Willis’s ‘Infinite Speed and Gradation’ Swell Control System
by
Colin Pykett
“If
any philosopher had been asked for a definition of infinity, he might have
produced some unintelligible rigmarole, but he would certainly not have been
able to give a definition that had any meaning at all”
Bertrand
Russell
Posted:
14 December 2011
Last
revised: 9 July 2012
Copyright
© C E Pykett 2011-2012
Abstract.
This article describes the 'infinite speed and
gradation' method of controlling swell shutters invented by Henry Willis III and
Aubrey Thompson-Allen in the 1930's. The system was different to any other
in that the amount the swell pedal was moved forwards or backwards from a
spring-loaded central position affected the speed of shutter opening or closure
rather than shutter position itself. Therefore it required organists to
develop a specific technique for controlling it effectively. Beyond the original patent, descriptions of how it
worked are rare and therefore misapprehensions are not uncommon. The most widespread
is that the
system offered a continuously variable speed of shutter movement depending on
how far the swell pedal was moved from its central position, whereas in fact
only five discrete speeds were available for closing the shutters and six for
opening them. Thus despite what its name implies and the claims made for
it, the system offered coarser speed control than most conventional methods.
It
is possible that the systems actually made differed from the descriptions in the
patent, in which several essential aspects were explained inadequately or not
all. For example it was necessary
to brake the motion of the swell shutters suddenly when the swell pedal returned
to its neutral position, yet the means to achieve this were incompletely
described. There was no description
at all of the claim that the swell shutters were accelerated or decelerated automatically
when they were near the fully closed position.
Nor was there any attempt to illustrate the electrical
circuitry involved, or the essential requirement to provide a visual indication
of shutter position at the console. An
effort has been made to remedy these shortcomings in this article. Some
potential improvements are also outlined which would bring reality closer to
original intention, including one which would provide much finer control over
shutter speed.
Whether
the system worked as intended or how reliable it might have been are
questions which are difficult to answer at this remove in time, particularly
as so few working examples now exist. For
instance air leakage would have led to major defects, illustrated by at least two
essential features of the mechanism (rapid braking and holding the shutters
tightly closed) depending on the integrity merely of a flimsy diaphragm which is
subjected to constant flexing. At
several points the patent rather overstated its case with claims which are at best not
self–consistent and at worst simply untrue.
Examples include “a perfect crescendo and diminuendo … as quickly or
slowly as is desired”, whereas in fact the speed of shutter movement was
limited to just a few discrete values. Moreover one is rendered speechless
on learning that
“there is no tendency of the operative parts to bind or stick”!
As it is now unusual to find an organ still fitted
with a system in proper working order, players and perhaps some organ builders might find the
article of some historical interest.
Contents
(click
on the headings below to access the
desired section)
Introduction
What
is ‘infinite speed and gradation’?
Some
personal opinions
Basic
principle of operation
Essential
elaborations
Minimising
overshoot
Holding
the shutters tightly closed
Shutter
acceleration and deceleration near to closed position
Console
shutter position indicator
Circuit
diagram
Potential
improvements
Concluding
Remarks
Notes
and References
Introduction
In
the 1930’s the Willis organ building firm in England, led by the grandson of
the famous founder of the dynasty, introduced a new concept in remote control of
the swell shutters (shades or louvres) of the organ.
Known as ‘infinite speed and gradation’, it was invented by Aubrey
Thompson-Allen who was also responsible for other innovations such as the
electromechanical combination capture system used by the firm in that era.
Thompson-Allen subsequently became a major player in American organ
building when he joined the Aeolian-Skinner Organ Company in 1949, and a firm
bearing his name still exists in the USA.
The
new method of controlling the swell shutters was unique to Willis and a patent
was granted in 1935. As the system
was electropneumatic no wind was required in the console, though it was not
exceptional in this respect. This
was an advantage when the console was detached and a long way from the wind
supply, and essential if the console was to be moveable as well.
In this we can see echoes of the continuing (if inexplicable) enthusiasm
for extreme detachment begun by Robert Hope-Jones in the closing decade of the
nineteenth century. He had incorporated a clever electropneumatic servo mechanism
to control the swell shutters in his first and famous organ at St John’s,
Birkenhead as far back as 1890 [1], and this organ was
playable from outside the church porch, swell shutters and all.
The
reason for this article is historical. Although
the infinite speed and gradation system still exists in a few organs and some of
today’s organists will have experienced it, I suspect the way it actually
works is mysterious to most players and perhaps even to some organ builders.
Its mechanism has almost never been described in the public domain beyond
the original patent, which skims over several important details and omits
others. This might be a reason why
it was often replaced by a more conventional swell control system when the time
came for an overhaul or rebuild. Therefore
this article attempts to throw light on a fascinating, if minor, backwater in
twentieth century English organ building which has almost disappeared into the
mists of time.
Though
I have tried to remain objective, readers might detect a certain exasperation at
times in this article. This is
because I was constrained to rely so heavily on the patent specification, a
document with numerous deficiencies, though this problem is far from unique in the
patent literature.
