SYTHESIZER TYPES & HISTORY

Synthesizers are complicated and expensive, so I’ve tried to simplify my collection by selecting just the most capable synths in each category. I see online photos of other people’s collections… massive numbers of synths in racks, with lots of duplication in capability. I can imagine many collectors are not actually very fluent in programming all of those keyboards. I’m not, I have to relearn things every time I switch to a different keyboard. Even with some of the Analogs it can be quite complicated. The modulation matrix in the Oberheim Matrix 12 is quite a bit more to deal with than say a MiniMoog, a Yamaha CS80, or a Roland Jupiter 8. Then you have the FM synths, which take a spectrum analyzer to really program them accurately. I suppose you could simply modify the presets a bit, that’s what most people do. But where’s the fun in that? Be more adventurous!

Yamaha CS-80, 1977

GENERAL SYNTHESIZER TYPES:

MODULAR FM – FREQUENCY MODULATION

MONOPHONIC ANALOG ADVANCED SAMPLE BASED

POLYPHONIC ANALOG PHYSICAL MODELING

WAVETABLE MINI SYNTHESIZERS

SAMPLE BASED VSTs TODAY

MODULAR

The first well known music synthesizers were the large Modular Synthesizers of the 1950’s and 60’s. They were constructed from a set of basic modules, including Oscillators, Mixers, Filter modules, Amplifiers, LFOs, Envelope Generators, maybe a few effects like a Reverb Unit or a Ring Modulator, and maybe an 8 or 16 step sequencer for arpeggios. The modules were mounted in a rack or case with a separate keyboard, and connected to form an instrument using telephone-exchange like patch chords. The first ones in the late 1950’s were huge, they took up and entire wall and were only available at large institutions like RCA Labs.

Don Buchla created a smaller, almost portable one in 1963 in Berkley. The first really portable ones were made by Robert Moog in 1964, which finally put the instruments into the hands of musicians. The University I went to back in the 1980s had a Moog Modular in an electronic music lab along with a number of large multi-track tape recorders. Bear in mind Modular synthesizers have no patch memory, you can’t save your work. To create a new voice you have to undo the previous one, and keep good notes. This is a big limitation of the early synths, in my view. The ARP 2500 series was the biggest competitor to the Moog Modulars, almost everyone was using a Moog at the time. Patching on the ARP was done with a large bank of sliders, above and below the modules, but it has a problem with a little cross-talk between the signal channels. In the 1970’s Roland produced two excellent lines of modular synths, the System 100m and the System 700, which inspired the modern Roland System 500 line of Eurorack modular synthesizers today.

RCA Mark II, 1957

(no keyboard, just teletypes to enter the programming)

Buchla 100 Series, 1966

ARP 2500, 1970

Moog Modular System 55, 1973

Roland System 100m, 1975

Despite being somewhat simpler in nature, Modulars have capabilities that would surprise you, even the old classics like the Moogs and the ARPs have hidden abilities. Studying patches on modern Modulars I’ve learned how to use a sequencer to modify each note you press so every note has it’s own character. You can use the sequencer to set up cascades of effects with the filter and reverb like Vangelis used. The ARP 2500 sequencer can also play sets of 3-note chords. Even a newer Analog keyboard can’t achieve that by itself, so don’t discount Modulars just because they seem like older technology. Some of the new modules can do really interesting things! We’ll get to Eurorack in a few minutes…

MONOPHONIC ANALOG

One of Moog’s technicians collected a set of basic modules from the shop in 1970, mounted them in a small box, and prewired all of the components together (no patch chords required) in the most logical order. This became the standard basic structure for almost every synthesizer that followed: VCOàVCFàVCA. He had constructed the first MiniMoog, which took the market by storm due to it’s lower price and ease of use. Robert Moog wasn’t pleased with the design, he felt the use of patch chords and constructing your own instrument was an important part of the process, but he couldn’t argue with the high number of sales. The MiniMoog was easy enough to use, anyone could sit down at it and get a reasonably nice sounding voice out if it in just a few minutes. Moog still produces a set of Monophonic Analog synths today which continue to be popular due to the unique quality of sounds they produce.

