Choosing an Electronic Organ
by Colin Pykett
Published in The Musical Times: January 1987
This version last updated: 16 June 2013
Copyright © C E Pykett 1987-2013
This article first appeared in The Musical Times in January 1987 and it is now reproduced here because of the number of requests received for reprints. The article was written at a non-technical level at the time when some electronic instruments were moving from analogue to digital technology. From today's perspective it therefore has some historical content. Now that all organs are digital, it was superseded by a more recent article entitled Electronic Organs, published in Organists' Review in 1998 and also available in full on this website. This introduced the reader to the updated technology in more detail. The point to bear in mind when reading the present article is that it was written some considerable while ago and no longer represents current practice.
The article was generally well received, though there were the predictable broadsides from the usual minority whose dislike of electronic organs verges on the pathological. The most amusing was by Christopher Kent of Reading university, spuriously adopting the platform of the British Institute of Organ Studies on this occasion, who wrote in the March 1987 issue that " ... most aurally sensitive musicians will find Mr Pykett's article on the purchase of electroniums a total irrelevance ... ". Presumably believing he had invented a new word, he must have been unaware that the Electronium was an instrument similar in appearance to an accordion which was first marketed by Hohner in the 1950's. They were used frequently by entertainment bands and I first came across one being used in a London pub in the 1960's. They were also used for more serious music making, and Stockhausen among others wrote for it. It is strange that a music lecturer apparently did not know any of this. Nobody who gave my article the most cursory reading could possibly have confused the instruments described in it with the Electronium. As Stanley Sadie, the editor of MT, remarked to me at the time, he always liked to give his correspondents enough rope to hang themselves with!
I am writing this article as a professional scientist and an amateur musician, one who has an intimate knowledge of pipeless organ technology coupled with a perhaps unschooled concept of what constitutes a good organ. On the other hand you, the readers, will bring a scholarly critique to bear to the musical aspects of what follows with nevertheless the possibility of being stimulated by some of the engineering considerations that I shall present. At some slight risk of trying to teach grandmother, therefore, I think we complement each other.
Nobody likes a good pipe organ better than I do: it is the instrument par excellence for interpreting the organ repertory and for providing the musical backdrop to Christendom. It is becoming more widely accepted, however, that when circumstances dictate that this ideal instrument cannot be obtained, the pipeless alternative has to be considered. An intriguing account of a decision-making process of this kind appeared in MT . This article will attempt to set out the major criteria by which one can assess an electronic organ.
At first sight advertisements for such instruments are perplexing. Prices for instruments of a similar size and tonal specification can vary by a factor of almost ten. Within this range none admit to the slightest shortcoming; they may claim to be indistinguishable from a pipe organ both from the player's and the listener's viewpoint, and together with other alleged virtues such as freedom from regular maintenance and virtually unlimited life one wonders why a new pipe organ is ever sold now. But many of those who have a wide enough experience of electronic organs would agree that the number of manufacturers actually approaching these ideals can be counted on the fingers of one hand. Thus it is instructive to try and analyze what makes pipe-organ sound so attractive, to see how these desiderata can be simulated by modem electronic technology, and to give pointers to the sort of questions that need to be answered by those selling electronic organs. To make this article useful, each section contains a number of questions that summarize its content and that can be asked of manufacturers. The degree of importance you attach to each question can, of course, only be decided against the background of your own requirements.
Regardless of such issues as extension and borrowing, any musical pipe organ is characterized by a huge number of independent sound sources (the pipes), each separately adjustable for initiation transient, tone-quality and power (the processes of voicing and regulation). The pipes occupy considerable volume in the building, giving significant spatial spread to the sound. Furthermore, such instruments are built by the best craftsmen who have served an apprenticeship during which they have absorbed the traditions and skills handed down over many centuries. Unless the manufacturer of an electronic organ respects and understands these factors and has succeeded in incorporating them to some degree in his product, it must follow that it will suffer as a musical instrument in comparison with its traditional counterpart.