Therefore I hope they will bear with me as I point these out and, in so
doing, try to present a more balanced representation of a piece of organ
mechanism whose workings have always been something of a mystery to many.
Hereinafter
the phrase ‘infinite speed and gradation’ will sometimes be contracted to
‘ISG’ for convenience in what follows.
What
is ‘infinite speed and gradation’?
The
ISG system is completely different to any other means of controlling the swell
shutters invented either before or since, because it requires the player to
develop an entirely new technique for operating them.
With all other systems the position of the shutters is always related to
that of the swell pedal (shoe) – when the pedal is moved hard over to one
extreme or the other the shutters will be either fully open or fully closed,
with a correspondence being maintained at all intermediate positions.
In the ISG system, however, this correspondence does not exist. One cannot infer the position of the swell shutters from that
of the pedal, and some form of indicator is therefore required at the console to
tell the player where they are. The
pedal itself is also arranged differently to any other in that it is spring
loaded, so that it always returns to an intermediate central or neutral position
when the player’s foot is removed from it, and in that position the shutters
stop moving abruptly (or, at least, they are supposed to).
Away from neutral the position of the pedal determines how fast the
shutters move; they close at a speed chosen by the player when the pedal is
moved towards the vertical by means of heel pressure, and they open when it is
pushed forwards towards the horizontal.
There
were also other more subtle elaborations, though whether they were actually
incorporated in all embodiments of the system is open to doubt.
This is because they were mentioned but not described fully in the
patent, and one is therefore left wondering whether they were successful,
whether they were deemed not worthwhile in practice or whether they introduced
unreliabilities. One elaboration
was that the shutters were said to remain forced tightly closed once they
reached that position, even when the pedal was released and had returned to
neutral. Another was that the rate of shutter movement was allegedly
reduced when they were near to the fully closed position, regardless of whether
they happened to be opening or closing at this instant and regardless of how
fast they happened to be moving previously.
This was said to be because the rate of loudness variation with respect
to shutter position at this point is greatest for any swell box, and this
acceleration-control feature was said to compensate for this perceived
‘defect’. (I rather suspect
that a more prosaic reason for its incorporation was merely to prevent the
shutters audibly slamming shut too often under the control of an inexperienced
player, and Hope-Jones had used an electropneumatic ‘brake’ many years
before expressly for this purpose). Another
problem is that the electrical circuit diagram is not entirely trivial if all
the claimed functionality was to be incorporated, yet this is ignored altogether
in the patent. Nor does the patent
refer to the need for some form of shutter position indicator at the console,
let alone how to implement it.
Some
personal opinions
I
have come across ISG systems from time to time in various Willis III organs, so
my excuse for offering a personal appraisal here is that I have tried quite a
few of them over an extended period starting back in the mid-1960’s. The first encounter
was on the organ at St Andrew’s church at Kingsbury near Wembley, north of
London, on which I practiced often while at university.
Interestingly, this instrument was built in about 1935, the year that the
ISG patent was granted. It was of two manuals with a detached console, though not
detached anything like far enough to have prohibited a mechanical shutter
linkage (it was only a few metres from the pipes). It is described in the National Pipe Organ Register [2],
though I cannot say what its current status is now.
A local anecdote of possible lateral interest was that Robert
Hope-Jones’s brother Frank, the famous horologist, once dived into the nearby
Brent Reservoir (also known as the Welsh Harp) to rescue someone from drowning.
Unfortunately he broke his nose while doing so and this was visible
thereafter on photographs taken later in his life [3].
I
have invariably come away from an ISG-fitted organ with the feeling that its
swell mechanism was somehow soggy and imprecise.
Perhaps this was merely that I never developed the necessary expertise to
control it properly, though I played the instrument at St Andrew’s regularly
over two years or so. Perhaps,
also, the mechanism in this and the other organs I have played was defective.
If so, one wonders whether this was the result of inadequate maintenance
or of fundamental design defects. Whatever
the reasons, I never felt able to execute rapid enough crescendi and diminuendi
when required, although slow variations were easy enough to achieve.
Nor did it seem straightforward to identify the exact moment at which one
should remove one’s foot and allow the swell pedal to return to neutral, so
that the volume remained at the level it had just reached.
Perhaps the best way to sum it up is that the arrangement seemed just too
idiosyncratic and with too much of a life of its own.
Many years later, after having now deduced how it (was supposed to have)
worked, these recollections now seem to be more explicable in terms of its
mechanism.
For
what it is worth, my opinion of the system is that it was to some extent a
solution looking for a problem. It arose from that period of organ building which began in
the reign of Victoria and which sometimes looked like technology gone mad as it
moved towards the theatre organ of the inter-war years.
Because the traditional organ was not immune from this, we therefore find
peculiarities thrown up like ISG. But
the system probably went a step too far. Not
only was it technically complicated, as we shall see shortly, but it demanded
that organists adapt their technique to it in a major way.