Another significant entry was the Arp 2600, designed with just the most important modules, and set up for teaching synthesis in universities. It was the very first synthesizer you could purchase in a music store, and the ARP 2600 has been cloned today by several manufacturers because of it’s versatility. It had the unique ability of inputting an electric guitar signal which could be heavily modified by the modules in the synthesizer. There were no effects pedals back then, so this was unique.

Moog MiniMoog, 1970

ARP 2600, 1971

POLYPHONIC ANALOG

Up to this point most of the instruments were mainly monophonic. They were only capable of playing one, maybe two notes at a time on a big synth. The next big innovation was creating a polyphonic instrument like the early Oberheim synthesizers. Oberheim stuffed 2, then 4 monophonic SEM modules into a single case in 1975. An 8 voice unit was produced by Oberheim in 1979.

Shortly after this, chip manufacturers began to produce microchips with complete sets of analog modules on a single chip, simplifying the design process significantly. It was relatively straight forward to produce a polyphonic synthesizer capable of playing entire chords of 4, 6, 8 or more notes. Chip manufacturers also began producing digitally controlled modules which allowed the synthesizer to finally ‘Save’ settings in a file. The Prophet 5 from 1978 was the first synth which could save an entire voice as a preset. I consider this feature to be essential, I don’t own any synthesizers without this feature. The Yamaha CS80 and the Roland Jupiter 8 are also exceptional synthesizers in this category (but without much patch memory). The Oberheim Matrix-12 from 1984 is one of the last, best examples of a Polyphonic Analog synth ever created, with 12 voices and more internal components than just about everything else. The Matrix-12 is one of my personal favorites, half analog, half digital, and with Midi and lots of patch memory slots.

Oberheim FVS-1, 1975 (4 voice)

Dave Smith Prophet-5, 1978 (5 voice)

Roland Jupiter 8, 1981 (8 voice)

Oberheim Matrix-12, 1985 (12 voice)

OK, now that we’ve covered some of the history let’s take a look at the how these machines work inside and see what the differences are. Remarkably, there are more similarities than differences.

MODULAR / ANALOG STRUCTURE

Modulars don’t actually have a set structure per say, you wire up the modules when you insert the patch chords. That said, there is a conventional arrangement that popped out near the beginning. The core set of modules is the well known VCOàVCFàVCA signal path which Moog established with the MiniMoog Analog synth. It held the absolute minimum number of practical modules and modulators, but no Effects. ARP went a little broader with the ARP 2600, with ways to cross connect everything along with adding two Effects, a Ring Modulator and a Reverb module. A lot of Modulars also included an 8 step Sequencer for arpeggios.

I’m just showing the basics below, not an entire block diagram. Envelopes and LFOs tap in here and there, along with several physical controllers like the foot pedals and mod wheels.

Some Analogs have more Envelopes and LFOs than others, some include Ramp Generators, Trackers for Keyboard tracking, FM and oscillator Sync, and a Modulation Matrix like the Oberheim Matrix 12. Others like the Roland Jupiter 8 have an Arpeggio module and a Chorus feature. The extra features are what you need to consider when selecting a synthesizer, the core signal path is pretty much the same across most synthesizers of the same type.

WAVETABLE

Somewhere in the middle of all of the development going on in the early 1980’s came the PPG Wave synthesizer in 1981. It expanded the capability of the basic oscillator from simple Sine, Triangle, Square and Pulse waves to add an entire table of new wave forms. The selected wave form could be repeated like the simpler wave forms, or it could be chained and mixed with others in interesting ways to create some really curious, evolving sounds. I really don’t have much experience with these. Equator2, my newest hybrid soft synth has this capability along with a number of other synthesis technics, but I haven’t experimented with it much yet.

The Sequential Circuits Prophet VS also fits in this category. It allowed you to select four different waveforms from the 127 in the wavetable, then a joystick directed the mix of waveforms used in your patch. They called this Vector Synthesis. Later on, the Roland D-50 also sported a joystick for blending it’s sampled waveforms.

PPG Wave, 1981

Sequential Circuits Prophet VS, 1986

WAVETABLE STRUCTURE

The Wavetable synths made a slight upgrade to the basic system, they expanded the number of waveforms from four up to 128 or so. A new block in the diagram is added on below. Vector synths allowed you to select four waveforms out of the wavetable which were then mixed based upon the position of the joystick. The diagram below shows just two VCOs.

