Our beginner's guide to analogue synths comes to the complex and thorny topic of modulation.

We've now seen how most individual bits of analogue synthesiser work. We've covered the VCO, VCF, EG and LFO, and what you've learned about those can be applied to just about any analogue synth, from a humble SH101 to a Prophet T8. So, now with the basic building blocks safely out of the way, we can go back to VCOs and look at some of their more complex functions.

A lot of what I'll be looking at will be of particular relevance to synths with two oscillators, though that's not to say you should stop reading if your synth has just the one. You're bound to come up against a pair of VCOs one day, after all.

Having said that, the first subject to come under discussion - Pulse Width Modulation - is by no means confined to dual-oscillator instruments. As we saw when we covered the VCO from a general viewpoint, many designs give you a square wave output whose width is continuously variable, this usually manifests itself in one of two ways. Either the selector that switches between the various available waveshapes has a variable pulse width control for the square wave, or there's a position on said switch that handles the pulse width as a separate entity from the square wave. In practice, both systems perform exactly the same synthetic task.

To recap on something I've already been into in a previous instalment, a square wave with a perfectly symmetrical shape creates a hollow sound not unlike that of a clarinet. As soon as you vary the width of the waveform, however, the sound becomes thinner and more nasal as a result of drastic changes in harmonic content. To be more specific, a symmetrical square wave creates only odd harmonics, whereas the same wave at one extreme of its pulse width contains a high number of both odd and even harmonics. The result is shown in Figure 1; analogue synth experts will realise that the harmonic structure is not unlike that of a sawtooth waveshape - except that there are even more harmonics. Not surprisingly, a good deal of variation in harmonic content exists between these two extremes of waveshape, and that's the parameter that a synth's pulse width control gives you.

But the real beauty of the pulse/square wave is that its width can be varied automatically, courtesy of your friend and mine, voltage control.

Automatic Control

Figures 2a and 2b show block diagrams of different sorts of PWM section. The first demonstrates Roland's method-whereby a slider has a switch that selects between LFO, EG and a DC voltage source - while the second shows the way most other manufacturers go about doing things. The latter incorporates that old favourite, the summing amplifier, in this case fed by a DC voltage source and a modulation source, governed by separate controls. The former, on the other hand, has just the one slider to regulate the level of voltage coming in, be it from a DC source (for setting pulse width manually) or a control source (for varying the width automatically).

It's pretty obvious that the system in Figure 2b is the more versatile, as it allows you to set the width of the pulse using the DC voltage source and then mix in voltage control as well. In other words, you can begin with a width that isn't symmetrical and sweep the pulse from that point, the maximum sweep amount being governed by the modulation level control. The difference is a subtle one, but it certainly makes the effect more controllable.

What does a modulated pulse width sound like? Well, that depends on the control source you're using. In both of the diagrams just discussed, the control sources are the sine and triangle output of an LFO or EG. There's a good reason for this, in that these two devices have now been adopted as the standard controllers by most of the modern synth industry, as a result of their producing the most musical effects. That's not to say a lot of fun can't be had from using other, less conventional sources (being the owner of a good ol' fashioned modular system, I can vouch for the usefulness of, say, routing a keyboard's CV to the PWM section so that pulse width can change across the keyboard's length), but as so often happens, sheer cost prevents most designers from implementing such electronic exotica.

But let's confine ourselves to the most commonly-found modulation sources. Selecting an LFO to do the job means you can sweep the pulse width between its two extremes at a rate determined by the LFO frequency. You can also decrease the amount of LFO modulation, if a somewhat subtler effect is what you're after.

There is a second effect associated with syncing, however, and it comes to the surface whenever you alter the frequency of the synced VCO. To demonstrate it to yourself, sync two VCOs together using a square wave output on each, set the filter so that the sound passes through unaffected (ie. with the cutoff frequency control at maximum), and make sure sustain portions of the EGs are up full. Now hold down a note on the keyboard and adjust the frequency control on the synced VCO (it's impossible to be specific here; most synths give the synced oscillator a Sync switch to identify it, some do things the other way around). As you make that adjustment, you should hear the characteristic heavy flanging sound.