In this it was not only a perfect example of the triumph of Machine over
Art, but of the contemporary public appetite for it. Although one could get used to it, I always felt irritated
that a mechanical contraption was deciding for me how fast I was allowed to vary
the volume of the music I was playing. Apart
from relatively few examples it is perhaps unsurprising that it more or less
died not long after it was born, and this tells its own story: it was clearly
not well liked on the whole, otherwise it would have survived longer in the work
not only of Willis but of other builders since.
It was of a piece with Willis III’s legacy in some other respects which
has been fascinatingly analysed by Jonathan Ambrosino in his perceptive review
of some recordings of the Willis organ at Liverpool Cathedral [4].
As mentioned above in the context of St Andrew’s, Kingsbury I can see
little justification for the use of other than a purely mechanical
shutter linkage when circumstances permit.
Etymologically,
I have problems with the mere name. Being particularly unkind, it is a double oxymoron.
‘Infinite’ is a word used glibly and far too often, without understanding and in
quite inappropriate contexts, and this is one of them.
‘Infinite speed’ is meaningless here, as is ‘infinite gradation’.
We shall see presently that the system as described in the patent only
offered five or six discrete speeds of shutter movement depending on whether
they were closing or opening respectively, far from infinite either in absolute
terms or in terms of the gradation between speeds.
In other contexts the similar label ‘infinitely variable’ is often
used incorrectly to describe, say, the volume control on an audio system when in
fact what is meant is that it is continuously rather than discretely variable.
Infinity does not come into it. But
we cannot fall back on this common inaccuracy of usage to excuse the promotional
zeal of the firm in their choice of title, because the system does
not even offer a continuously variable speed of shutter movement, let alone
infinite, and therein lies one of its main disadvantages. It only offers a
few discrete speeds. I have never seen this mentioned by those who have promoted its virtues,
perhaps because they did not realise that the shortcoming existed.
This is another reason for making clear later in this article how it
actually worked. In theory ISG
allowed for relatively fine control of shutter position at the expense of coarse
control of shutter speed, the reverse of more conventional systems which offered
a number of discrete fixed positions but retained better control over speed.
Neither is a perfect solution to the remote control problem.
A
somewhat bizarre way in which ISG reflected the novelty-dominated ethos of the
times was its fairly obvious parallel with the motor vehicle, and one wonders
whether this was unconscious or deliberate.
It is almost as though the organist was being encouraged to ‘drive’
the instrument. For example, the
accelerator of a car functions in a similar manner to that of the ISG swell
pedal at least as far as its forward movement is concerned – the further you
press the pedal of either the faster it goes.
Also both cars and ISG can be put into ‘neutral’, the word used in
the patent, and this is revealing in itself.
Then there is a sort of ‘kickdown’ aspect (my portrayal) which keeps
the shutters tightly closed when required, qualitatively similar to the kickdown
feature of the automatic gearboxes which were under development in the 1930’s.
Cars are exciting (to some), and in the heyday of the deus-ex-machina
organ it would not have been surprising to find some designers who reflected
these engineering enthusiasms in the instrument itself.
Yet another quirky parallel with the car was the use of actual dashboard
fuel gauges at the consoles of some Willis organs to indicate the position of
the swell shutters.
Basic
principle of operation
We
have noted that the
system was patented in 1935 by Willis and Thompson-Allen [5],
though the words ‘infinite speed and gradation’ do not appear in the
specification. This name was
presumably dreamed up later as part of a marketing strategy. Beyond the patent itself the only other place I have found
the system described at a technical level in the public domain was in one of
Noel Bonavia-Hunt’s books [6].
Unfortunately he obviously found it either too boring or incomprehensible
because he terminated his attempted explanation of it abruptly after a paragraph
or two, but not before stating that its modus operandi is something “the
reader may digest at his leisure”. I
found this as exasperating as that common tendency in university textbooks to
leave far too much as “an exercise for the student”!
Interestingly though, Bonavia-Hunt also identified a similarity with the
motor vehicle of his era (the 1940’s).
The
patent exposes the usual problems often found – it mentions some features
without describing them adequately or in some cases not at all, and ignores
others altogether. Moreover the
quality of the diagrams leaves something to be desired, and this difficulty was
amplified when trying to reproduce them here.
In common with most inventions, it is virtually certain that
modifications would have been made to effect subsequent improvements in
performance or reliability, though it would not be surprising if knowledge of
these has now all but vanished. Referring
to Figure 1 which shows the main elements of the mechanism in longitudinal
section, the swell shutters are operated by a single large pneumatic motor
enclosed in a chest 12. The electropneumatic means for controlling the motor are not
visible here and they will be described presently.
Basically,
the bottom board 11 of the motor moves the connecting rod 15 up or down, and the
pivoted swell shutters (not shown) are moved via the trace 19a and the lever or
crank 19. To my mind this
arrangement looks of doubtful viability because, as drawn, it is difficult to
see how the angular movement of the lever could be anything other than
(impractically?) small. Otherwise
the connecting rod would be subjected to significant lateral thrusts which would
throw strain onto the “packed guide” 16 (as the patent describes it).