SAMPLE BASED -ORIGINS

One of the earliest sample-based synthesizers was the Mellotron from 1963, which used loops of audio tape to play back the sounds. It was prone to the flutter and wow of tape recorders and wasn’t very reliable. The first really practical sample based synthesizer was the Fairlight CMI, which debuted in 1979. The Fairlights used digitally recorded samples of acoustic sounds to create voices in the synth. The software and light pen gave you the freedom to digitally edit and create your own samples quite easily. The Fairlight was way ahead of the technology of the day, but it’s cost was way up into the stratosphere. The new recreations of the Fairlight are also still quite expensive. The originals required a computer for operation back when computers were in their infancy. We’ll talk more about Sample-Based synthesis in a minute.

Mellotron Mk-II, 1963

Fairlight CMI LiX, 1979

SAMPLE BASED STRUCTURE

The first sample based synths were built on the same basic Wavetable diagram. The wavetable was massively expanded from very simple digitized waveforms to large recorded sound samples. It took a lot more memory to store the samples, but it’s really just an expanded wavetable. They also increased the number of samples used per voice, on average they worked their way up to about one sample for every 5 or 6 notes up the keyboard. The first synths of this type only used one, which is why some of them don’t sound very good. I’ve found you can bend a sample up or down about an octave before it starts to fall apart, it starts to sound strange rather than model the instrument you’re after. It’s like bending a human voice too high and getting a Chipmunk voice.

FM – FREQUENCY MODULATION

Complex harmonics were difficult to create in an Analog synth, making it hard to realistically replicate many acoustic instruments. In 1983 Yamaha released the first really new type of synthesizer, the completely digital Yamaha DX1 and DX7. John Chowning who invented FM discovered how to use one oscillator to modulate another, to create a massive number of harmonics. Frequency Modulation replaces the VCO Oscillator in an Analog synth with a set of FM Operators, arranged in an Algorithm. FM does a very efficient job of creating multiple harmonics, broadening the types of sounds the instrument can produce. The DX7 (using 6-Operator Sine wave FM) retailed for around $2000, which undercut the high cost of the premium Analog synthesizers of the day and completely took over the market. The DX7 sold over 200,000 units over it’s lifetime, dominating all of the music produced in the 1980’s. The DX1 was more iconic, two DX7 engines under the hood, an exotic LED display, but only 140 were manufactured. Later FM synths from Yamaha used 4-Operator FM which simplified the programming, reducing the price of the synths even more.

An advanced variation of FM was produced in 1999 in the Yamaha FS1R which combined 8-Operator FMX with high-harmonic wave forms and Formant waveforms as well. This combination was very powerful and is implemented in the current Yamaha Montage, but without the Formant waveforms. No one can figure out why Yamaha left Formants out… FMX works something like an additive method, you model a sound by reconstructing it’s harmonics one section at a time. FMX can produce a massive range of sounds, and with Formants mixed in that expands even further. Formants allow you to mimic acoustic instruments with a resonating body, like the Guitar, Cello, or the human voice. Formants on the FS1R are relatively unexplored territory.

Yamaha DX7, 1983

Yamaha DX1, 1983

Yamaha FS1R, 1998

FM – FREQUENCY MODULATION STRUCTURE

FM was really the first major departure from the standard model. FM replaces the VCOs with a block of digital Oscillators, VCAs and ENVs arranged in an Algorithm. Using one or two oscillators to modulate a carrier oscillator can create an enormous number of harmonics, which is enhanced by adding feedback onto one of the Operators. Most of the same basic components are here, but creating a sharp sawtooth, pulse wave or triangle wave takes some work. An FM synth can emulate a LOT of different sounds, far more than an Analog synth. Looking at the advanced, modern FM instruments starting with the Yamaha FS1R, Yamaha developed FMX which uses a set of very complex waveforms in addition to sine waves in the individual operators. This greatly expands the timbres possible on the instrument. The wavetable option actually started much earlier with the Yamaha DX11, a 4-Operator FM instrument.