Exaggerating the effect is easy: just boost the level of the synced VCO so that becomes the dominant tone. You could also try experimenting with different waveforms and combinations of waveforms - at various frequency settings.

A lot of sync effects are nicely suited to beefing-up lead lines, and devotees of the Jan Hammer school of heavy metal synth playing will already be familiar with the sort of things that are possible. You can also use them to create powerful bass sounds - especially if you're going to sequence them.

Voltage Control

Of course, any voltage-controlled oscillator also lets you create sync sweep effects automatically by using a source of varying voltage. Connecting a very slow LFO to only the synced VCO (let's call it VCO2 from now on, OK?) and setting the modulation level fairly high means that the effect I've just talked about will be heard as an automatic sweeping effect.

Alternatively, you could select some other modulation source for your sync sweeping, the most common of these sources being the EG. Most of today's synths let you route the VCF's EG to the VCO's, and by routing it only to VCO2, you allow sync sweeping to be governed by the setting of the synth's ADSR controls, and by the modulation amount. You may find you'll have to compromise between the filter's envelope shape and the VCO's in order to get exactly the effect you're looking for, but there's plenty of room for manoeuvre.

Touch-sensitive synths offer an even greater degree of control, as you can often route the voltage generated by the dynamic keyboard to VCO2, thereby putting the sync effect under (literally) fingertip control.

Most synths use sine or triangle LFO waveshapes for application to a PWM module, even if the LFO in question can generate all manner of outlandish waveforms. Figure 3 shows how this is achieved. You can see that, regardless of the waveform you select for pitch or filter modulation, the separate feed taken from the triangle or sine output of the LFO is always routed to the selector switch on the VCO's PWM section.

This back-and-forth modulation of pulse width is well-suited to the creation of chorus effects, and also comes in useful for thickening-up ensemble sounds such as strings. If your synth has just the one VCO, PWM can be an effective way of giving the impression you're using a couple of detuned VCOs. On the other hand, deep, fast PWM can sometimes give the impression of vibrato, as the human ear perceives the harmonic shift as a pitch one. And if you find this effect more of an irritation than anything else, you simply back off the modulation depth or LFO speed, and all will be well again.

In the past, synth manufacturers have seen fit to endow their instruments with just a single LFO, something that (obviously) restricted the possible applications of PWM because the section had to share the same rate as the vibrato (or whatever). Thankfully, today's designers have seen the light and given a lot of contemporary synths two or more LFOs. This really is a wizard wheeze, as it lets you set up different speeds for vibrato, filter sweep and PWM; the resulting range of possible sonic effects is, frankly, vast.

Dynamic Control

And so we come to the other most common form of PWM, namely that which uses an Envelope Generator for dynamic control. Here, the width of the pulse varies in accordance with the voltage output generated by the EG. As the EG voltage rises during its attack cycle, so the pulse width gets narrower. When the EG voltage falls during the decay cycle, the pulse width broadens out, eventually returning to its original width during the release portion, when the EG's output falls to 0V.

This technique comes in particularly useful for adding 'bite' to synth sounds such as bass guitar, harpsichord and clavinet imitations. And some of today's velocity-sensitive keyboards are fitted to synths that allow you to vary the pulse width depending on how hard you hit the keyboard, giving PWM a new lease of expressive life, regardless of how it's implemented in a synthetic sense.

And so to another VCO-related effect, that of phase-locked synchronisation. If you've never encountered that expression before, don't worry - it's often abbreviated to 'Sync' by synth manufacturers. For reasons that will become obvious, Sync never puts in an appearance on anything less than a dual-oscillator synth, and even then, designs that incorporate digitally-controlled oscillators (DCOs) rarely feature it, either.

