We continue our beginner's guide to synth programming with an explanation of what an LFO is and what it does.

So. Back to Basics has now helped demystify the theory that lies behind how an analogue synthesiser generates its pitch, and how it's possible to manipulate tone and amplitude shapes using the component parts that make up a synth's processing circuitry. But if you've been following the series with one hand on the magazine and the other on the keyboard, you'll have realised that most of the sound possibilities we've discussed so far have been a little on the static side.

Now, this wouldn't be a problem if it weren't for the fact that many acoustic sounds rely heavily on pitch and tonal fluctuations for their sonic interest. They're the components that do the most to animate acoustic sounds, and without them, synthesised voices aren't going to stand much of a chance in their attempts to appeal to the human ear. Which is why we're dedicating this month's instalment to the job of creating precisely those variations.

The synth module that's relied on most to perform these tasks is the Low Frequency Oscillator, a humble-looking device that's fitted to almost all modern-day synths, analogue or digital. Electronically, its purpose is to provide cyclic modulation effects of one form or another by generating rising and falling voltages. And as a glance at Figure 1 will tell you, the LFO is a free-standing device that does nothing more than generate various waveforms at subsonic frequencies that normally fall in the range between 0.1Hz (one cycle every 10 seconds) and 30Hz (30 cycles per second), though this can vary widely depending on the synth's manufacturing origins. Whatever its range, though, the LFO's function is the same - to provide modulation. Which is why Korg call their LFOs 'Modulation Generators'. Simple, really.

The Wave Factor

Although LFO waveforms look much the same as those generated by VCOs, they are in fact crucially different in that they're simply graphic representations of the way a voltage rises and falls. It's easiest to think of them as voltage 'shapes' similar to those that appear at an Envelope Generator's output, the only difference between the two being that whereas the EG's is a one-shot voltage that only exists on receipt of an incoming pulse, the LFO's voltage is a repetitive one.

The various common LFO waveshapes each have their own specific uses. Triangles are good for smoothly undulating sweeps of pitch or tone, Square waves are useful for providing automatic octave jumps and sequencer mimicking effects (as are the subjectively smoother sinewaves that are often found in their stead), while sawtooths, in either of their rising and falling variations (some synths have both), tend to be used for effects that can be most diplomatically described as 'off the wall'.

However, getting hold of the effect you're after isn't just a question of finding a synth that has an LFO with the right waveform and then bringing it into play. Because the waveform's speed and modulation level have just as big a part to play as the waveshape itself. Vibrato effects, for example, require a sine or triangle wave set at around 5-8Hz and a fairly low modulation level. Too low a speed and you'll end up with a siren, too high a mod level and your ears will be assaulted with something not dissimilar to the sound of a Star Wars zap gun.

In fact, the inherent flexibility of the LFO makes it well-nigh impossible to give a comprehensive breakdown of the settings that suit particular sets of sounds, so experience will make you the best judge of what's needed for a specific programming job.

The Age Factor

Classifying LFOs isn't that straightforward, however, because their waveforms behave differently depending on who designed the synth in question and when it was manufactured. The heart of the matter is the module's voltage output and how it alters with time, as we'll see.

Most of the voltages generated within a synthesiser start at a value of zero and move upwards in time to a positive value. The EG is a fine example of this, as we saw last month. But LFOs are the exception, in that some of their waveforms' voltages 'rotate' around 0V, going negative for one half of the cycle and positive for the other.

As Figure 2 should indicate, the sine and triangle waveshapes do just that. Now, whether or not manufacturers ever designed their LFOs to work this way intentionally is something of a moot point, but the principle has some foundation in the field of acoustic instruments. If, for instance, you were to look at a violinist's vibrato technique, you'd see it comprises revolving the pitch around the note being played at the time, an action replicated extremely well by a sine or triangle LFO wave being applied to the voltage input of a VCO. Some acoustic instruments aren't capable of true pitch vibrato, but can be used to create tremolo (a cyclic variation in tone and/or amplitude) instead. And again, this effect can be replicated by a sine or triangle LFO waveform. How? Well, applying a bipolar voltage to the CV input of a VCF and/or VCA is an action very similar in principle to, say, tone and amplitude variations on a flute revolving around an average level.

But if sine and triangle LFO waveforms are so good at recreating vibrato and tremolo effects, what do the alternative shapes do, and how? Well, the answer to this one is also complicated by the influence of history. In the past, manufacturers saw fit to make LFO square wave outputs jump between 0V and a positive value, creating a waveform we call, for convenience, 'unipolar'. As I've said, such a waveform is useful for creating note-jumps of various musical intervals, but what's important here is that the pitch of the note you play remains true at all times, with the interval jump tracking that note perfectly. This is a direct result of the square wave cycle's lower half being at zero volts, because (not unexpectedly) adding 0V to the voltage mixer of a synth's VCO makes no difference to its output, while adding a positive value during the upper half of the cycle results in the required interval jump.

But like I say, that was in the past. More recently, synth manufacturers have started incorporating bipolar square waves into their LFOs instead; a typical example of the new breed is shown in graphic form in Figure 3. Frankly, I can't help but wonder exactly why this new format has been implemented so widely, as it's not nearly as useful from a musical point of view. Cost has probably been the critical factor, I guess.

Why aren't the bipolars useful? Well, the crux of the matter is that if you modulate a VCO with one, the lower level of the interval becomes lower than the note you're actually playing on the keyboard. Which isn't, all in all, a particularly useful (or indeed musical) effect to have at your disposal.

Much the same argument applies to sawtooth waveshapes as well, even if the sounds they produce aren't necessarily affected by waveform polarity. Go for a unipolar variant if you can find one, but recent design trends might make the task difficult to accomplish.

