The evolution of the 900 Series
Bob Moog continually revised his 900 Series range of modules. For example, not only is his 1960s flagship Synthesizer IIIc significantly different to his 1970s flagship System 55, but even the Synthesizer IIIc itself evolved, adding and replacing various modules over the years.
901 vs. 921
Let's start with perhaps the most controversial improvement. The 901A Oscillator Controller converts the initial control voltages, from 1V/Oct, to doubling the voltage per octave. Each 901B Oscillator's wonderfully named Frequency Vernier dial then offsets this already-converted voltage. The upshot of this is that if you play an A2, and the oscillators are all 0.5 Hz out from one another, then you play an A3, they'll all still be 0.5 Hz out from one another.
In contrast, the 921A Oscillator Driver simply passes along the dial-offset 1V/Oct control voltages to each 921B Oscillator, which performs the conversion itself, after taking its own frequency offset dial into account. So if you play an A2, and the oscillators are all 0.5 Hz out from one another, then you play an A3, they'll all be 1 Hz out from one another, as their relative frequency offsets are doubled with each octave, along with the overall pitch. This proportional beating works exactly as it would if you sampled a single note from the 901A/B pair and played the sample back at different pitches.
While some people consider the 901A/B pair's linear frequency offset to be its secret sauce, this change has a big advantage: the 921B Oscillators can be detuned by a set number of semitones, that track consistently across the keyboard's range. This lets you play dyads or chords, with the monophonic keyboard or ribbon controller merely dictating the root note, and the other notes staying exactly, say, a major third above it. To my ears, a relentless use of only minor chords or only major chords with a disregard for whether they're in key is associated with some of the cheesier 1990s rave music that uses more modern equipment such as samplers to achieve the same effect, but Moog's 1970s modular synthesisers are very well suited to this.
The other improvements are more universally agreed upon as such: pulsewidth modulation and oscillator sync open up a whole range of possibilities. Indeed, pulsewidth modulation allows you to have fixed frequency beating after all, albeit by the same amount across a whole oscillator bank, and using an LFO to do it — an oscillator employing pulsewidth modulation sounds like two oscillators detuned by the same amount as the LFO's frequency.
The unpaired, standalone 901 module's 921 replacement is similarly more featureful, with the only loss being a built-in mixer to blend the waveforms together. This can still be achieved via a separate mixer.
903 and 903A vs. 923
The short-lived 903 offered just white noise. The 903A added pink noise. The 923 also added a highpass and lowpass filter, letting the user sculpt the noise a bit more without using up any other modules, and making better use of the available physical space. There are no downsides to this upgrade.
The 1960s version of the 904A Voltage Controlled Low Pass Filter sounds better to my ears, but the 1970s version can self-oscillate.
The 904C Filter Coupler allows the 904A lowpass and 904B highpass filter to optionally work together to form a bandpass or notch filter. This module is omitted entirely from the later systems. I believe this is the only instance of them actively losing a useful feature.
907 vs. 914
Wendy Carlos asked Bob Moog to create the 907 Fixed Filter Bank, to emulate the tones of various acoustic instruments. Everyone else used it as a graphic equaliser, or not at all. She also later commissioned Moog to modify it to use as a vocoder.
Moog later replaced the 10-band 907 with the 14-band 914. None of the original bands were removed or changed in the upgrade, which simply added two more at the bottom and two more at the top (though naturally the end lowpass and highpass filters were nudged along accordingly). Surely a good thing.
CP3 vs. CP3A
The CP3's mixer had a pretty significant issue: as it drew power from +12 V and -6 V rails, when the signal reached about 5 V, it would distort asymmetrically, with the negative part of the waveform distorting while the positive part remained unaffected.
The CP3A added more headroom, afforded by additional +15 V and -15 V power rails in later system busses. This is technically much better, but a lot of people prefer the asymmetrical distortion and consider it an important part of that big Moog sound.
It seems that Moog was no exception to the trend of successive synthesisers being generally more featureful and refined than their predecessors, yet losing some of the "warmth" and charm brought about by those original imperfections. Even in the 1970s, you could claim of the latest Moog modulars that "Things were better back when they were worse!"
- "Electronic Music Composition-Performance Equipment Short Form Catalog — 1967" Moog, 1967, p. 10
- "Moog 1971" Moog, 1971, p. 10
- "901-A Oscillator Controller" R. A. Moog Co., Jul 1966
- "900 Series service notes" Norlin
- "Synth-Werk Moog 901 A/B Clones" Robotmakers, Mod Wiggler, Dec 2017
- "Synthesizing Strings: PWM & String Sounds" Gordon Reid, Sound On Sound, Mar 2003
- "Behringer 904A Rewinds to 1967" Rob Keeble, Nov 2020
- "Dr. Robert and His Modular Moogs" Richard Leon, Sound On Sound, Oct 2003
- "Vocal Synthesis" Wendy Carlos, Secrets of Synthesis, 1987
- "Photo Archive I" Wendy Carlos
- "Mix and Attenuate" Rob Keeble