Op. 3 wrap up, and next projects

So now that I've done several conversions, I've generalized the process to the following:

Technology splits roughly into the following categories, which will determine how difficult the conversion is.  The list is, from least to most difficult

1) 2000's and later instruments - Fully MIDI capable, plug it in and go, with some minor annoyances.  I have found instruments from the late 1990s have quirks which are very annoying.  An example is a 1998 Johannus Sweelink 30 that sends an "all notes off" message on each piston.

2) 1980's instruments - These instruments have digital key action and scanners, but have analog tone generation systems.  The most effective way to convert these instruments is to use the scanners that are there, but replace the control computer with a MIDI capable microcontroller.  This is my Op. 4 conversion.

3) 1970's instruments - I'm about to start on a project with a Rodgers of this era (schematics are actually from the late 1960s).  Here, the tone generation is analog, and the key scanning is fully electronic, though not digital.  In this case, all of the wiring for the keyboards does go to one place, which allows you to use the wiring that is there, but all of the electronics must be replaced.  So the wiring task is simplified, but the amount of electronic work is increased. 

4) 1960s and before instruments - Here, the key action is electromechanical, and the only option is to strip it down to the switches and wire each key itself.  This is the Op. 3 instrument.


There were a number of goals for the Op. 3 instrument.  It is an instrument that happens to be going to my new teacher, Tom Mueller, but that's somewhat beside the point.  I wanted to experiment with a few things after a less than successful conversion of my Op. 2 (Zimmer) instrument.  One of the key things was to try to get a real moving stop action in place without spending a fortune on SAMs.  On eBay, I picked up two sets of Allen stop tabs.  The first one was from an older instrument.  These were more like traditional SAMs.  There's a coil for ON, a coil for OFF, and two terminals for a switch.  Add all of your own stuff to make it work.  It turns out I didn't use these, and they are waiting for another project (along with the Zimmer SAMs).  I got a second set from an 80's vintage Allen.  The SAMs for this system are interesting:




There's only one control input, high for off, low for on.  The problem is that the coil is always energized if the power is supplied,  So you need to control the power applied.  If you have 60 SAMs, each being 30 ohms (not to speak of the inductive load), and require 24V to drive them, you are talking about almost 50A surge.  So some careful design is require to control the power.  I have a 20A 24V power supply and some big capacitors.  The SAMs act quite quickly - on the order of 50ms, so the energy amount is manageable.  I have my controller programmed to provide 125ms pulses when there is a change, which seems more than adequate.  The power supply crowbars relatively quickly down to 15V output.

I ended up designing and building my own stop drivers.  The existing ones all had issues, and were really expensive.  It was also a good experience doing the firmware and hardware development.  It took forever, but now that it's done, I can use the same design elsewhere.




The board I built initially on one panel was two boards, each having 48 pulled up inputs and 48 outputs.  The outputs are designed by board population option to either have series output 300 ohm resistors to interface with the "new" Allen SAMS, or to have integrated output drivers (ULN2803A) which conveniently include the base current limiting resistor and the buck diode.  What I figured out is that it is pretty easy to put the scan chains together and have a single processor with 96 inputs and 96 outputs (of which I used 64)

So the system includes keyboard and pedal scanners from midi-works.com.  The stop driver and scanner as well as the scanner for all the pistons is from my board.

Incidentally, I've eschewed the mini display panels.  In my recent use, I've found myself using the monitor more and more, (probably because I have an instrument that won't fit on my console), but I've also found the mini display panels really difficult to get working properly.  More trouble than they are worth.  Plus, this goes to a home installation, which will be far better served by a monitor anyway.

On to the 1970 conversion!!  Some day, I'd like to do an Allen.  I suspect they might be more interesting because they kept the moving stop action way longer than Rodgers did.  Plus, I already have the driver hardware!

1987 Rodgers 2 manual conversion

I was presented with a mid 1980s Rodgers with the opportunity to do a hauptwerk conversion. This instrument still used analog tone generators as did the previous instrument I am working on, so originally o had thought I would convert as I am on the one from the 60s - strip it down to the switches, and use all new electronics. 

However once I took a look at the keyboards, I found that a serial scanning infrastructure was already in place. 

Upon doing more poking, this serial scanner structure was available for the two keyboards, pedal, stop action and Pistons. Even better, all the wiring was already in place!  So if I could figure out how to replace the computer with a midi interfaced microcontroller, I'd be golden. 

I was able to pretty quickly interface To a keyboard by itself. It is a simple chain of shift registers. The challenge was to figure out how the motherboard was wired for the pedal and Pistons. 

This shows the single board computer plugged into the motherboard. 

