Enter the digital domain

Previously we went through a brief history of synthesis, and started going into it in detail starting with why analogue synthesizers can have a sound that is fairly unique thanks to the manufacturing tolerances of the components that make up the basic synthesizer building blocks (Oscillator, LFO, Filter, Amplifier and envelope generators), now we move on to digital synthesizers, and explain a little more about how they work.

Now were going to move on a little with the birth of the affordable digital synthesizers.

In the mid 80’s, all that went out of the window, with the launch of a new range of synthesizers by Casio and Yamaha.  Lets first look at what Yamaha did to kick start the digital era, and then we will have a look at Casio’s system and compare the two.

Like analogue synthesizers, digital ones all start with a sound source. Yamaha’s DX series used a form of synthesis known as Frequency Modulation (FM), which was first introduced in a more basic form in the beginning of the 80’s in the more organ like GX1.  The GX1 was big and expensive, and was put together by Yamaha as a testing bed for they’re newly purchased synthesis engine.  The price at the time was a hugely expensive £10,000, more than the price of a small house at that time, and only a few made it into the U.K. with a few more escaping into the rest of the world.  It was marketed by Yamaha’s electronic organ division, and of the few that bought it; none really knew that the GX1 was to be the father of an all-conquering family of synthesizers in a few years time.  The GX1 was the forerunner of the DX series, which arrived towards the end of 1984.

Yamaha DX7 Synthesizer
The Yamaha DX7 introduced digital synthesis to the masses in 1984, allowing the user to create accurate digital re-creations of real instruments, as well as creating new sounds, which were never possible before. This new super-synthesizer also allowed you to store 32 of your own creations in it's on board memory, ready for you to recall and play, either from the keyboard. or from the new Musical Instrument Digital Interface (MIDI). The DX7 took the world by storm and sold out all over the world, and it's bigger and more expensive brothers, the DX5 and very rare DX1 also sold relatively well considering the price hike from the DX7. Soon the radio was filled with these new digital sounds, but it didn't take long for some people to notice that the sound was quite thin in comparison to the analogue synthesizers that were still being played at that point. This was overlooked however, as the DX series could play chords and up to eight notes at once from a sequencer, without huge cable looms hanging off the back, or needing more than one synthesizer.

Yes there were polyphonic analogue synthesizers around, but only one of these had MIDI built in (The Sequential Circuits Prophet 6), though others could have external interfaces added, such as Roland's Jupiter 8, but all of them were more expensive than the £1500 asking price (In the U.K. at time of launch for the DX7) and more temperamental than the DX and the monophonic analogue synthesizers, which were also quirky. The DX series spelled the end for analogue, despite the cold character of it's sound in comparison to the older generation.

So why was the sound considered colder and thinner than analogue? Well, lets have a look and see.

The sound in FM synthesis starts with the sound source, known as an Operator in Yamaha land. This produces a sine wave at the desired frequency, which in itself is not very interesting to listen to. This is passed to an envelope generator so shape the output of the operator, but aside from now controlling the output volume over time, the sound is still, quite frankly, dull.

So how did the DX7 take over the world? Because there was always 6 operators available for each note being played, and the operators were arranged in groups called Algorithms, and there were 32 of these, each one creating different possibilities for sound creation. How did it work? By using carriers and modulators. Getting lost yet? Well, I'll explain...

Lets look at a simple FM synthesis system with 2 operators, a carrier and a modulator. This would be shown as:

2 Operator F.M. synthesis block diagram

Operator 1 feeds into Operator 2 and modulates, or alters, the sound output of operator 2. In this case, Operator 2 is the carrier while Operator 1 is the modulator. The output is then sent to the amplifier before reaching the mixing desk.

What no filter?

Correct, no filter, and at the time there was reasoning behind this. Analogue synthesis generated a sound with a lot of harmonics, many of which had to be removed to get the desired sound, the filter did the removal and shaped the harmonic content of the sound, this is known as subtractive synthesis. FM on the other hand, built up its harmonics by adding sine waves of differing frequency together to generate more complex waveforms. If you want more harmonics, use more operators and/or wider frequency settings, for fewer harmonics reduce the frequency of the modulator(s) and/or use fewer operators. This is a process known as additive synthesis, as you are adding the most pure sound you can have with another at a different frequency to get a more complex waveform. Soon we will go into more about algorithms, and that's where things get pretty scary.

