Sonic Evaluation of a Small Moog Modular System

By Ben Luce

This study conveys the results of a short but revealing study of a modular synthesizer that consists of two modular Minimoogs inside the same cabinet, that resides at EMEAPP. One of these utilizes the original 901-B oscillators manufactured by R.A. Moog Inc, the other 921B oscillators manufactured by Moog Music Inc.. Some comparisons are made with the Minimoog. The following photos show this instrument in a set-up form, and then the front panel in detail:

The Moog Model 15 Modular Synthesizer, in set-up form.
The Moog Model 15 Front Panel.

Conclusions about the 901-B oscillators:

  • Sine Waves: As the following spectral graph shows, which was generated with Logic Pro X’s EQ plug-in, the sine wave output of these oscillators exhibit clearly audible 2nd and 3rd harmonics. Their audibility can be readily gauged by using the EQ plugin’s filter to remove the 2nd and 3rd harmonics, the effect of which can then be instantly A/B compared by turning the EQ on and off:
Figure Caption: Spectrum of a 440 Hz sine wave for a 901-B oscillator, showing the 2nd and 3rd harmonic content.
  • Rectangular and Square Waves: The oscillators have a master controller that enables the pulswidth of the “Pulse” waveform (rectangular waveform) to be adjusted, and can be used to create an extremely accurate square wave (with virtually no even harmonic contamination) but it was found difficult to obtain perfect square waves for all three oscillators at once: Specifically, when two of the oscillators (1 and 3) were adjusted to a perfect square wave, the 2nd oscillator sounded more like a sawtooth wave, as it contained very audible even harmonic components as well.
  • Triangle Waves: The triangle waveforms were all found to contain a significant (clearly audible) level of even harmonics, a phenomenon due to the particular rectifier style circuitry used in the oscillator, and which is also present, for example, in the Minimoog. The pulsewidth control on the master controller does not effect this.
Figure Caption: Spectrum of a 440 Hz triangle wave for a 901-B oscillator, showing significant (clearly audible) even harmonic content.

Conclusions about the 921B oscillators:

  • In general, the output amplitudes of the 921B oscillators are somewhat lower than the 901-B oscillators, especially for the rectangular and triangle waves, which are roughly 3 dB and 6 dB lower than their 901-B counterparts. This has implications for the levels of distortion produced downstream, and may lead some to conclude that these two different oscillator types sound different, when in fact they don’t really differ much sonically.
  • Rectangular and Square Waves: These oscillators also have a master controller that enables the pulsewidth of the “Pulse” waveform (the rectangular waveform) to be adjusted, and this can be used to create an extremely accurate square wave (with virtually no even harmonic contamination), and it was found that it was easier to get closer to (if not exactly to) obtaining perfect square waves for all three oscillators at once compared with the 901-B case.
  • Triangle Waves: The triangle waves for these oscillators a bit less even harmonic content – around 6 dB less relative to the 901-B case, which does make them sound significantly closer to true triangle waves.
Figure Caption: Spectrum of a 440 Hz triangle wave for a 921B oscillator, showing even harmonic content, about 6 dB less (relatively speaking) than the 901-B case.

Conclusions about the 904-A Voltage Controlled Filters (VCFs):

The Minimoog VCF in general exhibits both significant 2nd and 3rd harmonic generation, and does so very consistently, due to the (more or less) relatively high input level and the asymmetry of the regeneration amplifier in the VCF.

  • First, the upper 904-A VCF does not exhibit self-oscillation at very high Regeneration (Emphasis) levels, which is known to be consistent with how these filters were originally shipped by R.A. Moog Inc. The lower filter does self-oscillate – it was either altered or of later manufacture – but only at relatively high Cutoff settings – which is different from the Minimoog. The feedback loop for emphasis blocks low frequency feedback.
  • Secondly, when a pure (externally generated) 432 Hz sine wave was input to the VCFs, with Regeneration level set to zero, the first (upper) VCF exhibited some clear 2nd harmonic generation, which was maximum at a Cutoff setting of zero, while the second (lower) VCF exhibited some clear 3rd harmonic generation, which was essentially independent of Cutoff as long as the Cutoff setting was zero or higher. This difference between the two is a bit puzzling, but it appears from this, and various additional listening tests with various patches that were carried out, that the first VCF is perhaps defective, while the second is working more or less normally. This conclusion is consistent with the fact the 904-A VCF has a more symmetrical regeneration amplifier (output buffer) than the Mini, and hence should produce less 2nd harmonic distortion, whereas the ladder filter itself and the regeneration amplifier can be expected on general grounds to exhibit a consistent level of 3rd harmonic distortion.
Figure Caption: Output spectrum of 904-A VCF No. 1 when a 432 Hz pure sine wave was input with the Cutoff set at zero, showing 2nd harmonic generation.
Figure Caption: Output spectrum of 904-A VCF No. 2 when a 432 Hz pure sine wave was input with the Cutoff set at max, showing 3rd and 5th harmonic generation.

Conclusions about the 902 Voltage Controlled Amplifiers (VCAs):

  • Both VCAs were found to exhibit nearly identical 3rd harmonic distortion behavior very similar to that found in the Minimoog, which increases dramatically with the Fixed Control Voltage Setting. This is not surprising as the circuitry for the 902 VCA is quite similar to the MInimoog VCA. There does appear to be one fundamental difference though: The distortion the 902 VCA would appear to occur mainly in the third stage of the VCA, and varies with the loudness control voltage input. This is because the control voltage in the 902 VCA controls the bias current of the second stage of this VCA, which hence controls the input level to the third stage and not the input levels to either the first or second stages. And moreover, the low distortion level at low Fixed Control Voltage settings suggests that the first two stages do not contribute much distortion. In the Minimoog, by contrast, the first stage, especially in the Gen 1 VCAs, can contribute a significant level of distortion which is not dependent on the loudness control voltage. All in all, the characteristics of the 902 VCA appear to be more similar to the Gen 3 version of the Minimoog VCA, in which most of the distortion also comes from the third stage, and which depends on the loudness control voltage.
Figure Caption: Spectrum of the output of one of the 902 VCAs with pure sine wave input, at maximum Fixed Control Voltage Setting, showing significant 3rd and 5th harmonic content.
Figure Caption: Spectrum of the output of one of the 902 VCAs with pure sine wave input, with the Fixed Control Voltage set at just 3, showing minimal harmonic distortion.

Conclusions about the Mixer:

  • The Model 15’s mixer can contribute some 3rd and some slight levels of 2nd harmonic distortion if driven sufficiently strongly, as the following figure shows. This distortion depends on the mixer level settings, that is, it increases when these are increased.
Figure Caption: Spectrum of the output of the mixer with four pure sine waves input (all in phase, equal amplitudes), each with an amplitude equal to the raw output amplitude of a sawtooth wave from the 901-B oscillators, and at maximum mixer levels.

Overall Conclusion

The Moog Model 15, and by extension any Moog modular system using the same modules, has oscillator characteristics which are similar to those of the Minimoog, and can also produce distortion effects similar to that in the Minimoog. The inclusion of an active mixer in the signal path also further enhances the potential for distortion.


Special thanks to Timothy Warneck for helpful discussions.