Articles / Feedback Or No Input Mixing
Feedback (No-input Mixing)
by Muied Lumens
If you have ever pondered upon the mathematics behind those beautiful patterns that make up fractal images you have marvelled at the raw beauty of feedback. Many of those images are made up of very simple mathematical expressions, where the result of that expression is fed back in to itself (or iterated), over and over. Feedback is a phenonemon that exists all around us in various shapes and forms, and by experimenting with this in audio you will maybe gain some insight into it on a new level. Once you start really playing with feedback it almost begins to feel like a living entity. It is something that speaks to your intuitive side more than your intellect, because even with fairly simple setups, the behaviour of the fed back sound goes beyond what you can easily understand, and it simply boils down to how you interact with it in the moment.
But hold on...
First, a word or two of warning. Taking a signal and amplifying it, then reamplifying it again and again and again ad infinitum results in a very loud signal. It will almost immediately hit the ceiling of what your equipment is capable of reproducing (unless you delay it, more on that later). It is therefore essential that you protect yourself and your equipment from damage. Here are a few guidelines, but use your sense at all times...
There is of course no guarantee that your equipment is safe from harm by feeding it levels at the very top (and beyond) of what is was designed to be capable of, so don't use your most prised $10 000 mic preamps, OK? In general, gear should be able to deal with it, even guitar pedals. In fact, you can use lower gain guitar pedals as "buffers" or limiters for your line level equipment because pedals often deal with lower levels. The big risk is to the speakers (and mikes) and your hearing, so make sure you have them kept safe. Nobody but you will be held responsible if things go wrong though, so again use your best sense.
An additional note on using modular synthesizers with guitar pedals: these two often operate at very different levels, and while I must admit that I don't know if there is any danger of damage, please ensure this for yourself before you proceed.
Setting up your first session
Making a simple feedback loop is very easy. You take an amplifier and connect its output back to its input, and turn it up. At some point, the amplifier turns into an oscillator. You can't hear what is going on unless you split the output and that is where a mixer comes in. If you take the main output of a mixer and plug it into one of the channels you have the same thing happening, just now you can also route an other output to a speaker.
Mixers are very handy in feedback configurations because you can route and split signals and amplify them too. If you have a mixer with an Aux (Auxillary) out, and route that back into one of the channels using a cable, you can control the overall volume of the main output and whack up the channel and Aux gain to create feedback. In other words, the volume on the channel will be extremely loud, so you need to almost completely turn down the main fader to compensate for this, and keep a cautious finger on it while you increase the gain of your feedback. Make sure you understand this gain structure before you start, it's at the very foundations of feedback performance.
If your mixer has meters or overload indicators, they will come in handy as they can show you if there is a signal. When you have a stable tone going, experiment with volume changes and other parts of your channel, like the EQ, polarity invert, gain, etc. You will quickly get an idea of how things work. It is very hard to reproduce exact results over and over, so I recommend that you have something standing by to record with.
Here, the Aux out is connected back into channel one. The Aux on this mixer is pre-fader, meaning that the channel volume does not affect the Aux level. It is therefore easy to set up a loop and control its volume on the channel itself. The headphone output goes to some old computer speakers and the record out to my handy portable recorder. The sound this setup made can be heard here (dry recording, I only filtered out some subharmonics):
Once you have explored one channel, set up an other one, using an other aux out if possible. The extra channel in isolation will act as an other oscillator, and you can create some nice detuned drones this way, but interesting things happen once they start interfering with each other. One can act as an LFO on the other, you can get strange intermodulation going, or other unheard of squeals, unique to your equipment and setup. If this blows your mind like it did mine when I first tried it, you will slowly increase the complexity of your setup, adding external effects and processing until you have no idea what's going on any more! How you proceed is up to you, but there are no rules except for the safety ones. Always be aware of where your main volume output is, both in level and physical space.
Using the room
Adding a mic gives you literally a whole new set of dimensions to the feedback. It also adds some problems you should be aware of. When a mic comes into the picture, the unreliability of the system increases many fold, and your alertness should too. There are two simple ways of making sure that you are in control, more or less, and you should make use of both of them. They are the limiter and the delay.
The limiter is a safety precaution, and should be between your mixer main outs and your speakers, to protect from overloads that are too fast for you to react to. Limiters are more important in mic configs because of the added instability mikes add. When you feed a channel directly back into itself (i.e. internally in a mixer), it will be overloaded to its maximum possible signal level, and distort. This in itself is a limiter, and older cheaper gear with less headroom will do you a favour here. You do not want to push microphones to this level because it is normally far too loud for comfort and it could break your speakers, ears and mikes in one fell swoop. Be aware that limiters are actually compressors with a high compression ratio, so if you get excessive signal peaks combined with odd DC offsets and what have you, they may still be capable of breaking through the limiting ceiling.
Delays will slow down the development of the feedback, if you have them set up correctly. This can make it easier to "play" the feedback, because it gives you more time to react to changes, provided the delay time is long enough. You should start with a 100% wet signal, from the mic, through the delay (with no internal feedback) and out of the speakers. No direct signal from the mic should reach the speakers. The delay time is up to you, but 500 ms is probably a good starting point. Now take the mic in your hand and move it around the room (but be within reach or the main fader). If everything is set up properly and you have enough gain, you should be able to play the room! Different spots will give you different sounds or tones. I find that adding distortion makes it easier to achieve this at lower volumes, but your milage may vary. Also adding resonators, like placing the mic inside a bowl or tube will accentuate this effect. You may have to go quite loud to get it working properly, so make sure your neighbours are sane, sensible people.
