This page will detail the progress of my project to build myself a digital pipe organ. I have always loved the pipe organ, and have played for many church services and weddings over the years. I have always, though, found myself hampered by not having an instrument to practice on at home. A proper pipe organ occupies tens of square metres, and would cost around a million pounds to build, so obviously that was out of the question. However, recent advances in computer processing power have meant that it is now theoretically possible to build a high quality digital pipe organ. This page will show my trials and tribulations as I try to realise my dream: a pipe organ at home. I hope it provides some useful pointers and advice in the future to anyone else interested in doing this.
If you are trying to build a digital pipe organ for yourself, please don't hesitate to contact me, and I will be happy to offer whatever assistance I can.
Well, the bench is completed. I changed the design and just put a short piece of 2x4 between the middle of the two long side beams at 90 degrees, rather than have the third long beam down the middle, as it seemed ridiculously over-the-top. I just need to screw the feet into the legs and give it a varnish. Varnishing pine is always a worry - it's a terrible wood for going blotchy. I have ordered some mid-oak coloured water-based pine stain from wood-finishes-direct.com, as I really don't think anything sold in Wickes or Homebase will be any good. It will also need varnishing to protect it. I understand that it's better to use an oil-based finish if you use a water-based stain to avoid picking the stain back up again and streaking it. It turns out it's harder to find oil-based polyurethane than water-based these days. Anyway, I'll post the results here when I'm finished. If it all goes horribly wrong, at least someone else might not make the same mistake! Anyway, here it is as it stands:
Somehow - like an absolute idiot - I managed to mix up centimetres and millimetres when asking the timber yard to cut my plywood. I believe I wanted the length to be 860mm rather than 1100mm! In any event, they're too big. Still - at least you can do something when things are too big - it's a bit more difficult if they're too small, so I suppose I should be grateful for that! I needed a circular saw anyway, so I'll buy one to cut them down with - I want as straight an edge as possible, and I don't think my jigsaw will give me it.
I then marked out the stopped rebates for the plywood bases that the keyboards will rest on. These will be made of 9mm birch plywood. On the two shorter pieces of wood, the rebate is 9mm tall and 259mm from the rear end. On the two longer pieces of wood, the rebate is 9mm tall and 367mm from the rear end. This gives 33mm on the front of each piece with no rebate.
Showing the magnets attached to the pedal ends
The MIDI encoder circuit board from Bulgaria has arrived, so in the next few days I will start work on converting the old pedalboard to give out MIDI signals
Also, I have made one of the big purchases of the build - the four keboards for the manuals! I was hoping to have a bit more time to make a decision on these, but in the end the opportunity to save some money forced my hand somewhat! There were two choices. Firstly: buy a ready-made MIDI keyboard, take the case off, and remount it in wooden key cheeks; or secondly: buy a bare keybed from a manufacturer such as Fatar, who normally sell them to other companies to make their own keyboards out of. The benefit of buying a complete MIDI keyboard is price, and the fact that it needs no additional electronics to make it talk to the computer. The Fatar keyboards would have been a little better quality and mean I didn't have to modify anything before attaching it to wooden key cheeks, but they were about half as expensive again, and would have necessitated the purchase of another MIDI encoder board for each of the keyboards to send signals to the PC. In the end, I decided to buy the ready-made MIDI keyboards - I happened to come across an eBay seller with manufacturer-reconditioned 'B' stock of the "M-Audio Keystation 61es". They normally sell for around £100, but these were on offer for £74.95 each. He only had six left, and I wanted four, so I had to make a decision quickly. I'm going to have to modify them slightly to enable the organ's thumb piston rails to fit under each keyboard and maintain the Royal College of Organist's standard dimensions, but we'll cross that bridge when we get to it! I decided that if I ever wanted to upgrade the keyboards, I would probably want to upgrade to really nice wooden-cored ones, and if I wanted to do that, then I would have wasted money on the more expensive Fatar ones.
As mentioned above, the MIDI encoder circuit board arrived, and that turned out to be pretty easy to get talking to the computer - near enough plug and play. A bit of experimentation showed that the magnets supplied with it for actuating the reed switches were a bit on the weak side, meaning the pedals would have to pretty much touch the reed switches to activate them. Given that the reed switches are made out of delicate glass, this didn't seem the best idea, so I ordered some powerful N51 neodymium magnets from eBay. Here is the circuit board together with the reed switch array that will be mounted under the pedalboard:
While doing this project, I am trying to build up my tool collection. I saw a vintage Stanley No. 4 Bench Plane on eBay for £19, so I bought it, and have spent the last few days restoring it. It's a fabulous tool, back from when they used to make things properly! There is a fabulous video here (click link) on YouTube by the English woodworker Paul Sellers on how to do this. Below are the before and after photos.