After a couple of hours of poking around, I determined how the other two channels of pedal and Pistons were interfaced. On the right you can see two columns of shift registers. One is for the pedal and one for the Pistons. On the bottom is my microcontroller. 

Incidentally, the date codes on the chips indicate that this instrument was built in 1987.

The motherboard interfaced with everything at 12v. Fortunately, the shift registers were all cmos, and after data sheet review and some timing analysis, I determined how fast they could be driven. The biggest reverse engineering issue was finding how the interface between the computer and the shift register chains was made. Rodgers had about 4 levels of buffering. One level, a level shifter, didn't run at speed at 3.3v. So after removing that from the circuit everything worked.

The micro was programmed in assembly, and every byte of memory in the lower page was used. 

The lights interface was not reverse engineered or used, though it works on a shift register principle, but in the other direction. It would be straightforward to implement these. 

After finishing the interface, cleaning up was an order of business. All tone generation and switching circuitry was removed, as well as the inefficient and now hopelessly over specified power supply and amplifier. 

All these boards and associated wiring were removed

Resulting in a much cleaner interior

Showing up is half the battle

I have been burned out on practicing.  Since I played the Edwin Lamar arrangement of "Danse Macabre" in October of last year, I haven't sat down and played for hours.  There are two parts to this issue.  Firstly, I'm distracted by building instruments, and in my work on a major new sample set that I'm trying to get released.  Secondly, and more importantly, I realized that I was caught up in what all the cool kids are playing.  Difficult works by Reger, orchestral transcriptions, modern music...  I've realized that while these can be part of my practice, I do need to work on music that feeds my soul and where I can see progress.  So I learned most of the first movement of the B flat Mendelssohn sonata this morning.  It felt good to sit down for 3 hours and just work on something where I saw progress.  I also went back to my favorite Hauptwerk instrument, the CUI Casavant.  It is a great practice instrument, and since I incorporated a metronome into the sample set, it makes it that much more versatile and appropriate for learning music.

Showing up is half the battle.  Looking forward to showing up by doing something you want to do instead of what you have to do makes it way easier to show up in the first place.

Switches - experience is a great teacher

 Having worked with this style of keyboard on my last instrument, I was better able to work with this one. 

The organ was in a beach community, so there were a lot of sticky keys. I was able to clean the front bushings and coax the openings a bit wider and the keyboards are beautiful right now. I separated the keyboards from the switches this time. It makes it way easier to work on.  

The swell switch matrix is wired. I used 4 to 5 redundant contacts to make up for the low currents.  Fortunately on this keyboard the springs are already soldered so I don't need to go through that messy process as I did last time. 

Cable dressing on this one, based on experience, is much neater than the last instrument. 

Time to order more ribbon cable...

Tone generators

So it looks like essentially it is the equivalent of a 3 rank highly unified instrument. There were 85 notes of "flute"

73 notes of "diapson"

And 61 notes of "Celeste"

Just inferring from the parts, there is a base sine wave generator for the flute and additional filters for the diapson and celeste

These were on 2 racks with the thousands of wire connections.  You can see the 5 octaves of Celeste, 6 of diapson and 7 of flute. 

I'm glad I live deep into the digital age!!

Old time switching

It is amazing the primitive nature of pre digital designs. 

Key action. 

Pedal key action

Stop action (on the keyboards) and couplers. 

It is all about interconnecting thousands of wires. All done in semiconductor logic now. 

I have always held in admiration the scientists and engineers who were able to create television broadcast and receive systems with a handful of tubes and discrete components.

Opus 3 project - Old analog Rodgers Hauptwerk conversion

An old 70's vintage analog Rodgers came up for sale essentially for transportation cost.  I have been looking to try a conversion of an instrument like this.  These old instruments have pretty good bones, with well built pedalboards and wood core keyboards. 

Transportation was something of a challenge, as I had to get it out of there within a day, and no one was available to help.  However with a combination of our awesome hand truck, working out, a disregard for the finish on the case, removal of most of the interior parts, and judicious application of Newtonian physics, I was able to get it loaded into a pickup myself

This console seems well suited to a conversion.  The contacts on the pedal have multiple contacts, so it looks like I won't have to use reed switches.  The keyboards have some sticky keys, which I'll have to figure out, but they are identical in construction to the old pipe organ keyboards I used in my last project. 

First step was to remove all of the old analog oscillators.  These of course are unusable, and have to go to hazardous waste.  Even being an engineer in the field of electronics development, I'm still stunned by the pace of technology.  This was only 50 years ago!!

There is some additional removal of the now overspecified low efficiency power supply and breakout boards.  Once that's done, I can start work on the keyboards, installing ribbon cables as before.