SUBTRACTIVE TONE FORMING
Even today most electronic organs operate on principles developed about 50 years ago. These instruments generate an electrical waveform rich in harmonics for each key pressed, and those harmonics not required by a particular tone are removed by a filter circuit. For obvious reasons this is called the subtractive method of tone forming. The design of the tone-filter circuits is largely responsible for the quality of sound heard when a stop is drawn, and it is this aspect that demands a detailed insight on the part of the manufacturer into the harmonic structure of the corresponding pipe tone. Very little material exists in the scientific and engineering literature to guide this design process, and my own work is the only example known to me . You should therefore ask a salesman:
Tone filters must simulate the variation in tone-quality and loudness across the keyboard that occurs with all pipe ranks. For technical reasons it is impossible to make a single tone filter that can do this; to enable voicing and regulation to be achieved at different parts of the compass it is consequently necessary to use more than one filter circuit for each stop, each one being active over a specific part of the compass. You may find that fewer than four filters per stop (two on the pedals) is too much of a compromise for an instrument with serious musical pretensions. Many of the cheaper instruments use the limiting extreme of only one filter per stop, which imparts effects such as a thin 'sizzly' sound to reeds and strings towards the bass:
TONE GENERATION IN SUBTRACTIVE INSTRUMENTS
The electrical waveforms that feed the tone-filter circuits are derived from the tone generator. A wide range of tone generator integrated circuits ('microchips') are available in the electronic music industry today which considerably simplify the construction of an organ. However, almost without exception they are made to meet the demands of the large-volume, low-cost entertainment market and do not always produce genuinely musical instruments. On the other hand, their presence does not necessarily mean that an inferior performance will result. A more important issue is that of obsolescence if devices of this nature have been incorporated, since it is not unknown for chips to appear on the market and then vanish within a year. You should therefore ask:
Many such chips produce a quasi-equally tempered scale ('quasi' because production economies usually result in the appearance of perfect 5ths in all octaves). Even if this defect it not present, you may wish to specify a particular temperament to which the instrument should be tuned:
In poor tone-generating systems all octaves may be mathematically locked in tune. This can be most tiring to the car. Manufacturers realize this and sometimes provide a system in which adjacent octaves are 'unlocked'. If this feature is advertised you should nevertheless listen for locking occurring between adjacent-but-one octaves:
In investigating any tone-generator system, regardless of whether it uses microchips, the question of borrowing and extension should be pursued with as much zeal as with a pipe-organ builder. It is tempting for a manufacturer, if he thinks the market will stand it, to feed all the stops of a complete department (or even the entire organ) from a single set of generators. There are two effects to take into account in these circumstances. First, mutations cannot be tuned true; they can only be borrowed from what will probably be an equally tempered scale. The beats that will occur between a twelfth sounding stop and unison pitch may be tolerable but Tierces will probably be unacceptably bad. Second, stops when drawn together will not always produce the subjectively expected effects. Adding a Gedackt to a Salicional, for example, would produce a composite ‘sound’ in which both components would have lost their identity. Doing the same thing on a pipe organ would almost certainly reveal the slight out-of-tuneness that helps the ear retain a mental image of the two tones. With large tonal build-ups, the effect becomes progressively more important:
A cost-effective method of approximating to a large number of tone-generator ranks is to use what is known in the trade as ‘chorus’. This simulates an out-of-tune effect between stops that in fact share a common tone generator. It is not as effective as an equivalent number of real tone generators but it nevertheless can be acceptable. You must decide whether it is satisfactory for your application, but it is legitimate to ask:
In most tone-generating systems the electrical oscillators run continuously, but only when a key is pressed with a stop drawn are they allowed to drive the loudspeakers of the installation. Some instruments are poorly designed so that a residual background noise can be heard even when it is not being played. This is usually most noticeable with all the stops drawn and you should always test an electronic organ for this shortcoming. (The low-level breakthrough that can sometimes be heard, consisting of many different pitches simultaneously, is picturesquely called 'beehive' in the trade.)
