Gain Staging - Episode 2

Introduction Wakey Wakey! I'll come back to that later. Hi, I'm David Mellor, Course Director of Audio Masterclass and in this Sound On Sound podcast I'll be talking about Gain Staging. You're now listening to Episode 2 of this podcast series, so if you want to get the full picture, you'll need to listen to Episode 1 first. Episode 1 covers decibels, what Gain Staging is, my perspective from the old days of analogue through to VU meters and audio file formats. In this episode I'll look at Gain Staging during the recording part of the production process. Dynamic Microphones Like many aspects of audio, Gain Staging starts with the microphone. Let's start simply with a dynamic microphone, such as the classic Shure SM57 dynamic microphone. This is what I like to think of as the baseline professional mic. If you have one of these, you can make recordings of a professional standard. Conversely, if you have one of these and your recordings aren't sounding professional, it's not the fault of the mic. Actually, I'm going to start with the Shure SM58 because a). it's the same design, just with a different grille - that's according to Shure - and b). because Shure gives us some very interesting information on the SM58. So I said that Gain Staging starts with the microphone - well, that's not quite correct - it starts with the sound source. We have to know something about sound levels here, sound as actual sound travelling in air. As you know, sound levels can be compared in decibels, dB. Sound levels can be measured in dB SPL, SPL standing for Sound Pressue Level. Raw decibels are always a comparison between two sound or signal levels, two different sounds or signals or a signal before and after a level change. dB SPL is a comparison too, a comparison between the sound you're measuring and 0 dB SPL, which is taken to be the quietest sound the human ear can hear, or the threshold of hearing. In terms of sound pressure, which is like the pressure of your car or bicycle tyre, except sound, 0 dB SPL is set at 20 micro newtons per square metre, or 20 micropascals, which is the same thing. Your car tyre is 200 kilopascals, give or take, so sound pressure levels are tiny in comparison. Good old Wikipedia says 0 dB SPL is roughly the sound of a mosquito flying three metres away. Fine. I'd say a falling autumn leaf at 20 paces. My audiologist friend says she spent three years studying for her degree and I'm simplifying massively, but broadly, this is correct. Anyway, 0 dB SPL, that's our low level reference point as sound engineers, musicians and producers need to understand it. And where there's a lowest level, there must be a highest level. This isn't quite so easy to define. As sound gets louder, let's think of the sound being music. Firstly it's pleasantly loud, then excitingly loud, then it becomes painful and you want to go home. The level at which sound starts to become painful, the threshold of pain, isn't precisely defined in the literature, or rather, it's defined differently in different literatures, but 120 dB SPL is a reasonable figure to consider as being in the zone of discomfort. I nearly said 120 dB SPL is a reasonable figure to aim for, but I have my reputation as a responsible sound engineer to protect. That's 20 pascals by the way - your car tyre did 10,000 times better. So, how does this relate to the Shure SM58? I'm going to quote some figures from Shure. Remember that this is a dynamic microphone, there are no active electronic components inside, so the SPL limit of the microphone is when the diaphragm can physically move no further. Shure quotes the threshold of pain at 140 dB SPL, which I'd say is on the loud side but we'll go with it for now. Their measurements place the human voice up to 135 dB SPL, with a measurement taken 1" (25mm) from the mouth. A kick drum, bass drum if you will, may go higher than 140 dB SPL but Shure has never measured it more than 150. A trumpet however, 155 dB SPL in its upper register at 1". I'd have to say that you won't get the best ever recording of a trumpet at a distance of 1" but for a vocal or kick drum, well, that's in the range of normality. How will the SM58 cope? It'll cope just fine, its worst performance is at a frequency of around 100Hz. Note I said its worst performance - 150 dB SPL without undue complaint. This will give you an output of 1 volt - 1 volt! That's an awful lot of millivolts, which is normally what you expect from a microphone - a thousand in fact. That's just over 2.2 dBu. Capacitor Microphones Can your microphone pre-amplifier, or the pre-amp built into your audio interface handle this level into the microphone input? Possibly not. So to record a gorilla of a drummer kicking his hardest kick drum beats, you'd need to move the mic further away or put an attenuator between the mic and the preamp input - a pad, as we normally call it. I'll say more about pads in a moment. Using my trusty scientific calculator, at the 135 dB SPL given by Shure as the highest level of the human voice at one inch, the output of the SM58 would be a little over 170 millivolts or about -13 dBu, so your mic pre amp needs to be able to handle that. Don't worry, it can. I haven't tested them all, so it probably can. The Shure SM58 is a dynamic microphone, meaning that the sound is converted to an electrical signal by a diaphragm, a magnet and a coil of wire. Super simple. Nothing to go wrong. There isn't much to go wrong in a capacitor microphone either because all the main brands build their products to be very robust in the professional studio or live sound environment. But there are a lot