Dividers or Free Phase?
Although the choice of commercially made instruments is enormous, many enthusiasts prefer to construct rather than buy an organ.
Differing musical tastes and needs are such that no two constructors will have precisely the same specification in mind. Consequently, discussions on design tend to be endless; indeed, for this reason many self-designed organs never get further than the drawing board! One topic that is a hardy perennial is whether to employ bistable frequency dividers or turn to free phase methods of tone generation.
A large proportion of commercial organs are based on bistable dividers in one form or another. The advantages of this form of frequency divider are that each stage is identical, the system is highly reliable and it is relatively cheap to manufacture. Figure 1 shows a typical bistable stage which, if setup on a breadboard and fed with a square wave, will produce f/2 at its output. The circuit is ideally suited to fabricating several stages on one chip, the SAM 77 CMOS seven-stage package being typical.
The output is normally a square wave but the mark-space ratio can be changed by unbalancing the crossover components; employing a network at the output to modify the waveform will also alter harmonic content.
Several bistable divider stages are cascaded, the first stage being fed with a master frequency and successive octaves below being derived from the chain. The Top Octave Synthesiser outmoded this system in finding 12 or 13 correctly spaced semitones to feed into the divider strings and reducing the tuning requirement to one overall adjustment. This topic was covered in the May 1981 edition of E&MM.
The current method of TOS and divider strings therefore employs only one true oscillator, the remainder of the system being binary switching. The single tuning adjustment has given us even greater reliability, particularly in cases where the organ is frequently carted about by its owner. This feature can also be very useful if faced with playing with another instrument whose pitch cannot easily be shifted.
The divider stages are often encapsulated with switching circuitry so that the keyswitches handle DC rather than audio signals, thus reducing 'beehive' (the general hash resulting from capacitive pickup from wiring in a tight harness).
The square wave contains odd harmonics but is lacking in even harmonics. Normal practice is to amend the waveform by one of the methods mentioned previously.
Another approach is to mix the outputs of several divider stages resistively, using resistor values roughly proportional to frequency. This will result in a step-waveform or staircase and both the sound and appearance on an oscilloscope are similar to a sawtooth waveform. Adding the second and fourth harmonics to the fundamental tone provides what was missing previously — the even harmonics.
For the drawbar organ enthusiast, adding harmonics may not be the answer. He may prefer to try to eliminate everything but the fundamental. The square wave can be processed by a low pass filter, which effectively knocks off its sharp corners and gives an approximate sine wave. Mixing the filtered outputs either by fixed resistors or drawbars will provide the basis for harmonic synthesis, in direct contrast to using a complex waveform and various stop filters to obtain subtractive synthesis.
There are many variations on this general theme used in commercial instruments. The advantages have already been mentioned so let's examine the other side of the coin.
Cascading the frequency dividers results in complete phase-locking between octave-related tones. The TOS itself is also locked in phase where its various outputs are concerned as these are also derived by frequency division, even if of a more complex sort. Consequently, it can be argued that octaves sound 'dead' and that the tuning is in some ways too precise.
Compared with a pipe organ with several ranks of pipes, the divider organ with a single generator set may be seen as an extension organ. The borrowing that must take place to cover the stop system, including possibly mutations, cannot add any richness to the result. This objection may be valid, but, after the signals have been keyed, filtered and amplified, reverberation is normally applied. Quite apart from adding reality generally, the various spring reflections or bucket-brigade delays tend to counteract the strict phasing of divider systems.
Some organs feature an 'ensemble' tab; in this instance applying to the facility to multiply voices. This dedicated form of delay-line will, of course, totally destroy the phase-lock effect.
This type of generator is ideal for the critical organist who is used to playing an acoustic instrument. Playing a chord will result in random phase-relationships between the various sound sources.
It should also be mentioned that, however well it is maintained, no pipe organ is ever precisely in tune with itself due to changes in temperature, wind pressure and other factors.
A simple extension organ (with one rank of pipes to cover the complete compass of the stop array) gives a good chorus effect but the average pipe instrument will have several ranks from which to choose tonal effects. Playing a three-note chord with three stops drawn will bring nine pipes into play, for example.
Tiny random tuning errors and phase differences make the pipe organ sound majestic and brilliantly exciting in the upper register — an effect which is not particularly easy to reproduce by electronic means. Several divider generators can be used, each feeding a section of the stop tabs' associated filters. A Leslie speaker on Chorale, in association with a fixed speaker, will give a form of pseudo-chorus, or, various bands of the frequency spectrum can be modulated by independent, slowly-moving sine oscillators. The best approach, which is closest to pipe-organ analogy, is a free phase generator system. This consists of a set of totally independent oscillators tuned to each note of the instrument's compass. Tuning this type of instrument is, of course, a major operation as up to 100 separate tunings may be required.
Tuning can follow the normal practice of piano and organ tuners in that octaves are deliberately slightly mistuned to add brilliance.
A typical free phase oscillator is shown in Figure 2. It can be seen that a small time-constant can be added to retard attack and provide a small delay: the supply voltage is controlled by the keyswitch and the delay should vary according to 'length' of pipe. Large pipes take longer to speak than smaller ones and an arpeggio on a pipe instrument with 16', 8' and 4' stops drawn will prove that the 16' pipe may not have time to speak.
The greatest advantage of free phase, however, is that the various oscillators keyed will sound in random phase, and octave related sounds in particular will be greatly improved.
The disadvantages? Circuitry is not completely repetitive and a coil winder becomes a necessity. The lower notes of the compass will require large inductors and the instrument will generally be less portable.
For the serious musician, however, free phase offers many advantages. Several manufacturers (Conn, for example) consider this system well worthwhile.
There are, of course, other systems of tone generation including computer methods. As far as divider and free phase organs are concerned, I recommend listening critically to both systems for the differences that can be detected!
Feature by Ken Lenton-Smith
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!