The principles of feeders for organs historically change over time.  As
a first generation, I could put a human organ blower stepping on feeder
bellows bars.  These feeders produced constant pressure from weights on
the bellows.  The human blower could control his efforts and the
airflow according to the music produced by the organist.

A second generation would be bellows feeders driven at constant rate by
a crankshaft, typical in street organs.  The problem with this method
is that it supplies a constant flow rate, without respect to what is
needed.  The cranking speed has to be constant because it also controls
the musical tempo.  A simple solution is a spill valve to release any
surplus air, at the cost of wasted pumping power.

Better in large organs is control of the valves between feeder and
reservoir.  They are opened by strings to the reservoir lid when the
lid goes high and they let any surplus air return from the reservoir
into the feeders.  This improves power economy at low air consumption,
but at a cost of some valve-slamming noise.

As a third generation comes the use of a fan.  A properly designed one
can supply reasonably constant pressure even at varying flow.  The
problem here is that the system is rather costly when you care about
fan noise.  Things like a solid housing, journal bearings rather than
ball bearings, low RPM and consequently rather large diameter fan, or
multiple steps in the fan all mitigate fan noise.  Fan-speed control is
no option, since this affects pressure more than flow, which is not
what you want.  And the inertia of a fan rotor makes it impossible
follow load variations anyway.

I dare to push for a fourth, more high-tech generation, though
seemingly a step back to the origins: namely using cranked feeder
bellows, but now with speed control of the motor, separate from the
roll/book drive.  A sensor on the reservoir lid controls the motor
speed so as to keep the lid at constant height.  There would be little
problem with mechanical inertia, and even a small reservoir can hold up
while the drive accelerates during a loud passage.  But conservatism
presents a big inertia.  After some design effort, I implemented this
principle in my street organ a few years ago, still feeling lonely.  I
found it to be a spectacular improvement with regard to noise level and
power consumption.

See http://www.fonema.se 

Johan Liljencrants,
Stockholm, Sweden