I then used a hammer and chisel to square up the corners of the rebate
Most organs have rails under each keyboard with buttons on. These are called "pistons", and are programmable to quickly select certain combinations of stops. The Hauptwerk software is able to recognise MIDI inputs as piston "pushes", and so I need to find a way to get MIDI inputs from buttons into the computer. There are dedicated MIDI boards for this, but it seems silly to spend that sort of money for a few buttons. I have played around with "Arduino" microcontrollers before with my electronics hobby, so I researched whether it was possible to configure one as a MIDI device. It was, but it wasn't simple enough for me! I found a microcontroller called a "Teensy 3.0". This came with a dedicated MIDI library, and it also had the advantage of being programmable through the Arduino development interface, which I had a tiny bit of experience with. I have ordered a board to play around with and see what I can do.
The pin headers have arrived, so I very carefully soldered them onto the board with my finest soldering tip! I also found 3 "push to make" buttons in my parts bin and soldered leads to them to use for prototyping. Obviously, I'll get nicer buttons for the real thing!
I opened Hauptwerk with my now-programmed microcontroller plugged into my laptop, and went through the instructions to register it as a MIDI device. Telling Hauptwerk to accept events as inputs is ridiculously easy - you just right-click on the element you want to control, select "listen for MIDI event", and then press the button you with so control it with. The software does the rest. I selected a couple of pistons to control, and to my delight, my buttons worked! You can see the stops changing on this short video:
I connected the switches between input pins and ground on the Teensy board, using my breadboard, then loaded the Teensyduino Development Environment and typed the following programme in. (The bits of text which are preceded by "//" marks are merely commentary to assist programming - they do not form part of the programme.
I apologise that I haven't updated for a while. I have finally been getting round to turning the garage into a space where I can actually make this thing! I didn't include detailed posts on this, because it isn't really directly relevant to building the organ. Nevertheless, in brief, I cleared and tidied the garage, moved everything I could out of the way into the shed, found an old Record woodworking vice on eBay, and built a workbench to work on.
I still need to build a stand for the table saw I bought on Gumtree, and I think I'll need a router table in order to make the panels I'm starting to think the main organ case will be constructed from.
First things first - I need a pedalboard. There appear to be three options:
My woodworking is terrible, so 1) isn't an option. A ready-built MIDI pedalboard with the full 30 radiating and concave notes is in the region of £1,800, which is nearly my entire build budget, so that's out the window! So, I will find one and convert it.
I googled "Organ Restorers UK", and sent an email to every single organ builder that came up on the first 20 pages of google results, asking if they had any old pedalboards they would be willing to sell. And now we wait...
I have spent the last few days figuring out how I'm going to mount the keyboards in key cheeks. I got fed up trying to draw increasingly confusing things on paper, so decided to take a couple of days out to get to grips with Google Sketchup, which is a free 3D modelling programme (link here: www.sketchup.com) It's not the most intuitive thing in the world, but when you get the hang of it, it makes designing more complicated structures much easier. You can then print out a cut list with all your dimensions on it. This is what I ended up with:
As I don't have a router table at the moment, I made a couple of jigs to assist with this. Using my 9mm straight bit in my router, I made a cut in a piece of thin hardboard while it was held against a straight edge which showed exactly what distance the router cut from the side of its base. I then used this to work out where a fence would need to be to line the cut up with the edge of my rebate marks. Since all the pieces are exactly 64mm wide, this will be exactly the same for each piece of wood. I glued a straight edge to a scrap piece of 3x2, using my hardboard router reference as a guide. When these two pieces were clamped in the Workmate, with an extra piece of wood under them for support, this gave me a guide to run my router against. It was a bit crude, but it worked! I used the depth stop on the router to keep the depth of the rebates consistent over the four pieces of wood. The depth isn't critical - it just needs to be sufficient to screw the plywood into
Having had the opportunity to play with the keyboards I bought, I've had a slight rethink. Yes, a 4-manual instrument would be fantastic to have, but this is an instrument to practise on. How often am I playing a 4-manual cathedral organ in real life? Not very often! Plus, if I want that, then at some point, I will buy ready-made organ keyboards. They're very expensive - about £1,000 a manual for good ones - but it seems silly to make such a huge practice instrument out of cheap keyboards. I have therefore decided to proceed with only 2 manuals. If I want to upgrade later, I always can. This should significantly limit the expense, too.