Assuming the guide continued to do its job however, the rod would then
perforce be bent sideways over some part of its travel.
Depending on exactly how the rod was attached to the bottom motor board,
this itself could also be deflected out of parallel with the top one.
Both objections could be overcome by including some form of additional
swivel joint between the motor board and the rod rather than simply using the
two nuts drawn in the patent, and one wonders whether this was done in practice.
The requirement, which is straightforward, is for a properly constructed
crank to convert the linear motion of the motor into the rotary motion of the
lever, and attempting to do without it is akin to not providing the connecting
rods in an internal combustion engine with little-end bearings.
Figure
1. Longitudinal section of main
mechanism
The
connecting rod 15 passes through a second chamber 14 and a flexible diaphragm
18
before reaching the lever. It will
be shown later that preventing air leakage from both chambers (12 and 14) is
essential if the system is to work properly, otherwise it will be at the mercy
of the momentum of the heavy shutters which could continue to move and overshoot
their desired position when the pedal is released and returns to its neutral
position. Achieving the necessary
sudden braking function at the neutral position is an explicit design feature of
the system which is described in the patent, and it will be illustrated in more
detail presently. Without such a
brake the system would suffer from this major defect (overshoot) to the extent
it would be unfit for purpose. Not
only would there be the coarse control over shutter speed which is an
unfortunate feature of the system, but almost none over shutter position. That preventing this problem relied on the integrity of so
flimsy an item as diaphragm 18 was another reason causing me to question whether
we see reality in this patent. If
the “packed guide” 16 also became distorted as a result of the faulty crank
design mentioned above, air leakage into chamber 12 could also occur, leading to
the same result. Besides faulty
braking, another defect introduced by a leaky diaphragm would be the inability
to hold the swell shutters tightly closed, and this will be described later.
But
as an antidote to all this criticism, let us now look at the clever and novel
aspect of the design. To
open the swell shutters the connecting rod moves upwards as a consequence of pressure
wind entering chamber 12 through one or more of the six orifices 21.
These are of graded diameters and they are opened successively and
cumulatively as the swell pedal is pushed forwards towards the horizontal. The more orifices are opened to wind, the faster the motor
moves. Conversely, the shutters are
closed by injecting wind into chamber 14 through the five (not six) orifices
22
when the pedal is moved in the opposite direction.
The sixth and smallest orifice here (54) is used for another purpose
described later. Therefore in the
implementation as depicted, there are six discrete gradations of speed available
when opening the shutters but only five when closing them.
That the speed control was apparently so coarse, plus the fact it was
even coarser for shutter closure, are issues not mentioned in the patent. Nor is any discussion at all given concerning the absolute or
relative diameters of the orifices 21 and 22, though clearly these are vital
design parameters affecting optimal operation of the system.
In practice they would need to be chosen in conjunction with the wind
pressure in use, again not mentioned in the patent.
Figure
2. Lateral (transverse) section of
main mechanism
The
means for admitting wind into the orifices 21 and 22 is shown in Figure 2, which
is a lateral section of the same assembly depicted in Figure 1 but with the
addition of the electropneumatic valves controlling them.
One such can be seen at the bottom of the diagram but unfortunately the
drawing is rather obscure, so an attempt has been made to zoom in on the valve
subassembly alone in Figure 3.
There
is one such valve subassembly for each orifice, thus twelve in all, and the
design feature of most importance (besides the graded diameters of the orifices)
is that each one is isolated both from the wind supply and the atmosphere when
its electromagnet is not energised. This essential attribute is not mentioned in the
patent, but without it the shutters would overshoot the desired position when
the swell pedal was released. Again,
this is the problem mentioned above, and it is yet another aspect of preventing
air leakage from the chambers 12 and 14 (Figure 1) when the pedal is
at neutral. Air is admitted to the
respective orifice duct 25 via a purse valve 23, which is opened to pressure
wind in chamber 24 when the electromagnet 35a is energised.
However when the electromagnet is not energised, the orifice must not be
in communication either with pressure wind or the atmosphere, which the
arrangement in Figure 3 satisfies.
No
indication is shown of how the graded orifice diameters were realised.
Perhaps this was done using inserts of different sizes.
Figure
3. Expanded view of
electropneumatic control valve subassembly
The
electromagnet and its primary pneumatic valves 34 and 35 constitute a
conventional ‘action magnet’ of the type used in many other electropneumatic
organ mechanisms such as key actions. When
the magnet is energised channel 32 is opened to the atmosphere, thus valve
23
also opens by virtue of wind pressure acting on the membrane surface 26. When the magnet is de-energised, channel 32 is recharged with
pressure wind and valve 23 then isolates the orifice duct by moving back under
spring pressure plus any pressure difference existing across the valve.
The
swell pedal itself is shown in Figure 4. A
feature of interest is the flat spring 44 which returns the pedal to its
neutral position when released. Either
side of this it can move as far as the extreme positions shown in the sketch.