If you do not have a microphone handy, you can use other types of transducers, like headphones, speakers, piezo transducers and so on, in place of a mic. When using something that is not as sensitive as a mic normally is, apart from adding lots of gain, it helps to compress or distort the input to force some decent levels and add harmonics. You can of course also use electric guitars or other instruments, just laid flat on a table. Placing is all important, so slowly add one thing after an other, and move things around as you do. The more complexity you add the better, as long as you are in control of the main output.
Once you have set up a few loops that interact with each other - exploiting the nonlinearities that occur at the limits of the circuits involved and/or the acoustics of your room, instruments and even involving your own feedback, responding to the sounds that you hear - you have arguably created what is called a chaotic (or complex) system. This does not mean that it is completely random, but simply that its long term output is unpredictable, just as the weather and the stock market is.
If you are at all familiar with how chaotic systems work, you will know that small changes in input can result in a huge difference in output. You should be able to see this in action when you change a single gain stage in a multiple loop feedback setup for example. You may end up with a situation where you have a relatively large range which makes no difference and suddenly at a specific "hot spot" it will take off in a new direction. It is often around these hotspots that you can find interesting variations to play with, and even where the audio can take on that strange, alien quality and start changing without you interfering at all. As comes with the territory, these situations are very difficult to plan ahead for or repeat, but if you keep at it eventually magic will start happening for you.
Whether you are doing it for the sake of raw sampling material (you can get a wealth of great single cycle waveforms from mixer feedback - take a look at the waveforms!) or for a performance, the quality of the equimpent is almost unimportant. I say almost because as mentioned before, budget or old cheap gear is just as useful than modern, clean sounding hifi recording equipment. Lofi stuff will add more tone and character to the sound and hopefully limit the bandwidth too, as the evolving sound seeks the path of least resistance. That said, nothing stops you from using whatever you like, including elements that have nothing to do with feedback whatsoever. It is your playing field, conquer it!
So far I have assumed that the feedback has been created more or less in the analogue domain, but you can of course make software based feedback too. The rules do change a little bit though, and unless you put some work in, much of the unpredictability and charm of analogue nonlinearity gets lost. If you distort a digital signal you get a completely different result, and not one that may be immediately useful. I therefore usually stay within reasonable levels in the digital domain, and add other effects to bring out the nonlinearities. You could say that the same rules that go for acoustics apply here, which is to stay within the dynamic range. You may need a trigger to get things started, which can be anything that makes sound, anything from a short impulse to noise to a snippet from the Sound Of Music.
Not all software will let you create feedback loops, in fact, some software will prevent you from this altogether. Reaper, which is a DAW recording program, has an option that lets you switch off this safety measure (File > Project Settings > Advanced). Most modular software, like Max, Puredata, Supercollider, Plogue Bidule and Kyma will let you create feedback loops, but be aware of a particular difference between software and analog feedback systems. The software has to actually process the data it produces so if you create very short loops it will cost you in processing power, and can cause the program to behave strangely or simply crash. One simple way around this is to create an acoustic feedback path, as described above. The latency of the system will do you a huge favour here.
The great benefit of software is that you have access to a much larger pallette of feedback techniques. You do not have to limit yourself to straight audio paths, but can create FM feedback, intricate control dependent delay networks, noodles and all sorts of things that I can't even begin to think of. A software modular is in its own way a fun way to explore, and is discussably only outmatched in fun factor by a real modular, which brings us to...
The ideal feedback mixer in my opinion is a Matrix mixer, one which has as many outputs as inputs. Any mixer that has several Aux and Buss outs is great though. If you team this up with various effects and processing and take some time to explore how everything interacts, you have a very powerful modular setup that can make a wide variety of sounds. Having a couple of foot pedals to control volume could give you extra expressiveness for example, but in the end it all comes down to how you perform on your system, and that is not something you can learn from a wiki article. For inspiration, listen to Toshimaru Nakamura who coined the phrase "no-input mixing". Christian Carriere also did a piece by Arvo Pärt that you can listen to and watch here:
Other less virtuoso techniques (in performance terms) would involve using a sequencer to control the gain (and therefore the pitch) of loops for example. You could use an analogue sequencer for this, or a MIDI sequencer controlling CC values of various parameters of a synth or effects unit. Any synth that has inputs could become a target for feedback abuse! Other synths, for example Dave Smith's Evolver, has internal feedback paths that can create interesting sounds. Also making use of other not-so-obvious effects processors, like multi-band compressors, or even feedback eliminators will further expand your sonic territory.
Hopefully this has given you an idea of how much scope for exploration there is in the humble feedback loop. If you want to understand some of the processes behind this magic, I recommend that you read up on Chaos theory. If you like books you could do worse than read James Gleick's "Chaos" as an introduction to the theme. There are countless other books on the concept of feedback in areas from physics to biology and for a very different approach, try "I am a strange loop" by Douglas Hofstadter.
There are several threads on the forum, like this one, and this one discussing the theme, and especially this post which i recommend reading. Acoustic Interloper also has a wonderful recording from the 2012 electro-music festival posted in this thread:
Zero-input mixing performance Copyright (c) 2012 Bill Manganaro, John Driscoll, Jeremy Parson and Dale Parson on Zero-Input Mixer Collaboration
Released under the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.5)