I’m delighted to announce that we’ve hired Mark Demers from Spikenzie Labs in Montreal to work on the control system for the two Solari units. Mark really ‘gets’ the project, and is full of ideas for how we might proceed. He’s also sensitive to the particular technical and creative constraints we’re facing. This includes our wish to avoid making any irreversible physical changes to the Solaris. For example, if at all possible we will use the existing cavities inside the signs to install the new electronics, while bypassing (rather than removing or destroying) the existing components. With proper documentation it should be possible to return the signs to the condition in which they were originally received from ADM, should we eventually want to do so.
In the spirit of openness that has come to characterize the best of the maker community, Mark and I will be sharing our process with the readers of this blog. This will include insights into R&D, prototyping models, and schematics.
Right off the bat, then, here’s Mark’s first physical experiment. He says: “I made a mechanical model of the split-flap counter and zero sensor with our laser cutter and I’ve been using it to start working on the coding. It has a DC motor, micro-switches and rotation is controlled by the circuit. It is based on 40 transitions per zero trigger. (Similar to the ones in your sign which have 4 motor rotations with 10 transitions for a total of 40 transitions to sense per full set of flips.)”
What he’s tackling here is the technical specifications of the Solaris, in lieu of any documentation; historically, the Solari factory has not been particularly helpful to folks wanting to get their old signs working again. A major concern of mine is to try to reproduce the original behaviours of the Solari signs: how fast should they flip; in what sequence and direction; individually or all at once? (And what will our artist want them to do?)
Clues can be found online, since a few YouTube users have actually taken the trouble to record Solari boards in action around the world. I thought this video was interesting. The whole row starts changing, but they go straight to the next character without ‘zeroing out’ first (as we had initially thought – wrongly). For example, in the last (white) line, the word ‘KEYSTONE’ becomes ‘REGIONAL’ but the first E doesn’t move at all because it’s already at the right point in its cycle, ie KEYSTONE to REGIONAL. By contrast the ‘logic’ of this homemade one seems all wrong, and rather too fast. This one seems better but weirdly speeds up towards the end of its cycle. And to my mind this is absolutely, completely wrong.
Meanwhile, back to Mark: “I built a new model after seeing the actual display that you dropped off. The new one incorporates gearing and two different plains and rotational speeds for the counter and zero position sensor. I have also included diodes in the same arrangement, so that I could be sure that my counting circuits works with the actual displays….By adding the diodes I had to reverse the polarity of my de-bouncing circuit. There is also a bit more noise since part of the counting circuit is shared with the zero sensor, but I’m currently working though that. The code is at a point where I can type a character and the display updates to that character by first passing through blank.” (April 29, 2016)
“Here is the decoding of the Solari switches.”
Finally, for this omnibus update, I’m delighted to welcome Danica Evering to the project. Danica is an artist, curator, and grad student in my department at Concordia. She will be primarily responsible for logistical issues relating to our artist collaborator and the galleries we’ll be working with on the installation.
More updates as they happen.