I want to build a split flap display. I need the least expensive way to determine rotation. Any ideas?
Chris Spurgeon wrote 02/03/2019 at 22:51 • 1 pointHi all...
I want to build a split flap display board, the kind of thing that used to be commonplace in European train stations and airports. (Here's a YouTube video of the board at the Frankfurt airport -- https://www.youtube.com/watch?v=cj32w5z81Ak , and here's a wikipedia page about split flap signs -- https://en.wikipedia.org/wiki/Split-flap_display ).
Integral to making one of these signs is having a way to control the rotation of each character, so it stops at the correct letter. Stepper motors would be perfect, but given that I want to make a sign with at least a couple hundred letters, it became obvious to me that using stepper motors would be way too expensive (both in the cost of all of those motors, and the cost of the electronics to control each of them).
I was thinking a more economical way to go would be to use simple low cost motors, and turn off each character's motor when that character has rotated to the proper letter. That means I need a simple (and inexpensive!) way to detect that rotation. I'm looking for suggestions. Optical rotary encoder? A series of little reed switches on each letter that open and close in various combinations? Some sort of Hall effect thing? Some sort of rotary potentiometer?
I envision each character having about 45 different display possibilities (26 letters, 10 digits, some punctuation and misc symbols), which means I'd have to detect rotational differences of eight degrees or less.
Ideally I'll come up with some simple way to talk to each character, something like sending the desired letter to display to each character via I2C or something. But nothing happens without my first coming up with the most economical way possible to gain feedback / control of each character's position.
What should I do? What would YOU do?
THANKS everyone!
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
I you don't mind having a really slow update rate, you could connect all the displays to one single motor. Connect them in way that is similar to a dial combination lock (disks with protrusions that, once the meet, will turn the next disk). Then have the motor, turn back and forth and back and forth etc. This page describe how this works pretty well https://woodgears.ca/combolock/
Since flip displays only work by turning in one direction, you'd need to add some kind of ratchet mechanism. I think you also need to add some gears to inverse the direction on half the disks, or double the number of disks, but I'm not sure.
This turned out to be a stupidly complicated solution. Please don't use this (-;
Are you sure? yes | no
As others have said, conventionally these split flap displays would use little microswitches or ir photo-interrupters to count position during rotation and know the home position. It would be interesting to go analog, for instance if it were possible to use a potentiometer modified to be continuous rotation (or maybe use a multiturn pot and make sure not to over rotate in software) and that way all your controller would need to do is read the voltage via adc and it would know the absolute position by referencing a look up table. Of course though pots would be prone to oxidation and wear so this may all be a bad idea but still interesting to think about.
Are you sure? yes | no
Yeah, I've been thinking of maybe using pots.
Are you sure? yes | no
Why do you need to know the position if current in one direction can flip a magnetic disc, and a current in the opposite direction can flip it back?
Are you sure? yes | no
I think you might be thinking of a flip disc display, not a split flap display? https://en.wikipedia.org/wiki/Flip-disc_display
Are you sure? yes | no
Hmm.. Maybe I don't understand what you mean with split-disk display. The type of thing that used to be used for alarm clocks in the 1970s used to be called a flip-clock. The split-disk in the following video looks like a variant of a flip-disk (split but still bi-stable): https://www.youtube.com/watch?v=YmSJlQZByIQ
Are you sure? yes | no
Depending on the number of character places, today instead a large forest of sensors it may be cheaper for display controller to have a visual feedback - small camera placed at a point with unobstructed view of the display, watching it to know which place motor has to keep on spinning and which ones should be stopped.
Are you sure? yes | no
WOW I never thought of that! Interesting.
Are you sure? yes | no
simplest way might be two microswitches, one that detects zero and one that counts letters. drive with a cheap gearmotor. Use an attiny to count. Chain it all on an i2c bus or even custom chaining like neopixels do.
However there has to be an electromechanical solution, because these things predate modern electronics.
Are you sure? yes | no
For my project I'm using a single microswitch to determine the position once every rotation. Perhaps you find it interesting.
https://hackaday.io/project/163725-split-flap-display
Are you sure? yes | no
Ooh that's really pretty!
Are you sure? yes | no
Looking at that video, could you just replace the bar used to retain the top flap with a microswitch, and use that to count?
Are you sure? yes | no
Those 28byj-48 are pretty cheap too.
Are you sure? yes | no
If you can make or buy a conductive ink with a known resistance per square inch, you could print a tapered circular pattern on the side of each wheel and add a couple of copper brushes to turn it into a DIY potentiometer (or just add a cheap pot to the end of each wheel); you could also put a bar magnet on the end of each wheel and use hall effect sensors to determine the position, or print a conductive pattern on each display and use a different SMD resistor to identify it by brushes that touch it on its way to the front of the board.
Are you sure? yes | no
Is (part of) the back of the flap visible to the MCU? A barcode? One advantage is that on cold start it knows where the flaps are at.
Are you sure? yes | no
What is the size of the flaps? You may only need a very small, very cheap stepper motor that can be driven with very simple parts.
If you are dead set on not using stepper motors you would need a photosensor for the zero flap and another photosensor to indicate whole flap units. I envision the flaps held between two disks that form a cage, which would have reflective marks on each side for the sensors to indicate on.
Another possibility is to have a loose friction fit on a common shaft, and then a solenoid on each wheel that stops it from turning. either you could have a tab that catches the solenoid to zero the display on the first rotation and then releases it some time during the second rotation, or you could have it where the solenoid locks the wheel in places when the correct number comes around, which would require an index photosensor but would negate the need to clear the memory first.
Another possibility is a solenoid driven ratchet that moves one flap per pulse, but that may be too slow, and would still require an index.
Are you sure? yes | no
I hadn't thought of either of those solutions, thanks! I think they both have promise.
Buying a used display isn't really a possibility I'm afraid. These signs are becoming pretty rare, and used ones almost never show up for sale. And when they do, they are snapped up for many thousands of dollars.
Are you sure? yes | no
Why not have a micro switch or a photo sensor that detects when the next flap reaches the proper top position. Have another sensor for flap zero which has a small tab extending further than all other tabs.
I would probably just buy a used display - if there are still some that get thrown away.
Are you sure? yes | no
Ooh, I'm afraid these signs are getting increasingly rare, and are highly prized. It's really impossible to find one that's being scrapped. (I've tried!)
Are you sure? yes | no