The facility was originally conceived as a means of setting up intervals between VCOs without any of the unpleasant beating effects that normally accompanied such a practice. It works, too. Syncing two VCOs together means you can tune them exactly an octave apart without any of the tuning discrepancies normally associated with using two oscillators. In short, syncing means that the sound has all the pitch and tone qualities of two VCOs locked together, but with the pair of them in perfect phase.

Time to move on to one of the current synth scene's biggest buzzwords - FM synthesis. Thanks to Yamaha's excellent DX polys, complex FM synthesis is now available to more musicians than ever, even if it isn't presented in the most accessible way. But what many synth players don't realise is that FM is actually lurking in the depths of the humblest, dual-VCO analogue synth, albeit in a fairly crude form.

In fact, FM is not the mind-mangling concept many would have you believe. Essentially, it's a distortion effect that arises when one audio signal is used to distort another. Feeding the output of a VCO into the CV input of another creates just such an effect, and the rule of thumb is that the more signal you feed in, the higher the harmonic content of the modulated VCO. This leads us on to those well-known FM expressions, 'modulator' and 'carrier'. The modulator in this case is the modulating VCO, while the carrier is (surprise, surprise) the one being modulated.

In addition to modulation level, another parameter crucial to the operation of basic FM is the frequency ratio between the two oscillators involved. If they're tuned to unison or an octave apart, the resulting sound will be concordant (ie. in tune), but if the modulator's frequency is an odd one that isn't harmonically related to the other, the sound becomes discordant - hence all those wonderful 'clangy' tuned percussion sounds DX7s are so good at creating.

In case you weren't previously familiar with the term, the switching-in of one VCO to modulate another is known as crossmodulation. What the principle can do in sonic terms is give an analogue approximation (the closeness of which will depend on the individual character of your oscillator and filter sections) of digital clarity. It can't provide a vast range of high-quality acoustic approximations the way a proper FM synthesiser can, but it's certainly a step in the right direction. If only more analogue synth owners knew it was there...

As I've said, the higher the modulation depth, the richer in harmonics the resulting sound will be: this is a pretty effective way of adding harmonics to a sound without resorting to the usual analogue practice of filtering an existing waveform.

A word of warning, though. Introducing cross-modulation almost invariably knocks the sound out of tune. The reason for this is that the modulated oscillator sees the incoming signal as a DC voltage, because the modulator is in the audio frequency range. The summing amp principle shifts the carrier VCO up in pitch. This can easily be rectified by retuning the modulated VCO, but this, in turn, affects the overall sound because the frequency ratio between the two oscillators is altered as a result.

You can also come up against tuning problems that arise simply because your synth's VCOs have inherent inadequacies built into their design. They may sound in tune at first, but as soon as you apply cross-modulation, FM-type sounds soon acquire an excessively 'clangy' quality at either extreme of the keyboard, as the VCOs begin to go out of tune. True, you can get around this one by syncing the two VCOs together (as just discussed), but the problem here is that some synths offer only a choice of sync or cross-mod - not both at once.

In other words, using FM synthesis techniques on an analogue synth means experimenting with various different tunings - if you're keen to extract the best results. Mind you, many modern synths bypass all these tuning hiccups by virtue of having digitally-controlled oscillators, so you may not have to worry too much about the last paragraph or three.

After several months of comparatively trouble-free production, Back to Basics fell victim to the Curse of the Gremlins in E&MM June. First off was a typesetting error that resulted in a migration of parentheses, viz the sinewave clause that should have related to the triangle description rather than the square wave one it followed in print. That was a fairly obvious mistake that may well have been spotted by the alert amongst you, as was the artwork muck-up that resulted in the captions for Figure 6 being transposed, with sawtooth text referring to a drawing of a triangle wave and vice versa. Come to think of it, the sawtooth wave image could have been going through a rather more dramatic speeding-up of frequency, but we're not going to draw another one here, so you'll just have to leave it to your imagination...