But why does it sound thin and cold? Well, in the analogue days, the holy grail of all synthesizer manufacturers was to make the perfect oscillator, which could produce the perfect sine wave, the purest sound of all. Because what was generated by the operators was pretty close to a perfect sine wave (For the time), as the operators were created in silicone and the manufacturing tolerances that gave analogue synthesizers a warm and full sound were much less in integrated circuits, to the point that each operator on each chip on every synthesizer was as good as identical, and so was the sounds that each machine produced. As a result, two operators playing the same sound at the same frequency just made a louder sound, no drifting and no beating as a result, just a clean static sound. There were randomness parameters to try and introduce some drift, but much of the time these were not used, and when they were, they were at best a simulation of what had gone before, not an accurate replacement. There were no filters; no sounds generating self-oscillation, which made speakers go pop if the volume was too loud, nothing. But it had memories, MIDI, eight notes played at once, and six, yes SIX oscillators for each note played!  No wait, this is Yamaha land so operators for each note played. Ho hum.

Anyhow, enough delaying, lets get on to algorithms...

With six operators to play with, arranging them to make complex, realistic and interesting sounds was thought to be beyond the user at this early introduction to FM, so Yamaha made 32 arrangements of operators in a number of lovely and interesting patterns across the top face of the DX7's case. What many, indeed nearly all of the initial owners didn't know, was that these pretty pictures were showing them how the sounds the were falling over themselves to get hold of and play were actually constructed. In the simple example above, the two-operator arrangement could be considered a replacement for an oscillator in a traditional analogue synthesizer. So to make a three oscillator synthesizer, the algorithm would look something like this:
6 Operator algorithm
Each 'oscillator' is made up of two operators, a carrier and a modulator. Still with me? Well here is something to muddy the waters a little, in the above example, operator 3 has an option for a feedback loop, something which was supposed to act as a replacement for a filters self oscillation, but more often than not, sounded more like ring modulation and if used too much, made everything sound metallic.  However if used sparingly, the feedback could introduce elements that enhanced the sound and added character, the problem was it would be exactly the same, each time the key was pressed, on every example of a DX7! Remember, there are 32 different algorithms on the DX7, each capable of creating it's own range of sounds, from all six operators piled on top of each other, to all six in a neat little row and a few combination's in-between. However, that's all I want to cover on algorithms for the moment, so now we will move on.

Analogue synthesizers allowed you to meddle with the sound as you were playing, as every parameter that could be altered by the user was there to tweak on the front panel, however digital systems like the DX7 had so many possible parameters to edit, that to present them all on the front panel would take up an entire room. Not the most practical way of selling a synthesizer, and the additional cost of all these controls would be prohibitive. So as a compromise, the buttons for selecting which voice you wanted to play, also doubled up as a parameter or operator select button in edit mode, and a small Liquid Crystal Display (LCD) was provided to tell you what you were editing. Some have described this as wallpapering a hallway through a letterbox, and believe me, it is an apt description. Consider it this way, there are thirty two buttons to select voices and in edit mode, six of these are used to select which operator you are editing. The remaining twenty six buttons each have one editing parameter on them, which must be selected six times to edit that parameter on each operator. That's one hundred and fifty six different parameters that can be edited, assuming these buttons only have one, some had more(!), but these tended to be global and affected the whole sound. As a side note, Yamaha's service canter’s noticed that the majority of DX synthesizers returned for service or repair, still had all the factory default sounds loaded, probably because no-one could understand the synthesis, or that no-one could be bothered trying.

The net result was that many of the FM synthesizers sold sounded exactly the same, so the new wonder-sounds became boring and stale over time. Also, Yamaha introduced a simpler four-operator range to the market with only eight algorithms, and produced variant after variant of the four-operator design. The market was flooded, and other manufacturers were now stealing Yamaha's thunder, an example of which we will cover next time, but for now I will mention Yamaha's last stand on four operator synthesizers, the TX81z and the DX27.