The contacts, in the form of narrow blades rather than wires, are
arranged in two splayed sets 36 and 37. These are shown in more detail in Figure 5.
The contact set 37 consists of five blades whose positions are adjusted
using screws 46. These control
shutter closure. The shutters are
opened via the six contacts 36, similarly adjustable.
A common wiper bar 39, which must be supplied with power in a manner not
specified in the patent, activates the various electropneumatic valves
successively and cumulatively as the pedal moves. When the pedal is at neutral the wiper is clear of all
contacts, even though this is not immediately obvious from the representation in
Figure 5. However it can be
discerned better in Figure 4.
Figure
4. Swell pedal and contact
arrangements
Figure
5. Expanded view of swell pedal
contacts
Essential
elaborations
The
foregoing has described the basic operating principles of the system.
However a number of important elaborations are necessary for it to work
as intended and these will now be described.
As mentioned already, not all of these are discussed in the patent.
Minimising
overshoot
One
of the most important and necessary elaborations is the need to minimise
inertial overshoot of the swell shutters beyond the position they had reached at
the point the swell pedal was returned to neutral.
This braking function will now be examined in more detail. It
is only fair to point out that overshoot was indeed discussed in the patent, but
even so some important details were glossed over. Therefore I am grateful
for discussions with Lucien Nunes in arriving at the probable modus operandi of
this aspect of the mechanism. Given the vagueness of the patent however,
it is difficult to be certain whether the description which follows applied to
the ISG systems which were actually made. Although the concept is most likely
sound there were probably some practical issues limiting performance which will
not be debated here. It is not really up
to either Mr Nunes or myself to compensate for the shortcomings of an inadequate
patent specification! Nevertheless, here goes.
It
appears that the means whereby overshoot is minimised is to suddenly isolate
one of
the chambers 12 and 14 in Figure 1 from both the supply of pressure wind
and the atmosphere as soon as the swell pedal returns to neutral.
When this happens further motion of the motor board 11, mainly due to the
inertia of the heavy shutters, ceases rapidly owing to the trapped air on one side of it. One reason why the air becomes trapped is because of the
isolation of all the orifices 21, 22 and 54 in Figure 1 from both pressure wind
and the atmosphere, and this has already been discussed above.
However, to fully appreciate how the system works it is now necessary to
refer to the two valves 47 shown in Figure 1 and their associated
mechanism. The latter is only treated superficially in the patent.
Consider
a situation in which the shutters are opening.
The bottom board 11 of the motor will be moving upwards at a speed
depending on how many of the orifices 21 are admitting air to chamber 12.
However this motion could not take place if wind was trapped in the upper
chamber 14, so to allow it to evacuate to the atmosphere the
spring-loaded purse valve 47 of the upper
chamber opens against the air compressed by the moving motor board. At the
same time, wind entering chamber 12 is prevented from venting through its
valve 47 because the associated tube 50 is charged with
wind. There is therefore a net pressure (largely that of the spring)
keeping valve 47 of the lower chamber shut. But according to Figure
1 both tubes 50 apparently lead
nowhere. In fact the patent
specification implies that they are each controlled by an electropneumatic valve,
nowhere shown nor discussed further, which presumably opens the
tube to the atmosphere when the valve is not electrically energised and charges
it with wind when it is energised. The
design of this valve can only be guessed at, though it need only employ standard
electropneumatic technology of the 1930's. It might simply have been an
'action magnet' alone with no additional pneumatic relay, because the volume of
air required to move through tubes 50 is small if our understanding of
the mechanism is correct.
A similar process takes place when the shutters are closing.
In this case valve 47 in the lower chamber 12 evacuates the trapped air
to the atmosphere, and that in the upper chamber 14 is closed because of
pressure in its associated tube coming from its mystery electropneumatic valve.
There
is also another complication, scarcely hinted at in the patent, and this
concerns the electrical means whereby the mysterious electropneumatic valves are
energised and de-energised. The way
suggested in the patent (though using different terminology) is for the electromagnet
of the valve corresponding to the chamber being fed with wind to be connected to
the output of some form of logic OR gate whose
inputs are the electrical signals from the corresponding contacts of the swell pedal
(Figure 5). In other words, the
electropneumatic valve controlling purse 47 of chamber 12 (Figure 1) will be
energised when any of the
orifices 21 are admitting wind to chamber 12, and similarly for
the valve of chamber 14. In fact an OR gate is not required; all that is necessary is for the
valve to be energised when the smallest orifice is admitting air, because that
will also be the case for any number of enabled orifices. Therefore the electromagnets of the mystery valves could simply
be connected to the first of the contact sets 36 or 37 of the swell pedal.
The patent avoids making clear this simple fact.
Given
this postulated arrangement, when the swell pedal returns to neutral either of
the valves 47 can open against pressure in its corresponding
chamber. However the pressure in the chamber that was under inflation will
likely fall momentarily below that of the atmosphere owing to the momentum of
the shutters attempting to increase the volume of the trapped air.