Formants are another variation to the FM synth, but unfortunately only the Yamaha FS1R has this option. A Formant waveform is the combination of the FMX ALL1 harmonic waveform and a narrow band pass filter. You can emulate Formants on other synthesizers, such as using the 5-band Equalizer on the Yamaha VP1, but the FS1R is really the only synthesizer designed with Formants in mind as the dominant feature. The Formant waveform is selected as an Operator waveform, allowing you to mimic resonant body instruments like a Cello or the human voice.

ADVANCED SAMPLE BASED

In the late 1980’s another method of synthesis was developed by Korg, call Pulse Code Modulation, or PCM. Roland developed their own version and started production in 1987 with the Roland D-50. Yamaha’s version is called Advanced Wave Modulation, AWM. Basically, these synths use digital recordings of an acoustic instrument in place of the Analog oscillator, to match the sound of real acoustic instruments more closely. As computer memory increased in size and decreased in cost, the size of the recorded samples increased dramatically, improving the quality of the sound. This method of synthesis has continued right up to the present, some of the best synthesizers today use this method of synthesis.

The most dominant synth in this category in the late 1980’s was the Korg M1 (beginning in 1988) which sold over 240,000 units. It’s arguably the most prolific synthesizer ever produced. Later on in the 90’s a wide variety of Tone Generators were available which used EPROM chips to store the instrument samples. Roland had the Sound Canvas line, Yamaha had the MU line of tone generators, there was the Korg Wavestation, and the Edirol Studio Canvas. Numerous keyboards popped up based on the same basic digital technology. Yamaha’s AWM2 method dominated in the 2000’s in the Motif line of synthesizers, which continues in the Montage line today. Roland and Korg have their own versions of PCM/AWM today as well.

Lately I’ve been looking into the E-MU Morpheus and Ultra Proteus synths which were developed in the early 1990’s. They have a very unique 14-pole Z-Plane morphing filter which is quite fascinating. It creates wonderful pads and soundscapes like you find in a lot of movie soundtracks. Probably one of the most popular Euro-Rack filter modules, the Rossum Electro-Music Morpheus Z-Plane filter, is based on this very original filter design.

Korg M1, 1988

E-MU Morpheus, 1993

Tone Generators: Korg NS5R (1996), Yamaha MU80 (1994), Roland Sound Canvas SC-55mkII (1991)

Yamaha Motif-XF8, 2014

MODERN ADVANCED SAMPLE BASED STRUCTURE

To create an advanced sample based synth, manufacturers like Yamaha increased the number of samples used to create a voice. You can overlay up to eight samples on the Motif and the Montage. Some samples are used just for the Attack portion of the voice, some are used to create the main voice, another can be mixed in to create a high velocity portion of the sound, and others can be used to handle the key release, or they can model anything else you can dream up. Besides the controls on each part, the block diagram remains pretty much the same as the one we started with, there are just eight parallel sets of components now that mix together. The newest Montage M line allows you to overlay 128 parallel sounds… I can’t for the life of me figure out why you would need that many.

Looking at the diagram, I think you can see that a modern Advanced Sample synth can also create Modular/Analog sounds pretty easily. There is always a sawtooth, a square wave, pulse waves, a triangle wave and a sine wave in the sample set. Modern synths also include a large selection of Effects, which greatly enhances the quality of the sounds you can produce, along with arpeggios to enhance your playing.

PHYSICAL MODELING

In parallel with the first sample based synths, a few of the manufacturers experimented with Physical Modeling which mathematically models the characteristics of acoustic instruments. Yamaha was the first, with the Yamaha VL1 in 1993. The VL1 very accurately modeled wind instruments in every aspect, including reed squawks and performance timbre changes. It also modeled string instruments to a high degree. Technics made an attempt with the SX-WSA1 which uses a method halfway in between PCM and Physical Modeling, but it’s a rather weak entry into the field. Korg also entered the fray in 1997 with the MOSS system as add-on boards in the Z1, but the MOSS acoustic models were somewhat inferior to Yamaha VL. In addition to the two-note VL1, Yamaha also produced a 16-note VP1 prototype in 1994 which modeled percussive instruments like the piano, guitars, drums and plucked string instruments. It’s an impressive instrument, but only 9 copies are know to have survived. Physical Modeling is making a comeback today in VSTs, now that computers are capable of doing the sheer number of calculations required by the models.