The TX81z and DX27 introduced a twist in FM synthesis, the operators no longer just generated sine waves, they could generate other waveforms as well, such as partial sine waves, saw tooth waves and pulse waves. Now if you know your physics or maths and especially the bit about adding sine waves together, you will realise quite quickly that these additional waveforms are all mathematical derivatives of adding sine waves of different frequencies together. In effect, this gave the four operator synthesizers a sound range closer to that of the six operator synthesizers, without adding the layers of complexity. This occurred around 1988, four years after the introduction of the original DX7, and the same time as Yamaha launched the DX7 mkII, DX7 fd and DX7s, though oddly enough, the sound ranges of these two synthesizers was in reality the same as the earlier model, they just had better converters (Which improved the quality), and the MK II and FD models had the ability to play two sounds in unison (The fd also had a built in floppy disk for storing sounds), other than that they were identical to the original.

While the new waveforms allowed a wider sound palette at a lower cost, the lack of warmth was still evident as each synthesizer was as good as exactly the same as the rest of the production line, so sound reproduction was perfect every time. There were still no filters and the generated waveforms were as good as perfect, the synthesizer Holy Grail was here, but as the saying goes, be careful what you wish for, as you may well get it. The synthesizer world did get what it wished for; as good as perfect waveforms, but the end result lacked a little something.

The DX series also did one other thing, which changed electronic music, though many don’t know it.  When the specification for MIDI was laid down, manufacturers jumped on it as it would mean instruments could talk to each other with a set standard.  The DX7 was the second instrument on the market with the new MIDI interface, the first being Sequential Circuits Prophet 6, which was analogue in nature but digitally controlled.

But Yamaha played with the MIDI specification and set the Modulation wheel on a different MIDI command to the MIDI specification whereas Sequential stuck to the proper design.  As a result, the Prophet 6 could not send modulation data to the DX7, and if the DX7 modulation wheel was used on the Prophet 6, nothing happened.  Had the Prophet outsold the DX7, not much would have changed, but the DX7 sold in such huge quantities, if people didn’t want problems, they had to conform to the Yamaha implementation.  The MIDI specification was quietly changed in 1985, and manufacturers conformed to the Yamaha implementation from that day on.

An interesting side note, but what about Casio?
Casio CZ101.jpg
Well, in 1985 Casio launched their challenger to the DX series of synthesizers, starting with the rather diddy Casio CZ-101 with 4 octave mini keyboard, MIDI, the ability to play more than 1 voice at once over MIDI, could have a guitar strap attached, could run on batteries (So you would not trip over the power when using it like a guitar trying to look cool on stage) and had Phase Distortion (P.D.) synthesis.

Like Yamaha’s F.M. synthesis, Phase Distortion worked by adding sound waves together, but unlike Yamaha, Casio took the complexity of creating complex waveforms out of the users hands.  They provided a ready made selection of harmonically rich waveforms and provided you with a number of options for you to manipulate them.  Once again there was no proper filter, but they did make the envelope generators more complex and allowed you to use from 1 to 4 sound sources, however using any more than 2 sound sources halved the number of notes you could play at once from 8 to 4.  Like F.M. each sound source can be set to different frequencies, but there is a point at the end of the sound path that allows you to tweak the brightness of the sound.  The end result was a digital synthesizer that had an almost organic quality to the sound, and almost added the warmth of analogue.

I say almost, because it is still a digital system and behaves as such, although the ability to play 4 different voices over MIDI made it popular with the likes of Vince Clarke (Who throughout the late 80’s was a big Casio fan, appearing in many of their advertising literature).  Like F.M. the Casio synthesis system could make harsh metallic sounds, but could also create delicate timbres, which almost shimmer through a busy mix.

Sadly it arrived after the all-conquering DX7 and looked toy like in appearance, so by the time the bigger CZ3000, CZ5000 and the flagship CZ1 arrived, it was all sadly too late.  Even the last hurrah of the superior VZ1 and VZ10m in the early 90’s was not enough to give Casio headway into the professional studio’s, and Casio withdrew from the market, concentrating on home keyboards and taking with them one of the best synthesizer engines of the digital era.

Next time we will go into the next generation of digital synthesis, which arrived from the creator of the mighty Jupiter 8 analogue synthesizer, and stole the crown of synthesizer king from the DX7, but was it as good as it seemed?