Therefore the motion of the motor board is arrested more
or less suddenly because valve 47 of this chamber will be held tightly
closed by atmospheric pressure acting on the disc.
Holding
the shutters tightly closed
When
the shutters close completely, a circuit is made by contact 56 which meets
another 55 on the lever arm in Figure 1. One
or other of these is supplied permanently with power.
We are told that this results in wind being admitted to orifice 54 by
means of an electropneumatic valve, presumably similar if not identical with
those depicted in Figure 3 which control the other orifices.
When the swell pedal returns to neutral this circuit will remain active
and therefore pressure wind will continue to be supplied to chamber 14
indefinitely. In turn this will
apply a continuous force to the shutters to keep them tightly closed. However
the circuit must be broken if the shutters are ever to open again, and this is
done by a second, pneumatically operated, switch in series with the same
circuit. This second switch, 57, is
opened whenever pressure wind appears in duct 58, and apparently this occurs
whenever any of the six orifices 21 are supplying wind. It is best
to simply repeat what the patent says at this point:
“In
order to break the circuit through contacts 55, 56 as soon as one of the
orifices 21 is opened, wind is simultaneously admitted to the chamber 51 through
the duct 50, and a small motor 57 is moved by wind admitted through the duct
58
to effect the opening of a switch in the same circuit as the contacts 55,
56
…… ”
As
mentioned previously, the integrity of diaphragm 18 is vital if the mechanism
just described is to work properly. Otherwise
air will bleed through it and sufficient pressure will not be maintained in
chamber 14, thus the shutters will not be held tightly closed.
Shutter
acceleration and deceleration near to closed position
At
the outset, in its third paragraph the patent states that “ … the valves
[are] operated automatically so that the speed of opening or closing of the
shutters, when the opening is considerable, is made to decrease towards the end
of the movement … ”. This is
presumably intended to limit the speed of shutter movement when they are near
the fully closed position to prevent the loudness varying too quickly at this
point. Bonavia-Hunt makes much of
this in his own incomplete and vague description [6].
Among much else over which he enthuses, he states that “Mr Willis
applies to the swell engine an automatic accelerator device which makes the
shutters open at an ever-increasing speed as they approach the fully open stage.
When the shutters are being closed, a decelerator decreases the speed of
the shutter movement as the fully closed position is approached”. Engines, acceleration, deceleration and ever-increasing speed
– oh dear, cars again! How
revealing the contemporary terminology is today.
Unfortunately for us, neither Bonavia-Hunt nor the patent itself give the
slightest hint as to how these interesting phenomena arose from within the
system itself.
In
fact it would be quite easy to achieve, at least in principle, if some of the
various valves used for opening and closing the shutters were controlled, not
only by the position of the swell pedal, but by the instantaneous position of
the shutters as well. It is not incumbent upon me to improve upon an unsatisfactory
patent specification or to remedy Mr Bonavia-Hunt’s obvious lack of
understanding, let alone to posthumously extricate Messrs Willis and
Thompson-Allen from a
minor black hole, but one can postulate the existence of a set of switches
operated mechanically by the mechanism linking the main pneumatic motor to the shutters.
These would each be wired in series with at least some of the swell pedal
contacts and arranged so that the switches were automatically closed as the
shutters opened, and opened as the shutters closed.
By this means the shutters could be accelerated or decelerated near to
the fully closed position as the larger orifices in the sets 21 and 22 were
automatically opened or closed as the linkage moved.
An alternative scheme
could have involved the use of a pneumatic dashpot of suitable characteristics,
though the operation of the electrical system just outlined seems easier to
implement and predict.
Console
shutter position indicator
Bonavia-Hunt
states that “an optical indicator at the console tells the organist and
reveals the position of the shutters” [6].
A necessary function indeed but, somewhat surprisingly, one which is not even
mentioned in the patent itself. I
have already mentioned that in some Willis organs the console indicator was
merely a fuel gauge as used in contemporary cars.
Given this, it would not be at all surprising if the gauge was driven by
the type of sending unit used in the fuel tanks of such vehicles.
These used a float arm coupled to a variable resistor, and the arm could
easily have been coupled mechanically to the swell shutter linkage.
Alternative schemes using filament lamps arranged as a sort of bargraph were also used, illuminated as required by
mechanically-operated switches according to the position of the shutters.
The organ I played at St Andrew's, Kingsbury was so fitted in the 1960's.
Circuit
diagram
No
vestige of a circuit diagram appears in the patent.
The only discussion of the subject is as follows:
“The
electric leads connecting the contacts 36, 37 to the electromagnets for moving
the rods carrying the valves 34, 35, the electromagnets themselves and a source
supplying electric energy to the circuits can be arranged in any suitable manner
and comprise parts of any suitable construction, the wiring of these parts not
presenting any difficulty and not requiring any precautions not usually taken in
electrical installations of a simple and common character so that detailed
illustration of the same is unnecessary”.