The subject of tone generation has not been exhausted. One aspect not yet considered is that of the attack and decay characteristics of the tones. The reputation of early electronic organs suffered badly (and rightly) from their explosive or instantaneous attack. You should listen carefully for clicks and plops when pressing the keys, apart from appraising whether the attack and decay meets with general approval. There is no longer any excuse for inadequacies in this area:
A related but more subtle issue is that of deliberately-introduced starting transient such as chiff. For most electronic organs the introduction of chiff to a particular stop is more involved than adding another stop and it can therefore be expensive in relation to the rest of the instrument. Questions that should be asked if you desire this feature include:
The best answers to questions 4 - 9 above, in a musical sense, will normally be given by a manufacturer able to provide a separate generator rank for every stop, each rank consisting of the appropriate number of electrical oscillators. The term 'free-phase' is used to describe an instrument of this type which, because it has as many oscillators as pipes in a pipe organ of identical specification, tends to be expensive. It also needs regular tuning and because of the large number of components it may not be as reliable as a cheaper alternative.
ADDITIVE TONE FORMING
The subtractive method of tone forming used in the instruments discussed above is characterized by a generator waveform containing very many harmonics, some of which are then removed by a tone filter. Additive tone forming consists of building up the required sound, harmonic by harmonic, from a number of pure tones; thus the tone generator in this case consists of a set of oscillators producing the necessary number of pure tones. The Compton Electrone of the 1950's, with its rotating electrostatic generators, was an additive instrument. In principle it is possible to achieve more accurate imitation of a particular organ sound using additive synthesis, but only if sufficient harmonics (pure tones) are available. For the lower notes of a pedal reed, for example, over a hundred harmonics may well be necessary. Until the advent of the so-called computer organ (described later), this factor prevented additive instruments achieving much success, the practical limitations on the number of harmonics tending to produce a smooth, cloying sound immediately identifiable as 'electronic' and with insufficient bite to the reeds and mixture work. There are few additive instruments of the traditional sort around today, but should you encounter one then the questions posed previously can be paraphrased:
Just as with the subtractive instrument, it is necessary to make provision for varying the tone-quality of a stop across the compass. In an additive instrument this can be done by changing the harmonic proportions, ideally on a note-by-note basis. Normally this would be uneconomic, however, and one would expect to see 'voicing points' across the keyboard at which the required changes of volume and timbre are introduced.
All the electronic organs described above use what is known as 'analogue' or linear circuit techniques. In recent years, using the electronics technology developed for computers, a newer type of instrument has appeared which uses computer-type digital memory devices or even entire microcomputer systems for storing and reproducing sounds. Not surprisingly, they are called digital computer organs. (In using this term I am only considering those instruments which use computer-type technology for generating the actual sounds, rather than including those analogue instruments which merely use computer control for optional extras such as pre-settable pistons.)
A good deal of commercial secrecy cloaks the latest developments in this area but much can be gleaned from the patent literature. Thus there are two basic types of digital organ. The older does not incorporate a computer as such, but it uses such computer-related devices as memory chips. The sound of each stop is stored as a set of numbers (binary digits or 'bits') in the memory so that, on drawing a stop and pressing a key, that waveform is turned back into an analogue electrical signal which can then be applied to a loudspeaker. The later instruments incorporate one or more microprocessors, similar if not identical with those in a home computer, which endow the instrument with more versatility; for example, the C and C sharp sides of a pipe organ can be imitated by feeding the corresponding notes into separate loudspeakers - quite difficult on an analogue instrument. Furthermore, these recent instruments store the sound of each stop as a collection of numbers representing the strength of each harmonic rather than as the waveform itself. At first sight this is perplexing, but one reason for it is that voicing is made easier (and doubtless another is that it enabled the patents of some earlier systems to be bypassed).
