Roland A49 Keyboard Pitch Wheel Hack?

[Bo Deadly]

I have a Roland A49 midi controller connected directly to hardware synth with a MIDI cable. Meaning no USB or computer.

Unfortunately the pitch wheel (which is actually a sort of joystick that can control both pitch and modulation together) only emits 6 bit pitch bend messages. So when you bend, it's not continuous, it's stepped. Total fail. Apparently this does not occur using it as a USB controller. It's only when you're using it with old-school hardwired MIDI.

So I'm thinking about possibly hacking the A49 in some way such as by adding some 3.5mm TS jacks for pitch / modulation CV out. If the joystick is implemented using encoders with regular bushings, maybe I can replace them with pots.

Has anyone ever opened one of these controllers? I know nothing about keyboards. For all I know if I unscrew all of the screws on the bottom, the whole thing will just fall to pieces and I'll never get it back together again correctly.

Ideas? Is this a futile exercise?

 

[Bo Deadly]

Jackpot. It uses 10K pots. Here's a wiring diagram:

Very hackable. I should be able to make a CV without too much trouble but I'm not really familiar with how synth CVs work so I'll have to figure out how to add 3.5mm jacks for pitch and mod CVs but in a way that disables MIDI signals. Maybe a normaling contact that when broken just replicates the pots in their resting positions?

Note: To open just unscrew the screws around the outer edge.

Main board with ARM processor.

Pitch / mod assembly and panel controls boards. FPC cable is what connects to the main board.

Closeup of pitch / mod assembly and pots. This image correlates the most the schematic with the pitch pot (on the right) cable running over to the mod pot board / pot and then from there to the panel control board.

Closeup of mod pot. Looks like Alpha 9mm but reinforced with metal brackets.

Closeup of top of pitch pot shaft mounted in joystick.

 

[Bo Deadly]

Here is a PCB for hacking the Roland A-49 MIDI Keyboard Controller pitch and modulation "joystick" to output analog control voltages:

https://oshpark.com/shared_projects/B3W6PY65

This PCB plugs into the A-49 using the existing JST PH connectors and is wired to two 3.5mm jacks that allow plugging in standard euro-rack style patch cables which might be fed into the OSC1V/OCT and MODSRC of the Boog for example.

Here's a simple pic:

This makes the controls nice and smooth. Without it, the pitch control for example is stepped because MIDI only sends 7 bit codes of which only a subset are used because of the limited range of control (usually +-1 octave). There's maybe 10 degrees or so between each step change. Useless!

The mod is only engaged when the ring contact is grounded. Meaning when you don't have a patch plugged in, the A-49 exhibits the stock behavior.

The trimmers trim the range and offset of the CVs. For example, the pitch control would probably be trimmed to +-1V with an offset of 0. But the modulation control could be -1 to +3V or whatever. The board can output between -3 and +3V.

Note: This board is not trivial to implement. The analog switch IC is an extremely small package with 0.6mm pitch pins. It requires using flux and a solder wick to do it correctly. It is also easy to damage the ribbon cable of the A-49. This mod should only be performed by techs with the right gear and experience.

PCB BOARD 1.85"x1.6":

The PCB bolts onto the corner of the A-49 control panel PCB through an appropriately sized stand-off. A 4 pin female JST PH connector is soldered to the bottom so that it can just plug in. The female plug from the joystick then plugs into the other JST PH receptacle. So the modification is not destructive (aside from drilling holes for jacks but that is not required - you could just run a cable out).

SCHEMATIC:

ISSUES:

There is one not completely insignificant issue with this mod. The "joystick" on the A-49 has two resting positions depending on if the joystick is released from the left vs the right. Meaning there is a gap in the center position that is about 1/2 of a semitone. So if you trim it in one position and then accidentally cause it to move out of that position into the other resting position, the pitch will change and suddenly everything is sightly out of tune.

The fix for this would be to use a circuit like the one described ~ post #9 in this thread:

  https://www.gearslutz.com/board/electronic-music-instruments-and-electronic-music-production/1322003-bipolar-exponential-converter.html#post14918390

This circuit could be adjusted to the point where there is a "dead zone" in the center position. That would completely resolve the issue and work very nicely.

Another minor issue is that there is ~15mVpp of 250kHz noise from the DC converter. It should be harmless but it could easily be removed with a little LC filtering.