At
the risk of being unkind again, it
would be difficult to exceed the breathtaking disingenuousness of this
statement. If it were simply the
case that the swell pedal contacts were connected one to one to their respective
magnets, then it could be accepted. However
it has been shown above that significant additional complication at the
electrical level must exist in the system if it is to include elaborations
including minimising shutter overshoot, holding the shutters tightly closed,
shutter acceleration and deceleration, and provision of a console indicator.
When these are included it is necessary to draw up for oneself, if not to
expose it to public view, a complete circuit diagram.
When compiling this it becomes obvious that the electrical aspects of the
system, while well within the wit of man, are not entirely trivial.
Accordingly a suggested circuit diagram which covers all the aspects just
mentioned is included here at Figure 6. How
similar it might be to the circuit actually used by the Willis firm at the time
is probably a question without an answer, if only because such things are
subject to frequent modification and these are not always documented properly,
if at all. A
detailed description of the circuit will not be given as much of it would only
repeat the various issues discussed above.
However, note especially the vertical dotted line labelled “shutter
linkage”, which passes through all of the components which might be controlled
mechanically in some way by the position of the shutters.
Starting at the top of this line these are:
1.
The switch contacts 55 and 56 in Figure 1 which hold the shutters tightly
closed once they reach the fully closed position. They are released by switch 57 (also in Figure 1) which it
was suggested above could be opened purely pneumatically when the smallest
orifice in the set 21 receives pressure wind.
2.
A set of deceleration control switches controlling all of the orifices 22
except the smallest. In the absence
of any information at all in the patent, it was suggested earlier that these
switches are opened automatically, either simultaneously or successively and
cumulatively, as the shutters approach their closed position.
This would decelerate them regardless of how fast they happened to be
moving previously (unless only the smallest orifice was activated, in which case
they would continue to close slowly).
3.
A variable resistor or potentiometer wired as a potential divider which
sends a variable voltage to an indicator at the console related to the current
shutter position. In
Figure 6 the wiper of the resistor is shown mechanically attached in some way to
the shutter linkage. If the console indicator made use of lamps arranged
as a bargraph, as some
organs did, these could be controlled by a set of switches operated mechanically
by the
linkage.
4.
A set of acceleration control switches controlling all of the orifices 21
except the smallest. It was
suggested earlier that these switches are closed automatically, either
simultaneously or successively and cumulatively, as the shutters move away from
their closed position. This would
accelerate them to a speed corresponding to the number of contacts 36 closed at
the swell pedal (unless only the smallest orifice was activated, in which case
they would continue to open slowly).
Note
also the magnets controlling the two overshoot control valves 47, one in each of
the chambers 12 and 14. One or
other of these is activated whenever any
number of orifices in the set 21 or 22 is admitting wind respectively. This
would allow the
moving motor board to expel air from the chamber associated with the other
valve, the one which is not activated.
When no orifices are admitting wind, i.e. when the swell pedal returns to neutral,
neither valve is activated. As described
above, this brakes the shutters and prevents overshoot beyond the desired
position. These two magnets were
referred to as “mystery” magnets above because they are not shown in the
patent, and both they and their means of control are only referred to in the
vaguest terms in its text.
Figure
6. Possible circuit diagram for the
Infinite Speed and Gradation system
Potential
improvements
It
would not have been difficult to effect improvements to the system so that at
least some of its design defects were overcome, thereby bringing reality closer
to original intention. Only three will be discussed briefly here.
The
first modification removes reliance on the long term airtight integrity of the
flimsy diaphragm 18 in Figure 1. It has been remarked already that
significant air leakage at this point will encourage overshoot as
well as render it impossible to hold the shutters tightly closed, in both cases
when the swell pedal has returned to neutral. The means proposed is to use
two large pneumatic motors coupled together mechanically instead of just one, each motor being responsible for
either closing the shutters or opening them, but not both. The interiors
of both motors would be provided with multiple orifices for air entry as before (either
the set 21, or the set 22 with 54 in Figure 1) plus an exhaust valve
47, but neither motor would
need to be enclosed in an airtight chamber. Most importantly,
diaphragm 18 could be disposed of. Other advantages of using
unenclosed motors would include ease of access for inspection and repair,
although it should be remembered that pneumatic leatherwork tends to last longer
if it is enclosed in some way. However the enclosures would not need to
be airtight (indeed, they must not be in this case).
The
second suggested modification is to provide a proper crank mechanism to convert the linear
motion of the two coupled pneumatic motors into the rotary motion of the shutter operating
lever (if one were to be retained).