I have also purchased some pine to practice making the key cheeks before I cut into my oak!
Having read around the subject, I decided I wanted to build the best organ I could for the least amount of money. That might sound an obvious statement, but It's possible to spend almost nothing on a project like this (playing the free, limited version of the software with an old MIDI keyboard plugged into your laptop, for example), or to spend tens of thousands of pounds (buying beautiful wooden keyboards, a top-of-the-range, dedicated PC to run it, making a custom console out of hardwoods etc.).
I want my organ to look like an organ rather than a motley collection of keyboards, but equally, I'm not bothered by stunning design. Although, as a pianist, when playing the piano, I am a stickler for a perfect key action, I am not going to throw money at the keyboards in the first instance. I'm going to buy reasonably cheap MIDI keyboards, to prove that this will work, before spending an awful lot on anything better. I believe they will hold a decent resale value, and anyway, it never hurts to have a few keyboards lying about the place as a musician! I'm hoping that in the end, my build will come to £2,000 or less. That's a lot of money, but incredible value when you consider that a comparable off-the-shelf electronic organ would be in the region of £35,000 to £50,000 or more.
I have been making an assembly to hold the reed switches in place beneath the pedalboard. Given that the pedalboard is concave, and the switches need to be a constant distance from the pedals, I thought this would be easier to make in sections. Below is a picture of the first section on its own, and another picture showing how it will be mounted to the pedalboard. In the photo on the right, the pedalboard is on its end, with the ends of the pedals at the top. The L-bracket I have added to the base of my switch holder will have velcro on it, and this will attach to the base of the pedalboard, which will keep it there by virtue of its own weight when placed flat on the ground. Call me sentimental, but I didn't like the idea of drilling into beautiful old oak unless it was absolutely necessary! And, this way, it allows for slight adjustments as and when necessary to ensure everything lines up as well as possible. You can see the grooves I routed to hold the reed switches, and the magnets will be placed on the end of the pedals, to activate the switches when the pedals are depressed.
I will be giving the bench some coats of polyurethane to protect it and letting it cure for a month before rubbing it out to a satin finish with steel wool and paste wax. The next job is to remove the MIDI keyboards I have bought from their plastic casing, and think about how to make some wooden key cheeks for them.
Well, Christmas and New Year have been and gone. I have been working on removing the old electronic contact system from the pedalboard and generally cleaning it up. There is a lot of grime on it, and it is tricky not to take the varnish off with it, so it's rather slow going. I have also started work on the organ bench, basing the plans loosely on the bench on this website (which is a clickable link). I'm awaiting the arrival of the printed circuit board I ordered from www.midiboutique.com in Bulgaria so I can start modifying the pedalboard to give MIDI signals to a computer, as sadly the contacts were not in a good enough condition to use - I think they would have given too much "switch bounce" to reliably turn on and off a MIDI signal.
The circuit board I have ordered will allow me to use magnets on the bottom of each pedal to turn on a reed switch when the pedal is pressed. When the board detects a switch activated, it will send the relevant MIDI signal to the PC. Anyway, below is the bench as it "stands" at the moment. It still needs screwing together, and planks for the top need gluing, shaping and attaching.
I thought it was time to bite the bullet and start cutting the keys - they have a front which is quite deep which would interfere with a piston rail. I unscrewed the bottom of the case from the top, but left two screws in, which kept the keyboard secure in the base. This was to assist with stability while I worked on it. There were three square cable sockets which needed unplugging in order to completely remove the top, as there is a separate circuit board for the modulation wheel and volume controls. It was a simple matter of unplugging them before completely removing the top.
Once the keyboard was open, it was easy to see the springs attached to each key. The easiest method I found to remove these was poking a small screwdriver down them, and flicking them out of the channel in the back of each key. Then, I removed each white key by gently pulling up on the back of the key, where shown, while also pulling reasonably firmly towards myself.
I've finally got a new circular saw. I'm going to have to make some sort of sacrificial table top to cut on, I think. The blade is very coarse, so I will probably get some chip-out when I cut. I will get a better blade at some point, but I'm not too worried for now, as these bits won't be seen at all.
