The steppers are far too slow to write the time in under a minute so I delay it for a couple of mins between each draw. Mostly to give the very hot motors a while to cool off and to give the etch a sketch a break!
Tags: computers, mpegs, retrocomputing, robots, toys
That's something construction articles rarely mention: that stepper motors run really hot, like 100°C or more. They should be bolted to something large and heat conductive all the time.
Also, they have no torque.
Stepper motors can have plenty of torque. You're not using the right ones if you think they're universally wimpy. In the video, notice the boxy one on the left which flips the etch-a-sketch over, with the machined end caps and the body composed of many layers of stacked sheet metal? That type is usually a lot more torquey than the cheaper stamped can types being used to run the X-Y.
Also, when they hit 100C it's often because you're driving the windings with far more current than is truly required for the work you're getting out of the motor. Simple stepper drive circuits tend to be very wasteful of power, particularly at low or zero RPM. A halted stepper with energized windings is developing close to its peak torque (stepper torque curves look like 1/speed), and is often capable of drawing as much current from your power supply as it's able to give. Since the motor's ability to develop torque (and therefore draw high current) falls off as speed goes up, the guy who did this clock might be able to keep the motors cooler by drawing faster, provided the etch-a-sketch can take it.
It's possible to design or buy more sophisticated "microstepping" drive systems which modulate winding current. The main goal is to drive the windings with closer approximations of sinusoids, which allows smoother motion and finer stepping. But a nice side benefit is that if you don't need full torque at stall or near-stall speeds, you can reduce drive current and keep the motor much cooler.
I get some of the best engineering advice here.
Thanks; I'll definitely look into microstepping drives next time. I didn't use them before because the additional angular resolution wasn't required in that particular application. But it was the braking heat problem that really doomed the project; H-bridge chips actually exploded a few times, when they ought to have been within limits according to the data sheet. I was using NEMA 17 motors, but a very cheap driver board. (You would think it wouldn't be that hard to build a robot capable of driving up a fifteen degree ramp.) This is what happens when a programmer builds hardware: electrical fires, eventual understanding of the purpose and benefits of fuses, and ultimately, perhaps a better programmer.
I used to think consoles exploding on the bridge of the Enterprise were unrealistic . But an experienced robot builder learns that they really do blow up sometimes just like this .
I am reminded of my favorite scene from Farscape:
I would note that capacitors can work as a particularly exciting form of fuse
I like this very much due to the medium. That said, it's basically a reimplementation of a pen plotter, without the ability to lift the pen, and with "tilt" instead of "paper feed and cut."
Refreshes and drawing are expensive on this build (as factors of time and heat), and the clock being drawn seems to make the worst of the tradeoffs in-place.
It seems that an analog clock would be more-funner in that it could persist for a whole hour between refreshes while still presenting a perfectly cromulent clock.
It would go something like this: Ever hour, on the hour, the clock would tilt and thus blank. It then would draw a basic watch face consisting of one large circle, optionally with marked 5-minute intervals (or 15, or whatever).
It would then return to center from the point of the current hour, thus drawing an hour hand.
After that, it would plot a course outward to a point somewhere between center and the hour-ring, thus marking the current minute. For the next minute, it would step clockwise (on an arc, of course) and draw a line back to center. This loop gets rinsed and repeated 30-or-so times (depending on how fast the first clock face gets drawn).
And then obviously, for the next hour, you tilt the thing to refresh, draw another simple clock face, and rinse and repeat.
The minute-dial would darken over the course of the hour, which is perhaps even advantageous in that one could tell at a glance -- by darkness alone -- how much time has elapsed in the current hour. Wear-and-tear is minimized. And it becomes a useful timepiece instead of a study in masochism.
Bonus points for moving the clock face around strategically, to help prevent "screen burn" over time.