The
third modification will increase the number of speeds available, limited to only
5 or 6 in the system as described in the patent. Instead of activating the
various orifices 21 and 22 cumulatively, they could be treated as binary
pneumatic arrays. To explain this, consider the five orifices comprising
set 22 and denote an activated orifice (opened to pressure wind) by the digit '1' and one
not activated by '0'. Then the slowest closure speed would be obtained, as
in the original system, using the activation pattern
00001. This is a five-bit binary number equating to decimal 1. The
next fastest speed would then correspond to decimal 2 which is 00010 in
binary. This means that only the second smallest orifice would be opened
to wind, all others being closed. However this pattern cannot be
obtained using the system described in the patent, in which the next lowest
binary number obtainable after 1 is 00011 (decimal 3). This jump, from 1 to 3
omitting 2, is one of many in the system as described originally, and it explains why speed control is so coarse
. But by using fully developed binary activation, subsequent
patterns after 3 would be 00100 (decimal 4), 00101 (decimal 5) and so on up to 11111 (decimal 31). Thus activation in this manner would enable up to 31 shutter closure
speeds to be obtained using the existing five orifices in set 22, and up to 63
opening speeds for the six orifices in set 21. However, in reality this
takes us from the ridiculous to the sublime - while 5 or 6 speeds are barely enough, 31 or 63 are far too many and some intermediate value would be more
sensible and practical. Fifteen
speeds would be a realistic upper limit, and an interesting aspect of binary
activation is that these could be obtained from only four orifices instead of
the five or six described in the patent. Even just three orifices would still
deliver seven speeds, more than in the implementation originally described. Thus the mechanism could be simplified yet much
finer control of shutter speed would result. A minor downside
of this approach would
be the need for some form of electrical decoding system between the swell pedal
contacts and the orifice control magnets so
that the position of the pedal could be converted into the
equivalent binary pattern. This would be trivial using today's digital technology, though in the 1930's electromechanical decoding using relays would
have been necessary. Similarly easy, this would probably not have presented much of a difficulty to
Thompson-Allen with his evident flair for engineering, but perhaps the reason he
did not do it was simply because binary circuits were almost unknown in those
days outside the exotica of telephone exchanges.
Concluding
Remarks
The
infinite speed and gradation method of controlling swell shutters has been
described mainly by referring to the patent granted to Willis and Thompson-Allen
in 1935. There was not much choice
because, beyond this, little collateral information exists about a system not
always employed even by Willis and (to the best of my knowledge) never by other
builders even after the patent had expired.
It is now rare to find an organ still fitted with ISG, and even then it
is difficult to
be certain that the system is in proper working order.
It
is possible, indeed likely, that the systems actually made differed to some
extent from the descriptions in the patent.
This is because several essential aspects were explained inadequately or
not all. For example it was
necessary to brake the motion of the swell shutters suddenly when the swell
pedal returned to its neutral position, yet the means to achieve this were
incompletely described. There was
no description at all of the claim that the swell shutters were accelerated or
decelerated automatically when they were near the fully closed position.
Nor was there the slightest attempt to illustrate the electrical
circuitry involved, or the essential requirement to provide a visual indication
of shutter position at the console. An
effort has been made to remedy these shortcomings in this article. Also
some potential improvements have been suggested, including one which would
provide far finer control over shutter speed.
Whether
the system actually worked as intended or how reliable it might have been are
questions which seem impossible to answer at this remove in time, particularly
as so few working examples now exist. For
instance air leakage leads to major defects, illustrated by at least two
essential features of the mechanism (rapid braking and holding the shutters
tightly closed) depending on the integrity merely of a flimsy diaphragm which is
subjected to constant flexing. Misapprehensions about how the system
worked are legion, no doubt because at
several points the patent made claims which are, at best, not
self–consistent in that they are unsupported by explanations elsewhere, and at worst
simply untrue. Examples include
“a perfect crescendo and diminuendo … as quickly or slowly as is desired”,
whereas in fact shutter speed is not continuously variable at all - only five
discrete speeds are available for closing the shutters and six for opening them.
Moreover one is rendered speechless on learning that “there is no tendency of the
operative parts to bind or stick”! That
the system has been consigned to little more than a footnote in history is
therefore not surprising - the customer has had the final say.
Notes
and References
1.
Hope-Jones’s swell shutter actions are described in detail in my
article “Robert Hope-Jones: the evolution of his organ actions in Britain from
1889 to 1903”. Available for
download as a PDF file from this website. Clicking here
will give you more information before you decide to download.
2.
See NPOR entry N15779 at:
http://www.npor.org.uk/cgi-bin/Rsearch.cgi?Fn=Rsearch&rec_index=N15779
(Accessed
16 October 2011).
3.
The anecdote concerning Frank Hope-Jones’s heroic exploit was retailed
to me by a correspondent who wishes to remain anonymous.
An article elsewhere on this site describes Frank’s life and work in
more detail. See “The ‘Other’
Hope-Jones”, C E Pykett, 2006 (read)
4.
See http://homepage.mac.com/glarehead/ambrosino/paper-liverpool.html
(Accessed 16 October 2011).
5.
“Improvements relating to Swell Louvre Actions for Organs”, Henry Willis and
Aubrey Thompson-Allen, British patent specification 428,448, accepted 8 May
1935.
6.
“The Modern British Organ”, N Bonavia-Hunt, Weekes, London, 1947.