I then downloaded two pieces of free software - the "Teensyduino" development environment, and a programme called "MIDIox", which shows you what MIDI events are being transmitted to your computer. I plugged the Teensy board into my laptop, fired up the Teensyduino software, followed the instructions, and the in-built LED started to flash, showing everything was working properly.
I then replaced each key back in the keyboard, and reattached the springs. You will see from the picture below that I could not reach some springs while the keyboard was still mounted to the base. I took this opportunity to release the keyboard entirely from the base, unscrewed the circuit board, and finished adding the springs back. Then ,I plugged power into the keyboard, connected it to my laptop, and made sure it still worked with the Hauptwerk software. Fortunately, it still did! I repeated these steps with the other keyboard. I will be keeping the extra circuit boards with the modulation wheels and volume controls on them. While the keyboards do not need them in order to work, they seems to use 10K ohm variable resistors. These wires may be useful to hack into later on in order to operate a master volume control on the organ, or swell pedals.
(And if you ask, "Which one's which?" I'll get very upset! That took days!
I thought I'd make a start on the pine version of the key cheeks. I got some redwood pine milled to 64mm x 30mm at my local timber yard, and then cross cut it on my mitre saw to 2 lengths of 400mm and 2 lengths of 292mm. You can see the masking tape I wrapped round the wood to minimise tear-out.
I then set about installing the pedalboard reed switch array in the holders I had made. I had thought about using epoxy to permanently attach them in the grooves, but after temporarily fixing them with gaffer tape, I realised that this solution was plenty good enough - they're never going to be moved - and it means that if and when a switch fails, it will be an easy task to replace it. So I secured each reed switch with two strips of gaffer tape, and then topped the whole lot off with a layer of masking tape. You can see the black velco to attach it to the base of the pedalboard, where there is a corresponding strip. I also glued the magnets to activate the reed switches in the appropriate place on each pedal. I had been going to use superglue, but at the last minute, I decided to use my new hot glue gun. This was a good job, as I somehow managed to start gluing them in the wrong position. Fortunately, isopropyl alcohol is absolutely brilliant for removing hot glue, and it was a simple job to rub each magnet with a cotton bud soaked in it, peel it off, and reposition:
Then it was just a matter of attaching the reed switch blocks to the pedalboard with the velcro, and ensuring that the Hauptwerk software recognised it, which went without a hitch.
I bought a plastic cutting disc from Homebase, which was intended to be used with a Dremel. I don't have a Dremel, but I did have an old tiny pillar drill for drilling holes in circuit boards. I thought I would see if I could get away with just using that, and it turned out that I could!. Using this had the added advantage that I could set the height of the cut, and leave it set, so each key would be cut at exactly the same height. I cut the front of each key off, level with where the sides of the key met the front. Then I sanded each cut surface on fine sandpaper to remove any burrs.
I then uploaded the programme to the microcontroller, and tested out my buttons in MIDIox. I could see that I was getting multiple events being transmitted for each button press, probably due to my cheap and horrible switches having a lot of ontact bounce. I think that Hauptwerk has an inbuilt switch de-bouncer, but I thought I may as well try and improve things in the programme anyway, and so I changed the line that says "delay(25);" to read "delay(50);" This is the delay in microseconds that the microcontroller waits before carrying out the next action. It gives the switch contacts a chance to settle down. This didn't solve it completely, but it made it significantly better, and it shouldn't be a problem.
Listening to organ music on YouTube, I came across some software called "Hauptwerk" (link here: www.hauptwerk.com). It is a remarkable piece of software, originally designed in 2002 by Martin Dyde in Britain. In 2006, he developed it further with his company, Crumhorn Labs, and in 2008, sold the rights to Milan Digital Audio in the United States.
Essentially, Hauptwerk is designed to allow the live performance of pipe organ music using MIDI (Musical Instrument Digital Interface) controllers, which then, through your computer, play very high quality recorded sound samples of many pipe organs from around the world, depending on what notes you have pressed, and what stops you have selected. Each rank of pipes (each stop on the organ) is individually sampled, so you can choose and select them as you play, altering the sound of the instrument, exactly as you would on a pipe organ.