The Option Factor

So much for LFO basics. One point worth bearing in mind is that synth manufacturers have seen fit to incorporate a number of extra facilities into their designs as the years have gone by, and one of the most common is the trigger sync facility. What this option involves is routing the keyboard's trigger or gate output to a reset input on the LFO, so that every time you play a new note on the keyboard, you automatically reset the LFO waveform to the front edge of its cycle (see Figure 4). And although this is a facility of somewhat questionable usefulness, it does at least mean that square wave octave jumps (say) will always be in time with your playing.

There are problems with trigger syncing, though. Let's say you wanted to set up a slow, undulating filter sweep using a sine or triangle wave. With trigger syncing, each new note played would reset the LFO, and the end result would be more akin to an EG sweep rather than anything else - not a happy state of affairs. Unfortunately, there are gaps in the Howell Knowledge of Synthesiser Specifications, and I can't say off-hand which models feature LFO trigger syncing arid which don't. Try before you buy.

A further option, and one you'll find on almost all of today's polysynths, is delayed modulation. Basically, this allows you to execute a gradual fade-in of vibrato automatically, with the Delay Time control being used for precise adjustment of that parameter. Figure 5(a) sheds the light. You'll see that the LFO's output is routed to a VCA, whose output level is controlled by an EG. The EG's attack time is effectively the delay time, because the VCA's output level increases as the EG voltage increases, and Hey Presto, the LFO's waveform is gradually faded into whichever bit of the synth's internals it's been connected to.

But again, things aren't problem-free. The release time of the EG in question is always set to minimum, which means that the vibrato disappears as soon as you take your fingers off the keyboard; something that can be a bit disconcerting if your envelope-shaping EGs are set fairly long.

However, a variation on the delayed vibrato theme has been developed in an attempt to overcome just this problem. Figure 5(b) shows the layout of a typical 'present-day' delayed modulation patch. In it, an EG is routed through an inverter that turns its voltage upside down, so that when a note is pressed, the output shoots down from 0V to a negative voltage, and the decay/release envelope portion starts to enter the proceedings. With the attack time in this example set to instant, it's the decay/release time that sets the vibrato delay. The decay and release portions are tied together so that if you keep your fingers down on the keyboard, the modulation fades in. Likewise, if you just stab at the keyboard using a sound with a long release time, you'll still hear the modulation increasing even as the sound fades away.

Obviously, the latter system is the more flexible, and you shouldn't experience too many difficulties tracking down a synth that's equipped with it.

It's also worth noting that a lot of models equipped with a delayed modulation facility also have a Delay Reset option, so that each new note played retriggers the LFO's inverted EG. This makes good sense on a monophonic synthesiser, but causes the odd hiccup or two when applied to polysynths. Why? Because it means that every new note you play over a chord starts the delay cycle all over again, and as a result, the chord you're holding is robbed of its modulation as soon as a new note arrives. The alternative is to give the new note instantaneous - rather than delayed - modulation. Unfortunately, on all but the most expensive polysynths that have a separate LFO for every voice, there exists only an either/or situation.

The Touch Factor

Nowadays of course, you can introduce modulation manually by applying extra keyboard pressure, in addition to implementing it automatically through electronics.

The theory behind this is actually quite simple. The LFO's output is again fed to its destination via a VCA, the output level of which is controlled by a second voltage derived from pressure sensors mounted on the keys. The harder you press the keyboard, the higher that secondary voltage becomes, and the greater the proportion of the LFOs output is passed through the VCA. And as if by magic, more modulation makes itself apparent.

Still, a simple theory doesn't always mean an equally straightforward execution, and the circuitry for this particular function is a bit on the complex side, not to mention expensive too. Thankfully, recent technological advances have made velocity-sensitivity easier and cheaper to implement, and I can't see it being too long before it becomes de rigeur for any decent, self-respecting polysynth. And that's got to be good news for musicians.

Moving on to a slightly more obscure variation (obscure in that it's confined to modular synths and the odd overdraft-inducing poly), we come to LFOs that can be modulated by an external control source. The golden rule with these is that the higher the incoming voltage, the greater the LFO speed will be. It's good for special effects but not much else.

Then again, while the VCLFO may not be the most commonly found synthetic device, a larger number of instruments do feature a VCO that can be adjusted to perform the function of LFO, which can then, of course, be voltage-controlled. Figure 6 shows some of the control voltage waveforms that can be created using just such a technique, which should in turn give you some idea of the sort of effects possible.

Lastly, we come to the vexed question of exactly how many LFOs are actually desirable on a modern synthesiser. It's pretty clear to me that a single-LFO setup isn't going to give you much in the way of inherent versatility. If you want a slow, sinister filter sweep, you have to forego that nifty bit of fast vibrato you've just set up - which is a bit of a shame, all things considered.

Luckily, dual LFOs are now the rule rather than the exception, though in a number of cases, the second oscillator is nothing more than a triangle or sine wave generator. If you can stretch to it, the ultimate source of LFOs is probably Oberheim's Xpander, which has no fewer than 31 of the little beasts.

Well, that's about it for the Low Frequency Oscillator, though I for one didn't imagine such a simple device would take such a lot of explaining. The way I see it, no amount of journalistic meandering is going to tell you all you need to know about it, so I'll stress the point yet again that nothing beats hands-on experience. After all, plenty of acoustic instrument players spend years perfecting their modulation technique - so why shouldn't synth players?

I know it may seem easy just to add a bit of vibrato here and there, but there's a lot more to LFOs than that. Just compare the modulation styles of Jan Hammer and Vangelis to see (and hear) what I mean...