Manufacturers offering the second and newer scheme are more numerous, at least in the UK, than those marketing the older system. This may be related to the fact that one system was developed at Bradford University under the auspices of the National Research and Development Corporation (as it was then) who subsequently offered licences for commercial exploitation. Since the sounds are stored as a set of harmonics, it is clear that this represents a form of additive synthesis. However, modern computer technology does away with the limitations placed on the number of harmonics that plagued earlier analogue additive organs and a capability of synthesizing tones from several hundred harmonics is theoretically possible. One feature associated with both types of computer organ is that very accurate reconstruction of sounds is possible, in principle, since the sounds from a real pipe organ can be encoded and stored in the memory. In practice the ability of a manufacturer to do this must be questioned since he may have obtained the digital data representing the stored sounds from another source. Just as with a pipe-organ builder who is expected to be able to voice the pipes that he may have purchased elsewhere, you should ask:
A number of other questions related to voicing and tone generation, comparable with those posed previously, are also legitimate:
There are further questions that are specific to the digital organ. One concerns the accuracy with which information can be stored in the memory and hence the accuracy with which it can be turned back into ordinary electrical signals to drive loudspeakers. All the stored waveforms are represented by sequences of numbers (‘binary words’) and these have to pass through a digital-to-analogue converter before they can be amplified and fed to the speakers to produce sound. If insufficient bits are available for each word, spurious harmonics will appear in the final sounds. That will be most noticeable with those stops with few intrinsic harmonics, such as flutes, which may be audibly impure especially in the lower parts of the compass. Another effect may be a gritty sound on the lower notes of loud reeds. These shortcomings will start to become more noticeable when a digital number representation using fewer than 12 bits is used:
A question that also relates to the number of bits, though in a different way, concerns the amount of memory available to store each waveform. Limitations here will cause corresponding limitations to the number of harmonics that can be used to synthesize a particular stop. We have already noted that the storage of very large numbers of harmonics is one of the theoretical advantages of a computer organ, so it is a pity if this advantage is not capitalized on in practice:
The issue of special-purpose microchips is particularly important with computer organs. Not only will they use computer-type chips which are constantly being upgraded and replaced with newer types, but some instruments might use chips specially designed for it and only available from a few sources. Thus you must get an answer to this question if you are interested in keeping an instrument for a reasonable period:
The spatial spread of sound from a pipe organ has already been mentioned and it follows that an electronic instrument ought to use as many loudspeakers as possible to imitate this. Multiple loudspeakers will also allow separate units to be used for different stops, and possibly for different parts of the same stop. A multiplicity of loudspeakers will also reduce cross-modulation and other distortions which arise if too many signal sources are fed into too few units. Organs currently available range from those with one or two loudspeakers in the console to those by manufacturers who collaborate with the architect of a building in the incorporation of large pedal speakers at the other. The choice is dictated by economics and the environment in which the organ is to speak.
When trying an organ, one should listen for background noise. ‘Beehive’ has been mentioned in the context of instruments using subtractive tone forming though it can arise in any other type. Other objectionable noises include clicks when the pistons or stops are operated. In the latter case some instruments are quiet when the stopkeys or drawstops are manipulated without notes sounding whereas clicks are heard if notes are being played. Such defects are entirely the result of poor engineering and should not occur even in the cheapest instruments. However, it is probably unreasonable to insist on a silent background from the loudspeakers when the instrument is not being played. Low-level hiss, just audible in a quiet building, would generally be acceptable and comparable in any case to escaping wind from a pipe organ. Hum at the mains frequency should not be tolerated under any circumstances since it causes unpleasant beats with quiet pedal stops (notably around GG in the UK with 50 Hertz mains). Finally, it helps to reduce palpitations in an elderly congregation if loud thuds and cracks do not occur when switching the instrument on or off.
The Swell pedals should not control the volume over an unrealistically large range and they should incorporate a tone-control effect so that higher frequencies are attenuated more rapidly than low ones. This should be arranged so that there is a tolerable imitation of the highly critical control obtained with a real Swell box near the fully closed position. It is also not difficult to incorporate means of 'slugging' the action of a Swell pedal so that, no matter how rapidly it is operated, the loudness changes at a rate compatible with the speed at which heavy shutters could move in a pipe organ.
1. D. Parkes: 'The Organ at St Nicholas's Durham', MT, cxxvii (1986), 165
2. C. Pykett: 'Tone Filters for Electronic Organs', Wireless World, lxxxvi(1980), 1537, 1539