of components inside. Cutting it short, the capacitor microphone needs to be powered and it needs an internal amplifier to boost the very weak signal from the capsule. That's the very short version and the key here is that there's an amplifier. An amplifier has active components and active components can distort or clip if given too much level. So where a dynamic microphone such as the Shure can handle absolutely amazing sound pressure levels, a capacitor microphone needs more careful consideration. Like I chose the Shure SM57 as a baseline professional microphone (OK, SM58, but they're almost the same) I'll choose the Neumann U87 as my baseline capacitor microphone. The reason I choose this is because a), it's good,without a doubt, and b), it seems to me to be in the centre of what capacitor microphones can be. Other capacitor microphones diverge in various measurable and subjective ways, but the U87 seems to me to be bang in the middle. So what I'll do is compare it with Shure's figures on how loud we expect sound to be. 135 dB SPL, that's Shure's figure for a vocal at one inch. Note that you don't have to sing that loud, it's just that some people can and do. What does the Neumann U87 have to say about this? 117 dB SPL - that is the maximum sound pressure level for 0.5% distortion. It's quite a lot lower than that figure of 135 dB SPL from Shure. So what are we going to do about this? Well, firstly, it's a matter of custom and practise. It is almost standard practise to sing into the Shure SM58 from a distance of one inch or even less. We saw Mick Jagger doing it back in the 1960s and we've been imitating it ever since. No, in reality, in live sound, there's always a battle against feedback, or howl round, if you prefer to call it that. The closer the mic is to the sound source, the less the risk of feedback. If you want to be technical, the loop gain needs to be less than one. But if you want to be practical, just get the mic very close to where the sound is coming from. And what we see in live sound or on TV, we mimic in the studio because a), we just do, and b), it gives the sound texture we expect. But that's not what we do in the studio with the Neumann U87 or any other similar capacitor microphone. We place a pop filter between the singer and the mic and the distance from mouth to mic will be somewhere around 20 to 25 centimetres, give or take. That's eight to ten inches, approximately, in the King's units. Sound levels decrease with distance. This is sometimes given as six decibels per doubling of distance, but it depends on the directivity of the source. But it does decrease and the difference between one inch and ten inches is significant. The result is that with most singers singing at normal levels, the U87 won't be bothered in the least. But try it with an opera singer and you may have trouble. Now, I'm not entirely sure that Shure (sure that Shure) measured an operatic soprano to achieve 135 dB SPL, but I can say from practical experience that standing close to an opera singer is a good way to explore the threshold of pain. So what do you do? As I said earlier, you can increase the distance between mouth and mic, but this will affect the sound, maybe for the better, maybe for the worse. But you will want to optimise the distance to get the best subjective sound quality for your purpose. So, there has to be a better way, and that is to switch in the mic's pad. Now, PAD passive attenuation device. I don't know whether that was ever the official designation and I might just have been making up that phrase when I first used it in 2003 and the internet then had no reference to it. But anyway, that's what it is and that's what it does. It's passive, in that it doesn't have any active components and requires no power. It attenuates the signal or takes down its level. And it's a device - a device. You can get one built into a cylinder with a female XLR connector at one end, which is the input, and a male XLR at the other, which is the output. Just handy for connecting between the microphone cable and pre-amp input. But the Neumann U87, like most other quality capacitor microphones, has one built in ready at the flick of a switch. So where the U87 can handle 117 dB SPL normally with the ten dB pad, it can handle127 dB SPL again for 0.5% distortion. This should be enough for most purposes. Gain Staging First Step So here's where you can start gain staging. You have a singer in your studio singing into the highlight of your microphone collection, your Neumann U87. The singer is belting out like Ethel Merman on steroids. Listen carefully. Can you hear any distortion, a fizzy, crackly sound that isn't part of the human voice? Yes? Then switch in the pad. Crystal clarity should be what you hear. If you can hear a difference, then yes, you needed that pad. With experience you'll hear easily when you need to switch the pad in. But you might say, wouldn't it be easier to leave the pad switched in all the time? Well yes, it would be. But this is where we get to the heart of gain staging. The main purpose of gain staging is to avoid distortion or clipping and at the same time optimise signal to noise ratio. When you switch in the pad on a microphone, you're lowering the signal level going into the internal amplifier. All amplifiers produce a certain amount of noise, which will stay the same. So your signal level is closer to the noise level, giving you a reduced signal to noise ratio. If you need the pad this won't matter. The signal level, even through the pad, is high compared to the noise. If you don't need the pad, you risk hearing more noise. Pad