the electronics behind the noises
Andy Honeybone looks at the birth of effect electronics
It was a good many years ago when yours truly pushed a handful of components into a sheet of corrugated cardboard and wired up his first fuzzbox from an 'ingenuity unlimited' design in another magazine 'Home Stuffed Rodent'.
Resplendent in its metal Oxo tin and nuke-proof ex-GPO jack sockets, the unit was the envy of my friend because any note plucked on a connected Watkins Rapier was immediately transformed into a pukka Jimi Hendrix lick. Simple but brilliant, the tone was mellow and sustained until, with a hiccup, the note stopped short in a most unmusical way.
This was in fact a design feature because the circuit was based on a Schmitt trigger. This marvel-of-what-you-can-do-with-two-transistors is an electronic switch which converts an incoming 'nice' signal into a rectangular waveform. It exhibits hysteresis (you can get pills for it) which means that once the input level is high enough to trigger the switch, the switch will not turn off until the input level has dropped to a value some way below the turn-on threshold. This cleans the signal but also removes any envelope information. When the guitar note has decayed to the point where the switch will no longer trigger — bingo, the note stops dead.
A fuzz box is thus a unit which takes a signal and squares it up giving a harmonically rich output with altered decay characteristics. The early Marshall design was considered to be 'the King' and sported a couple of OC76 germanium (not geranium) transistors in a high-ish gain configuration. This simple circuit works on the principle that if the gain of the fuzz box amplifier is, say, one thousand times, and the input signal is 40 millivolts, the output should be 1000 x 0.04 = 40 volts.
As the unit is powered by a 9V battery, this range is impossible and so the signal is clipped top and bottom for the period while the output signal is 'outside' the 9V range. The flat edges make the waveform square and harmonically rich, and as the amplitude (volume) remains constant while clippling is occurring, longer sustain is also introduced. From this scheme you should be able to see why the guitar volume control needs to be full up to get the most effect from a fuzzbox. What goes on inside a fuzzbox is a low power version of the output stage of an overdriven instrument amplifier.
With the advent of integrated circuits, a new level of unnecessary sophistication was available and the op-amp with inverse paralleled diodes in the feedback loop was all the rage. This configuration was supposed to be capable of a greater intensity of fuzz than previous designs but tended to be rather uncontrollable and produced hum, whistles and burps after the breakup of a sustained note. In an attempt to emulate valve distortion, field effect transistor (FET) amplifiers have been used in the form of CMOS logic inverter stages biased for linear operation. This arrangement does seem to give a more gentle onset of distortion but valve sound synthesis is a very subjective area.
Fuzzboxes aren't only for the guitarists of this world as Keith Emerson was quick to find back in the days of the Nice. Playing minor seconds (two notes one semitone apart) on Hammond organ plugged through a fuzzbox gives a superb tugboat effect and holding a bass note and slightly pulling out a higher pitched drawbar gives a good circular saw.
Moving on some, it's time to delve within the diecast boxes that flange and chorus. The fable goes that flanging was originally produced by playing identical tapes on two tape recorders in tandem, mixing the outputs and slowing one by pressing on the 'flange' (the outside rim) of the supply reel. True or false, it does illustrate that the effect is achieved by adding a varying delay to the signal and mixing it with the direct original to cause certain frequencies to be cancelled. The effect is that of a comb filter and, unlike phasing, the notches (cancelled frequencies) are harmonically related — odds cancel, evens reinforce. Chorus simulates multi-voices and differs only in requiring longer delay times.
The electronics of the delay line divide neatly into analogue bucket brigades and digital memory clocking techniques. The first is cheapest and therefore a good place to start. The name bucket brigade stems from a water analogy of the line's operation which looks at the delay chip as a row of buckets passed from hand to hand in the old time fire fighting tradition. The input is sampled and a value corresponding to that sample is stored in the first bucket. An intermediate cycle then 'tips' the contents of the first bucket into the next and then fills the now empty bucket with another sample. The pattern repeats with new samples arriving at the beginning of the chain and previous samples advancing their way to the last bucket which is the output. The speed of sampling and hence delay time is under control of a clock oscillator running in the range 10-500kHz and a Low Frequency Oscillator modulates this frequency to provide movement within the effect.
As with any sampling system, the two major gremlins are prevention of aliasing and elimination of the clock frequency from the output signal. The first is a foldover problem where the input signal contains frequencies higher than the sampling rate and the difference is reflected into the signal as twitterings and distortion. The effect is analogous to watching a film where the wheels of an accelerating stage coach appear to start to spin backwards. A movie camera is a sampling system and aliasing is observed when the frequency of the wheels exceeds the number of frames per second — the sampling rate. In the case of a flanger or chorus unit, a filter at the beginning of the delay line will prevent this occurence. Filtering the output is also necessary to remove the clock signal which, although often out of the audio range, might saturate following circuitry or give trouble by beating with tape recorder bias frequencies causing unwanted whistles.
With each sample being slopped from bucket to bucket, a fair amount of noise is to be expected from an analogue delay line and the most common way of achieving a respectable signal-to-noise ratio is to build in a mini dbx. If the dynamic range is compressed before the delay line and expanded afterwards, the overall dynamics are unaltered but noise from the line is expanded downwards out of hearing.
The delay can be achieved by digital methods but as musical signals are analogue quantities they have to be converted to and from digital at the start and finish of the works. The input is sampled and converted to a number which is stored at a memory location. Step two sees that number shifted to the next location and a fresh sample pushed into the first. The process repeats with each number moving one step along the line for each new sample. Numbers reaching the end of the line are converted into voltages having been suitably delayed by their circuitous passage. It's exactly the same as the bucket brigade except it's numbers being pushed around instead of a charge. The digital advantage is that however long the line may be, the signal emerges at the far end exhibiting exactly the value with which it started — no degradation and no noise (well, only quantisation distortion).
Compressors were mentioned in passing and a closer look is worthwhile (so it says here). Firstly, the term refers to the contraction of dynamic range — making quiet things louder and noisy things softer. The worst example of a compressor in action is the automatic level device on a cheap cassette recorder. The best examples are completely transparent to the ear but help to create slinky bass lines, larger than life bass drums and controlled vocals. Compressor circuitry has a Voltage Controlled Amplifier at its heart and the control voltage is derived from the output of the VCA itself. This feedback loop constantly adjusts the gain of the VCA to try to keep the output constant. If the input to output signal ratios are constant over a wide dynamic range the unit is a compressor but if only input signals above a certain threshold are reduced, the unit is a limiter.
The difference is brought about by doctoring the feedback voltage from the output signal envelope follower. Varying the amount of feedback alters the slope (efficiency) of a compressor while the use of non-linear circuitry gives a limiting response. If frequency selective circuitry is substituted, the same basic design can be used to limit high frequencies in the input signal such as sibilance (de-essing).
The attack and release times of a compressor are often made variable and require careful trimming to avoid 'breathing' effects where you can hear the noise level increase as the gain rises during a momentary pause. The best compressors have a degree of 'intelligence' which adapts the time constants to suit the input material and prevents the horror of full gain with no input signal by incorporating some form of noise gate.
Next month — inside the Cyclotron.
Feature by Andy Honeybone
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!