It's clearly out of the question for anyone working part time just for fun even to contemplate trying to reproduce one of the great fairground organs. My intention was never to do that. What did appeal was to try to simulate or replicate the effect of one of those instruments but by using modern techniques and materials as much as I could. This decision is far more one of pragmatism than anything else, because the moment you start to think of the manual skills involved in producing reed and flue pipes similar to those used in the real instruments, you realise that there is very little hope of replicating them without replicating the investment in time and training that the original organ-builders had to make.
My initial research found only scant information about the way that church-organs are built. Fortunately I knew a little already and was also able to pick the brains of one of the organists at Ely cathedral who had formerly been an organ-builder. What I learnt confirmed my view - anything I did would have to use modern materials and it was going to be unrealistic to aim for a copy of the instruments. The best I could hope for was to build a machine that owed a lot to them but was new and distinct with its own personality and features. With modern technology and facilities available, perhaps it could be seen as a continuation of evolution down a different branch. It might be possible to recycle sets of pipes from derelict church organs and if possible I would love to do that, but it couldn't be a fundamental part of the plan.
A clear place that I would differ would be in the playing mechanism. Traditional steam organs use a form of punched-card to play the music, the punched cards being fed into a reader mechanism that detects holes in the cards and then uses air-pressure to drive pneumatic valves that allow air into the pipes and other actuators for the rest of the organ. Given my own skills and background, it was quite clear that I should use MIDI files in place of punched cards and computer-controlled valves or solenoids for the playing mechanism. As soon as I started investigating MIDI software for Linux, I found "timidity", and another piece of the puzzle slotted itself into place. The instrument could easily be a hybrid with some of the music tracks being played for real on the mechanical instrument and with others being synthesized electronically then played through amplifiers and speakers. The more parts of the instrument that were completed the less the electronic synthesis would be needed, allowing progress to be made slowly whilst still getting musical results that would be worth hearing.
Overall the instrument should have a number of playing sections, just like the fairground organ. Different sets of air-blown pipes, whether flue (easy-ish to make) or reed pipes can be used to produce different sounds. Church and fairground organs alike have used this tactic for centuries. Since fairground organs are intended to produce a rousing sound that can be heard for miles, they have a very different remit than their devotional counterparts. Percussion or anything else that makes a pleasing din is welcome and subtlety of the soundscape takes a back-seat.
Stage one of the design was to work on controller software and electronic drivers for the electromechanical parts. Following that, the xylophone would be the first instrument to be made fully playable and the initial experiments on plastic-waste-flutes would probably be the next to be put into full production.