Twenty Two Servos And An Awesome Clock


We can never get enough interesting clock builds, and [ebrithil]’s servo clock (Deutsche, Google translation) is up there with the best of them. There’s twenty-two servos in this clock, moving time forward with the mechanistic precision only an Arduino project can.

The digits in [ebrithil]’s clock are constructed like seven-segment displays, only instead of lighting up LEDs, servos turn small bits of paper that are light on one side and dark on the other. Turing the servo 180 degrees changes each segment from one shade to the other, making for an electromechanical seven-segment display.

The servos are controlled by an Arduino Mega connected to a DS1302 real-time clock. One problem [ebrithil] had with this project is having the segments rotate slightly every time someone turned on a light attached to the same circuit. He solved this problem by running the circuit off a perpetually charging cell phone battery, allowing him to take this clock on the go without losing time.

27 thoughts on “Twenty Two Servos And An Awesome Clock

  1. The second video doesn’t seem to show any kind of improvement with the alignment of the cards. I think this is a great idea, however I find it quite difficult to read at all.

  2. shows an alternative option, and indeed a few articles on flip-dots have run here on HAD. The idea is the flipped element, in this case the 7 segments, is magnetic. A pulse from a solenoid coil, north or south, flips it between showing the black or coloured side. A small permanent magnet, weaker than the solenoid, is enough to keep each flipped element still between deliberate changes.

    Uses no standby power, just enough to pulse a solenoid for a small while to do the flipping. Sensibler than servos, unless you happened to have a LOT of servos about. There’s articles in the HAD archives about methods people have used to recondition and, I think, make these from scrap. The servo clock here might take well to being adapted like this.

    You’d either need an H-bridge to drive each coil, or use the coils in opposite-facing pairs. Or use a split -ve, 0, +ve power supply. You don’t have to drive all the coils at once, so if you stagger it the power supply can be modest.

    1. I have a few of an older version of those for a project:
      (Not from flipdots, as these are yellow, no idea who made them, they are quite old)

      They are awesome to read, quite easy to control. But the main problem is, they are horribly expensive. You can easy pay over 100 euro for 1 digit of 25cm.

      You could DIY this, but DIYing servo’s is easier I think.

      1. That’s what I was thinking of, DIY flipdots.

        I imagine they’re expensive cos it’s a small market that they’re the only company to serve. Flipdots presumably last long enough to not need repeat ordering, but there’s a large amount of legacy out there.

        But I can’t see why an electromagnet and a disc on a spindle should be expensive to make! Shouldn’t even be difficult, after a bit of experimenting to optimise the production and design. Maybe a 3D printer would help, with tiny magnets attached to the plastic discs. For a 7-segment, maybe use sewing pins mounted to paperclips, or some Meccano, Fischertechnik, or even Lego bits for the joint. Or a paperclip and a washer.

        You could even have the segment suspended from a bit of cotton thread. As long as there’s enough magnetism to keep it rigid, stop it flapping in the breeze.

        Just seems easier and cheaper. For a permanent build, it seems a waste of 22 servos. Mechanically much simpler too, lots of things to go wrong in a servo, in an outdoor cold, hot and wet environment.

  3. Didn’t someone say something about trickle charging lithium batteries back in that other thread? It was an explosion type warning. Isn’t this the same thing he’s doing here?

    Would a big capacitor, or even small lead-acid or NiMH be better? AIUI it’s only there to absorb transients in the power supply. A big capacitor should do that, and maybe increase the input voltage a bit to start with, before the regulator.

    1. If there is a regulator it hasn’t enough voltage overhead to regulate, that simple. Disconnect that battery and safely dispose of it, and design a better power supply!

  4. No reason as to why you cant drive every servo at once. If you need to you can get an i2c servo (**cough** pwm) driver from a multitude of places such as adafruit. This way you can set them all at once and not have to worry about driving the PWM

    1. Power consumption.

      I have a project where I used dozens of servos. The power requirements were kinda high. More than what I liked. After careful design I realized I only needed to drive two or four servos at the same time at any given moment. Suddenly my power requirements became more manageable, ergo cheaper.

      1. Correct me if I am wrong, but if you stop providing signal to a servo doesnt it go limp? The basket of servos I have do. So in the cas eof arduino code if you call detatch it will not be powered.

        I understand the power consumption issue but if it were me I’d take an ole ATX/laptop psu and use that :)

        Just some ideas.

        1. “ATX/laptop psu”
          I encourage you to try that.
          At the very nasty pwm inductive loads of 10+ servos doing things at the same time your voltage supply _WILL_ break down. (what is what happened in the second video as ebrithil told us)

        2. Well yes they do go limp. In my application there wasn’t a whole lot of opposing force against most of the servos except some friction. Basically a servo at rest wasn’t going anywhere.

          Unless there’s a strong breeze on this guys project I don’t see a lot of opposing force against his servos either.

  5. Why not 22 glow engines with electrically operated clutches? Maybe 22 solenoid operated gas valves inflating balloons? Surely there must be a way to make this less efficient. :-D

    1. I once contemplated a clock where the segments were high-voltage arcs.

      Um, yeah.

      I should get onto that while CRT/TV flyback transformers are still common. Yes I know, there are a few minor technical hitches to overcome…

      1. I think you meant to say “HELL YEAH!”

        Apart from the obvious, tedious deadliness etc, I’d worry about the arcs repelling, or maybe attracting each other. Maybe you could time-multiplex them so nearby arcs don’t come on at the same time, and twiddle with polarity so everything’s attracted in the right direction.

        Other thing with these, size is a function of voltage! So make it robust enough that you can have segments feet wide just by dialling it up.

        I can’t think of any museum, restaurant, living room, or anything short of operating theatre, that wouldn’t want at least 2 giant sparking deadly arc clocks.

        Hey maybe you could start with an analogue clock. Have a dial with 1 centre electrode and 60 outer ones. A smaller ring of 60 for the hour hands too. You could even drive the electrodes mechanically, use a real clock mechanism with the hands brushing against the relevant electrode. So you’d just need a real clock, an HV supply and a few dozen metal pins.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.