Now, I have come across MIDI many times as a pianist, and I have to say I was never overly impressed with it for reproducing the sound of a piano. However, the organ, with its notes being simply either on or off, rather than any shade of musical colour as on the piano, is an ideal candidate for MIDI. Also, advances in computer technology mean that latency (the gap in time between playing a note and hearing it, while the computer figures out what it's doing) has been greatly reduced due to 64-bit operating systems letting up to 128GB of RAM memory now be addressed. Essentially, it is now easily possible to build a computer which can hold an entire organ's pipes in it's quick-access memory to be played by you on a MIDI keyboard and pedalboard.
I was impressed, and I wanted one! So I set about doing all the research I could.
The last few days have been taken up with finishing off the pedalboard varnishing and getting it to talk MIDI to the computer. The shellac I have used t isn't the most hard-wearing of finishes, but it was what they used when the pedalboard was made, any points it loses on the durability front are more than gained in ease of application and renewal. I applied six coats of shellac to the pedals, and then turned my attention to finishing the bench.
I had bought some dark coloured pine stain, but was aware that pine can be a really difficult wood to stain, so I tried it on a bit of spare wood from building the bench. The results weren't good. Terrible blotching and uneven coverage. Worse, in fact, than this photo makes it look:
I have ordered some European Oak from my local lumber yard to make the side pieces, and will use 9mm birch ply to make the lengths that the keyboards will sit on. The ply won't be seen, and, if necessary, I will reinforce it somehow if flex becomes an issue. There will be a stopped rabbet joint in the base of each oak length, which will accept the ply, and it can then be screwed into the oak from the base with countersunk screws. Not shown on the model yet are the piston rails, which will be 4mm oak ply, screwed along the base of each birch ply sheet at the front with some brass screws. There just isn't room, sadly, for any more invisible mounting solution for them - they would all be fouled by the switches that will be in the rail. At least, I haven't been able to think of any yet. Having practised a couple of rabbet joints on some scrap pine, I realised that in order to clamp it down well enough for my router with the equipment I have, my oak needed to be at least 40mm thick. I ended up buying 45mm thick oak, at quite considerable expense! One thing that this project is bringing home is that the right tool for the job is absolutely imperative, and cheap tools are a false investment. For some reason, it seems to be a lesson I need to learn time and time again...!
Having given the pedalboard another good clean, I decided that I really should re-varnish the playing surfaces before I go any further. It's one of those things that, if I don't do it now, will never get done. So, I sanded all the pedals back to bare wood with 120 grit sandpaper, and then 240 grit. I will put a few coats of shellac over the top of it. I think it looks an awful lot better. The keyboards have arrived, too. I've plugged them into my laptop individually and made sure they work.
Well, the microcontroller has arrived, and I've had a chance to have a bit of a look at it. I suppose the clue should have been in the name, but it's TINY! They don't come with pins soldered on, and I couldn't find any in my electronics stash, so I had to order some 0.1" pin pitch headers to solder onto it so I can plug it in my breadboard for prototyping.
I worked out the measurements for the plywood key beds. I want a 3mm gap each side of the ends of the keys before the key cheeks, and I need to take account of the depth of my rebates in the key cheeks. I asked my timber yard to cut these pieces for me, as I didn't want to be trying to deal with an 8' x 4' sheet. I asked them to cut me one piece 1100mm x 259mm for the top keyboard and 1100 x 367mm for the bottom keyboard.
An organ builder in Wakefield, Yorkshire got back to me and said he has something from an old organ built by JW Walker & Sons that he would be willing to part with for £100. I have had emails from other restorers quoting between £300 and £1,000, so this got my interest. He has sent some photos. It is cosmetically bad, but nothing a bit of elbow grease won't solve. More interestingly, he is visiting Birmingham in the next few days, and has offered to bring it to me for an additional £25, so I won't even have to hire a van and go up to Yorkshire! He will be arriving in the next couple of weeks, and the first job will be to assess whether the electrical contacts left on the pedalboard can be re-purposed for giving signals to a MIDI-encoder, or whether I'll have to go down a different route to get it to somehow talk with the computer. Then, I'll set about making it look a bit smarter! I rather like the idea of an old organ forming the basis for this project - hundreds of feet will have played it over the years. It feels grounding. And slightly humbling in a weird way.
Anyway. Shortly, this beauty will be mine!
So, I decided that the sensible thing to do would be to accept that it is pine, and not oak, and not try to make it look like something it isn't. So I bought some "Antique Pine" coloured stain and gave it two coats of that: