is down around 0.0015%. Obviously, guitar amps designed for distortion have much higher amounts of THD. And also obviously, is that the more you turn stuff up (increase the power), the more you increase harmonic distortion.

But, THD is really only a small part of the harmonic distortion story. There’s also the “sound” of the harmonics added, the math of their pattern, that make a huge difference.

Even vs. Odd Harmonics

I made a video about this next bit, so you can watch the video and skip ahead, or watch it and then read so you understand it all that much better.

Quickly, let's look at the way a string vibrates.

A vibrating string is very complex. Back to our C, if you fret and pluck a C on a guitar, you'll get a nice loud fundamental, vibrating at 261.63hz. Let's round that to 262 to make the math easier.

So, we have a string vibrating at 262hz, but it's also vibrating at twice that - 524hz. But it isn't vibrating with as much power, so this 1st harmonic is much quieter than the fundamental.

There's also a harmonic vibrating four times as fast as the fundamental — 1048hz.

When these vibrations all happen on one string, the result is a much more complex waveform than any fundamental or harmonic by itself.

There are also other math things happening there. There's an E, which is the third, and which is around 1.25 times the fundamental.

Fundamental and the third - the basis of a major chordThese harmonic relationships that sound good to our ears tend to be even number multiples, often called even order harmonics. Our ears tend to not like odd number multiples - 3, 5, 7, etc. These particular odd harmonics sound kinda ugly to our ears — the 7x is especially dislikable, and they tend to square the wave off...

In general, our ears think even harmonics sound better than odd. In general, tube equipment generates a lot of even harmonics. Does that explain to a large extent why everyone likes the sounds of tube amps?

Many Things Explained

Understanding some of the math of harmonics also explains why distortion seems to make something sound brighter: because what you're adding is harmonics ABOVE the fundamental, and those harmonics stack up and increase the apparent high frequency tonality of a sound. It also explains why too much harmonic distortion can sound harsh and painful — it's causing a lot of high frequency activity, and our ears don't like that very much.

Now, some of you might be thinking, "Even on a really good day people can only hear up to 20kHz. If I have something at 8kHz, then its harmonics are at 16kHz and 32kHz and other frequencies, all high, and most of them beyond the range of hearing. How can this possibly affect what we hear?"

The answer is that we can sense frequencies we can't clearly hear, and very high, over 20kHz frequencies can affect the equipment we are using, especially digital stuff, and we can hear that effect.

Some stuff you just have to take on faith, you stumpy bastard.

SO... download a PSC demo or buy one, flip it around to the backside, turn up the PREAMP gain until you hear some distortion. Then swap around the three different sets of tubes we've thoughtfully included, and you'll hear the quality and frequency response of the distortion change. This is because we've modeled the different harmonic distortion characteristics of them into the PSC.

On the AIP, you can do this: Switch on the PSP and turn up the input gain until you hear distortion.

Go around to the back panel and click between TUBES , TAPE and SOLID STATE to hear three different variations of harmonic distortion. Turn up the INPUT TRIM to make the effect more easily heard. Don’t forget that turning up the trim can add a lot of gain and make things louder. Use the OUTPUT TRIM to readjust output gain down.

Next week, we'll talk about why cranking things up causes an increase in harmonic distortion, and we'll start talking about some recording techniques that take advantage of the physics involved.

LET'S START WITH BIAS

We're going to start this series of educational posts discussing Bias. There are reasons I want to start here, rather than something more elementary like dynamic range or “what is a dB” or some such. If you understand bias, you’ll understand a lot of other concepts, and things like dynamic range and harmonic distortion will actually make more sense when we get to them. And if you understand bias, our plug-ins will make more sense to you, especially since almost all of them have a tweakable bias control on them.

LOTS OF AMPLIFIERS

Analog recording equipment is made up of a bunch of components, things like tubes, transistors and transformers, etc. And digital plug-ins are all simulating the characteristics of those analog components.

Generally, in a piece of analog gear, no matter if it is an EQ or a compressor or a mic preamp, the heart of it, the thing that makes it work, is some sort of amplifier circuit. So, for the rest of this article, when I write amplifier or component, I am NOT referring to a guitar amp, or a mic preamp or a stereo power amp; I'm referring to a little circuit thing stuffed down in all the analog gear you will ever run into. It might power a preamp or an EQ, or control the gain of a compressor—all of these are amplifiers of some sort.  A recording console has thousands of amplifiers in it.

Amplifiers in equipment can be based on tubes, or on solid-state components like transistors or OP amps, or some sort of combination. Obviously, if you've got a bunch of amplifiers in a device, they're going to contribute a lot to the sound and character of the device, which is why tube EQs and compressors sound "tubey" and Neve EQ's sound "Nevey." The amplifiers inside the gear impart a particular sound.

AUDIO CIRCUITS HAVE A SWEET SPOT

Amplifier circuits of any type—tube or solid state—actually don't want to work properly. In some cases, they don't want to work at all. They are very particular about the amount of input fed into them, and they can be very particular about power in general. And unless power is handled just right, a component might not work, or work like ass, or work inefficiently and burn out quickly. They have a sweet spot.

If you feed in too little power, you’ll be below the sweet spot, and for a lot of components, they simply won't pass sound, or if they do, they're very quiet, or really noisy. If you feed in too much power, you’ll be above the sweet spot, and while the component will work, it might be distorted or otherwise bizarre-sounding.

Weird shit happens outside of the sweet spot. It’s like frying eggs. If you set the frying pan’s temperature too low, your eggs are going to be sitting in oil, getting all disgusting without getting cooked. Nice. Oily raw eggs. If you have the frying pan crazy hot, when you drop in the egg, the oil will come splattering out, making a mess, burning the egg and your face off (if you decided to lean over the pan like an idiot). The sweet spot of the pan is the right temperature, such that the egg cooks just fast enough that you have control, and you get the egg that you want.

LINEARITY AND NON-LINEARITY

For many amplifiers, the “sweet spot” is when its response is LINEAR. You’ve probably heard this term. Basically, when a circuit is linear, the signal that comes out of it is the same as the signal that feeds into it. Now, if it’s an amplifier, the signal coming out might be more powerful (louder), but if the amp is linear, the frequency response of the output closely matches the frequency response of the input. In simplest term, what goes in is what comes out.

If the level of power you feed in is BELOW the sweet spot area, the response is NON-LINEAR, if the device even passed signal. If you go OVER the sweet spot, the response is also non-linear, and what comes out of the component isn’t the same as what went in.

How is the output different if the component is non-linear? Well, there can be a lot of things different about the two signals, from changes in the frequency response to changes in the envelope, but the thing engineers are usually looking at when they want to discuss linear/non-linear is Harmonic Distortion.

We’re going to spend a lot of time on harmonic distortion, but not today. For now, all you need to know is that if a device is behaving in a non-linear manner, harmonic distortion typically increases.

Recap:
Linear: what comes out is the same as what goes in
Non-Linear: what comes out has been changed, and is different from what goes in.

AMPLIFIERS ARE LAZY

Now, here’s the problem, and this is true for many of the components in a piece of audio equipment. They only behave in a linear way across a small range of power. In many cases, this range is TINY. Outside of this range, the component is non-linear. So, the big trick to designing a circuit is to make sure all of the different amplifiers are getting a power level that makes them linear. That sweet spot level might be different for many of the components involved. Again, if you’re below that tight power range, the component might not even work, and if you’re above it, the component will add distortion.

So, for most amplifiers, there needs to be a BIAS signal added to it, and this makes the amplifier play nice with the audio signal. The type of bias signal can be very different depending on the component, and the circuitry involved can be different, but in general, all bias signals push an amplifier or a component towards efficient, linear performance.

BIAS: THE GUN TO THE HEAD

Bias for some amplifiers or components is basically a gun to the head. As an example, to get an analog tape deck to record, a super high-pitched, very powerful bias signal is mixed in with the much weaker audio signal and actually printed to tape. This bias signal is so strong that it forces the magnetic particles on the tape to actually record. Some types of transistor-based amplifiers also need to have a bias signal mixed in with the audio input signal, and then the bias signal, which you don’t want to hear, is filtered out.

In this case, bias is like... going to a birthday party place when you're a little kid and you want to go in the Ball Pit or use the Bouncy Castle or something and there's a height requirement, and you're too short. Your head needs to come up to a certain line by the door, and if it doesn't, no Ball Pit for you, you stumpy little bastard.

But you have special Bias Shoes that add a few inches to your height (they add power). You put them on, and now you appear tall (powerful) enough to get into the Ball Pit (linear amplifier performance).

BIAS: SETTING A CAR IDLE

Other types of amps use bias differently. In this case, the bias is sort of an efficiency adjustment. A device might work with a wide range of bias settings, but again, there is a sweet spot where it works best.

A way to think of this is to think about a car idling. When your foot is off the gas, with a normal gas-powered car, the engine runs but it doesn’t put out so much power that you can’t control the car by just holding down the brake. In fact, if it is set right, you should be able to drive the car, albeit very slowly, just by the brake. If the idle is set too high, when you lift your foot off the brake, the car jumps forward. You can set the idle so high that the car can’t be stopped even with the brake slammed down. The idle can also be set so low that the car coughs and stalls, or that when you step on the gas, it dies. If you set the idle just right, the car purrs like a kitten, can be controlled by the brake at really low speeds, and when you punch the gas, it takes off and there’s plenty of power to drive with.

A bad idle setting is hard on the engine, hard on the transmission, burns more gas and makes the car hard to drive. Likewise, for some types of amps, such as a tube amp, if the tubes are biased correctly, the amp is quiet, has plenty of power when you need to rock out, and the tubes have a long life. Set the bias wrong and you’ll burn out tubes, cause excessive distortion, or the output might sound dull and lifeless.

ADJUSTING BIAS

First of all, what you SHOULDN'T do is open up your mic preamp or your vintage tube compressor, locate the trim potentiometer that adjusts bias and then dick around with it. In physical audio gear, bias is generally set at the factory and it's not something the average person should deal with. Now, as gear ages, bias settings can drift, and as they do, the performance of the piece of gear will change. In some cases, the drift might make things sound better, and in other cases, it might make it sound worse.

But with virtual equipment, like Korneff Audio's Talkback Limiter, or Pawn Shop Comp, there's a bias potentiometer that you can adjust. And as you might think, if you turn the bias counterclockwise, the circuit's performance changes one way, and if you turn it the other, it sounds different yet another way.

What you're doing, in analog terms, is adjusting the overall output and linearity of the circuit. With one of our plug-ins, you're adjusting values in a computer algorithm that will change the harmonic distortion characteristics of the signal. Depending on the plug-in and the audio signal you're feeding it, you might even get changes to the envelope of the sound, the attack and release of the compressor, etc.

So that is BIAS. It’s a signal that makes an analog audio device work efficiently and have a linear output. If it's not set right, things will either not work at all or sound distorted or like ass in general.

Now, go be the damn audio genius I know that you can be.

If you have questions, feel free to post them up on Facebook or use the contact form up top and send us an email.