out, pad in. You're the engineer and you're in charge. The quality of the end product is down to you. There is, though, something I need to add here. I don't have a Neumann U87 to hand, although I've used plenty in the past. I do though have an AKG C4114, which is the U87s nemesis in its EB version, so it's quite old. Here it is with the pad switched out, I'm as close to the mic as I can get. Wakey, wakey! It really is difficult to hear any distortion, maybe just a moment on the A of the first wakey, but the waveform doesn't look at all clipped, and it could just have been my voice cracking. With the -10 dB pad switched in and the level corrected - wakey, wakey! I really can't go any louder than that. And as a why not, let's hear the sure SM57. Wakey, wakey! I think we can say that other than that operatic soprano I mentioned earlier and that trumpet recorded at one inch, you probably won't need to use the pad very often. If you're going to use your expensive capacitor microphones as close mics on drums, then the pad will be your friend. The Pre-amp So we've passed the first stage of gain staging. Let's move on to the second stage of gain staging, the pre-amp. Audio electronics - digital audio electronics. The people who make all these lovely devices and software for us are experts, absolute experts. Experts so expert that they don't feel the need to shout about how expert they are. They just get on and design and build incredible equipment and software for us. And as well as all that electronic and digital expertise, they know something else. That their products will work better and more reliably if they make it as nearly impossible as possible for nothing to go wrong. This is why I said in episode one of this podcast series that gain staging can be simple. It has all been made simple for us and there's hardly anything that will go so wrong that it'll affect the quality of your work. If you don't clip when you record and don't clip when you bounce, then that's almost everything you have to do. Oh, and pay attention to the level you're putting into plug-ins, because that can be a factor too and I'll consider that in a future episode in this series. So let me first consider the situation where you have an audio interface with an integrated microphone pre-amplifier. You plug your mic directly into the interface and your interface connects to your DAW via USB, thunderbolt or maybe even FireWire, if you're old school. It goes like this - pre-amplifier, gain control, then stuff, then the metre in the track of your DAW. Think of your pre-amplifier gain control being linked directly to the meter in the track you're recording in your DAW. I am by the way, thinking about a pre-amplifier with just a gain control, not a gain control and an output level control. I'll come on to that in a future episode. One more thing, your pre-amp, or audio interface, like the microphone, may have a pad. If you can't set the gain low enough to avoid the red light on the meter, then you can switch this in or switch in the pad on your microphone, if it has one. I prefer to pad the mic first and then it will only rarely be necessary to pad the pre-amp. Now this is all you have to do. Get your sound source going, give them a cup of tea first if you have to, raise the gain on your pre-amp, watch the meter in your DAW and stop raising the gain when it gets close to the top. That's all you have to do, it really is. Nothing has gone wrong. Nothing will go wrong. And this part of your gain staging is done. Of course, there has to be more nuance than that and there is - headroom. Headroom I'll try an analogy. I'm 6ft 2" tall. Suppose I walk into an old English pub with wooden beams and the floor to the lowest point of the ceiling is 6ft. I have no headroom. In audio terms, I'm going to clip and in real world terms, I'm going to clip my head. OK, I don't like that pub, so I'll try a different one. The floor to ceiling distance is 6ft 2". I fit exactly - I should never bump my head. But suddenly I find a spring in my step. I extend upwards, higher than expected and ouch. Although in theory there was enough height, in practise there wasn't. What I need is literally headroom. So I take my business to WeatherSpoons, where I should probably have gone in the first place. Other pub chains are available. The floor to ceiling height in this particular 'Spoons is 10ft. And there's no possibility that I will ever bump my head. I can't even jump that high. I have enough headroom. What if the floor to ceiling is 20ft? I have more than enough headroom. And now it's time to get back to audio. So your singer, well lubricated, with a tea-like infusion of some kind, is ready to do a soundcheck.They start singing and you start raising the gain control. You could, if you wanted, raise it so far that the meter in your DAW peaks at -1 dB FS, or -0.1 dB FS, if you like and that will give you a perfect recording. No clips, no distortion, no problems. But the thing is that it's a well known fact in audio that however loud a performer is during the soundcheck, they're always louder in performance. So if you didn't leave some headroom at the top of the meter, then when you go for a take, you'll clip. You'll have to lower the gain and ask your performer to go again, which is something you really should never have to do for any purely technical reason. So instead what you do is allow for this and set your gain lower so that there are some unused segments at the top of the meter scale, more unused segments than you could ever possibly need. You'll never need to ask your performer to go again for the reason of setting the gain