Yamaha VL1 w/ Breath Control, 1993

Yamaha VP1 w/ the Editor, 1994

Korg Z1, 1997 with MOSS technology installed as an expansion card.

PHYSICAL MODELING STRUCTURE

Physical modeling gets really interesting when you look at the internal components. It starts with a simple impulse wavetable where you select the ‘Driver’ waveform for the VCO, which then feeds into one or two delay loops. Tuning the Delays and setting up the internal Filters and Attenuators in the Delay Loops to get exactly what you want is immensely difficult. Most synthesizers that use physical modeling set you up with a specific starting point near where you want to go. That way, it’s not so difficult to program usable sounds. The output from the Delay loops is mixed and fed into an output Filter and an Amplifier just like we did before. All of the components for an Analog or Wavetable synth are inside if you bypass the delay loops. With the delay loops, the simpler sounds of the Analog portion of the synth are transformed into something magical. It begins to sound just like a plucked string, a blown reed, or any other acoustic instrument sound. Coupled with a number of integral Filters, Attenuators, and Inharmonicity modules, you get the sounds you normally hear from an acoustic instrument.

The Yamaha VL1 also adds a large number of new components onto the core modules. Most of these add-ons modify the ‘Driver’ waveform to mimic the mouthpiece, bow, or finger pluck of the instrument you’re trying to model. These add-ons are linked to controllers so the player can mimic acoustic instrument articulations as they perform. The Yamaha VP1 has none of these add-ons, it’s a more general acoustic modeling instrument which makes it a bit more difficult to program, but far easier to understand. You can see the internal block diagram for the Yamaha VP1 from the editor below, which really only shows one half of the model at a time. The VP1 models plucked or struck instruments, the VL1 models wind and string instruments.

VL1 Controller Parameters

VP1 Block Diagram

MINI SYNTHESIZERS

In the 1980’s and 90’s advances in electronics were making significant strides in reducing the size and complexity of building a synthesizer. Along with the Tone Generators, several micro-sized synths emerged, which really started to change the synth world and reduce instrument prices. A few of these early models were the British Wasp (a crazy anomaly in 1978!) the Roland TB-303 in 1982, the Kawai K1m in 1988, the Yamaha VL70m in 1996, the Nord Micro Modular in 2000, groove boxes like the Yamaha DX200 and AN200 emerged in 2001, then things exploded in about 2010 with the Blofeld Waldorf, Korg Volka, Arturia MicroFreak, Arturia Micro Brute, Behringer Pro-VS, and on and on. There are so many now I can’t even begin to list them all!

EDP Wasp, 1978

Roland TB-303, 1982

Kawai K1m, 1988

Yamaha VL70m, 1996

Nord Micro Modular, 2000

Yamaha DX200, 2001

There are Hundreds of Mini’s Today… and Eurorack Modulars have gone nuts!

MODERN MODULARS

Going back to Modular Synths for a minute, in about 1996 Doepfer Musikelektronik released the very first Eurorack-format modular synthesizer system, the Doepfer A-100. They followed it up with a series of compatible modules in 1997 and 1998. The format really started to grow in popularity in the 2000’s, with over 700 modules from about 80 manufacturers in 2013. Today (2023) there are over 15,000 different modules from over 1000 manufacturers! I’ve stayed away from that deep, deep rabbit hole and stuck with my low-cost virtual Modular system instead. The system pictured above is easily over $25,000. You can get really deep into it really fast, one of my friends dropped over $5000 on modules in just a couple of weeks putting his first system together. I can put a virtual system like that together on my computer for a tiny fraction of the cost. I also have the Nord Modular G2 from 1998 which did everything in software, then downloaded the setup to the keyboard. All the wires are on the inside! A friend of mine in Germany tells me I’m missing out on the raw immediacy and sound quality of the hardware Modulars, though.

Doepfer A-100, 1996

Nord Modular G2, 1998

VSTs TODAY

The first Soft Synth I had experience with was the Yamaha SYXG100, which modeled a Yamaha MU128 Tone Generator with the PLG150-VL Acoustic Modeling card installed. It appeared in about 1998, the PLG cards debuted in about 1997. In the 2010’s we saw a massive surge of virtual synthesizers implemented entirely in software. These virtual synths started by emulating the older styled synths, but have expanded radically into new areas like Granular Synthesis, massively parallel additive models, Modular Synthesis, and every technique you’ve never heard of. There are literally hundreds of VST modules now to choose from, in every variety and flavor. VSTs are relatively inexpensive compared to the hardware synths, which helps musicians who are on a budget. Many can even be obtained for free! For myself, I prefer hardware synths which make up most of my collection. Besides, you only have so much space on your computer screen to work with. Crowding my DAW and recording software with a bunch of VSTs just doesn’t work that well for me.

There is so much variety available today! If you can’t find what you’re looking for you’re not looking hard enough. Or, you might be overwhelmed with the sheer number of options available. If you find yourself a bit confused, there are a number of excellent guides online to dig through. The older classic hardware synths are also seeing a comeback today, but some are at exorbitant prices as collectors snap them up. Other newer synths are reworked versions of the classics, coming out from Behringer and Korg, with modern improvements like a USB Midi interface and Effects.

Another angle here are the VST Modular synths. I wanted to work on an ARP 2500 which clearly is way out of the budget at well over $100K just for a center console. A friend of mine has one of these for real, a beautiful classic ARP 2500 modular from the annals of history. I went digging around and I found a very nice set of ARP 2500 modules in the Voltage Modular library for the low price of just $49 from Cherry Audio. I have over 200 different modules available in Voltage Modular now, more than the 160 modules my Nord Modular G2 has. And then I saw the open source VCV Modular system… Oh My Heavens! It’s a free modular synthesizer system with over 2200 free modules available! There is so much available for it, it blew my mind! With such a HUGE selection of modules, most of them free, how in the world can you justify buying an equivalent system in hardware? You can’t, can you? I’m struggling here, I really prefer hardware to a soft synth, but this option has me on the fence now. I’m still sorely tempted to purchase a complete set of Roland system 500 modules I see for sale, but the Voltage Modular system I already have covers all those bases and then some. What would you do?

Analog Labs Sylenth1 VST

Voltage Modular VM2500 Patch, 2020 (software clones of the original ARP 2500 modules)

Cherry Audio Moog 900 series modules for the VM system.

VCV Modular Patch on YouTube by State Azure, 2023

My latest acquisition is a Cherry Audio GX-80 soft synth which is a combination of the Yamaha GX-1 synthesizer used by Keith Emerson in the ‘Fanfare for the Common Man’ and the Yamaha CS-80 used by Vangelis. They’re based internally on the same Analog electronics. In the same market space, Arturia has really gone all out in modeling classic synths on your computer. Their V-Collection 9 has 32 virtual classic synthesizers in it, including a lot of the ones talked about in this article… and then some!

CONCLUSION

I think you can see from the simplified diagrams that most synthesizers are very similar on the inside at the core level. The differences come in with all of the extras like the arpeggiators and controllers you can add on. Because of this, you don’t need every single synthesizer out of a particular category to get access to most of the sounds. In reality you can get almost exactly the same sounds out of each of them. I know I’m going to get in a lot of trouble for that statement, but as a patch programmer I have a different perspective from the musicians out there who rely mainly on the presets. I can program almost any sound I want out of each of them. Yes, some filters or oscillators are smoother or grittier than others, but if you work at it you can program almost identical patches with each different synthesizer.

As a result, I’ve tried to limit the synthesizers in my collection by only acquiring a couple of units out of each category. I try to figure out which one in a particular category is the best, then I go after that one. Well, cost does come into play as well. I would really like to get an ARP 2500 Modular, but I don’t think I could afford over $100,000 for one even if I did find one for sale. Oh well. I did find a surrogate in software. You don’t need to spend a lot of money or put together an expensive collection of synthesizers to make good music. All it really takes is a basic instrument and some talent. But, if you’re into collecting keyboards, go for it! Everybody it seems likes something different. Judging from the photos I’ve seen, nobody has the same hardware!

Isao Tomita, Moog Modular IIIc