too high. At this point, the natural question would be how much headroom to leave. Back to VU. If you remember from episode one of this podcast series, I talked about 0 VU and its relationship to -18 dB FS on your DAWs meter. We don't need a VU meter yet and that's something I'll come onto in the next episode but -18 dB FS on your DAWs meter - now this is something worth considering. Suppose you were a broadcasting corporation, in Britain perhaps, perhaps a British Broadcasting corporation and you had spoken word contributions coming in from probably hundreds of contributors of all levels of technical expertise. And when technical expertise is at a minimum, then its importance often isn't realised. What you don't want to run things efficiently is content coming in that's clipped because the gain settings have been too high. So you advise settings that are never ever or at least hardly ever going to go wrong. And even with massive caution over levels, there won't be any problems with noise. Here's the advice - set the gain so the meter bounces around -18 dB FS peaking up to around -10 dB FS. With these settings, nothing will go wrong. OK not never nothing, if that's acceptable English, but very, very rarely. And these are good settings for us to use too to record spoken word as I am now, or music. Now, it doesn't have to be exact, you don't need to get finicky over this. Think of all those non-technical people setting their gain controls - how finicky are they going to be? Bouncing around -18 dB FS mostly you don't even have to bother where the peaks are, this will get the job done. Audio Examples Now, time for some examples. Like a modern day goldilocks - too much, too little and just right. Since this is a podcast, well, you knew that already, but since this is a podcast, levels have to be set a certain way, which again I can go into in future, but for now I have to be clear that I'll need to adjust my finished levels to match the technical requirements of the podcast. But all of my examples will be in proportion with each other. I'll start with what can go wrong - clipping. What I have is some audio I recorded in Abbey Road Studio 2, so I think we can accept that the technical quality is good. I'm going to take this original clean recording and pipe it through my pre-amp at line level rather than mic level but the results will be similar. Line level - that's another gain staging thing for me to cover later. So I'm setting the level going into my DAW using the pre amp's gain control. First let's hear what happens when my DAWs meter bounces around -18 with the peaks going up to -10. I wandered lonely as a cloud that floats on high o'er dale and hill when all at once I saw a crowd, a host of golden daffodils, beside the lake beneath the trees, fluttering and dancing in the breeze. Nice and clean, but what if I push it a bit and peak at -1? I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. It's still clean, this is what we expect, but for an original recording 1 dB is far too little headroom for any kind of comfort. I suppose I should do this, it goes against all my instincts but I'm going to push the gain into clipping. This will be unpleasant. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees fluttering and dancing in the breeze. That was unpleasant. So we've explored the Goldilock zone and the too much zone. Let's try the too little zone. Here's a recording peaking at -20 dB FS. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. And at -30, dB FS. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. And finally at -40 dB FS. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. 40 decibels of headroom is massively too much for speech. It might not be for sound effects recorded in the field but that's another story. As we can hear the level got quieter and quieter, unusably so, but if you had recorded, say, an interview at these low levels, you'd have to boost them up again. So I'll do that I'll normalise them to peak at -1 dB FS the same as my example with almost no headroom. We don't need headroom here, so this is okay. I'll play the -20 version followed by the -30 then the -40 all normalised to the same level. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. I wandered lonely as a cloud that floats on high o'er dale and hill, when all at once I saw a crowd, a host of golden daffodils, beside the lake, beneath the trees, fluttering and dancing in the breeze. Noise - we hear noise, even in the -40 example. It isn't the most noise I've ever heard and in fact you might not even notice it played on speakers, but it's noise nonetheless and professionally we don't like it. So this is why we can't go too low with the gain and therefore the recorded level. Even with 24 bit digital audio with its amazing signal-to-noise ratio and dynamic range if we record too low, then the end result will be a noisy recording. Ending So that's it for now. You know all you need to know to make a great gain stage recording with nothing to go wrong. You can record your whole 100 track magnum opus like this and everything will be perfect. In the next episode, I'll be looking at gain staging in the mix and I will be bringing back the VU meter. I'm David Mellor, course Director of Audio Masterclass. I'll see you in the next episode. Thank-you for listening and be sure to check out the Show Notes page for this episode where you'll find further information along with web links and details of all the other episodes. Oh and just before you go, let me point you to the soundonsound.com/podcast website page where you can explore what's playing on our other channels.