Help me find the motor for my clock!

April 23, 2007 12:38 PM

Help me find a motor to really make time flee!

I'm trying to make an (unorthodox) clock which involves a large motor of some sort turning a loop of plastic film, like a belt. Because this is a clock, the film will have to move very accurately (1.5 millimeters per minute or so), which means that the motor will as well.

What are my options considering the kind of motor I should use? Should I use a stepper motor? Should I use a non-stepper DC motor and pass it through some kind of gearbox? Use worm gears? Are DC motors' RPM speeds fairly accurate enough to be relied upon for use as a clock?

Thanks!

posted by provolot to science & nature (22 comments total)

I don't think a DC motor's RPM is accurate enough to keep time. The speed of an electric motor is determined (primarily) by the voltage you put into it, so any fluctuations in voltage result in you losing (or gaining) time. If you're running this from a good, regulated bench power supply (a $200+, 75lb item btw), you might do alright.

And 1.5 mm/m? I can't think of a DC motor that will move slowly enough for you to do this without a couple gears.

Likewise, I imagine that using film (like celluloid photography film) will result in a good deal of slippage.

If you're not looking for a particularly accurate timepiece, this might work. But, I wouldn't expect it to keep very good time.
posted by Netzapper at 12:58 PM on April 23, 2007

The best way to do this will be to design a gearbox to use a very high RPM and low torque motor. The greater the gear reduction, the more precise you can make this. The crucial part will be having a rotary encoder in this gearbox. You'll probably want to put this at a stage where the revolutions are quite numerous, so you can use a less finely graded (and therefor less expensive) encoder. You'll then need a micro-controller or something to monitor the rotary encoder's output and pulse width modulate the input to the motor to keep it accurate. You could connect this to the net via a computer and use the network time protocol to keep your clock obscenely accurate. This will also let you do nifty things like adjust for daylight savings.

If you're not willing to go through all that trouble, the bare minimum is going to be the big gear reduction and a stable power supply. But at 1.5mm per minute, I doubt you'll find much success.
posted by phrontist at 1:14 PM on April 23, 2007 [1 favorite has favorites]

I suggest finding something that has already solved the problem, then adapting that to your needs. For example, use the innards of a mechanical clock. The minute hand can be expected to make one revolution per hour. Affix a wheel to the minute hand's axle to turn the belt. Adjust the size of the wheel to match the length of time displayed by the belt. If it doesn't provide enough torque to move what you want to move, you could gear it down.
If you need much more accuracy and torque than that, you'll probably start looking at stepper motors.
posted by leapfrog at 1:30 PM on April 23, 2007

Is AC motor OK?
The synchronous motor does this very well. Several electronics supply catalogs have them- for example, but I don't know if they have enough torque.
posted by MtDewd at 2:00 PM on April 23, 2007

The AC syncronous motors that most electric clocks are built around are sensitive to line frequency, not voltage. Operate a 60Hz clock on nominal 60Hz power, and it will stay syncronously locked to power frequency, as long as there is power. If the utility company's operating frequency drifts even slightly from the standard 60 Hz, the utility company can insert Time Error Correction cycles to correct for drift, but this is a fairly rare event, given the effort put forth by all utilities to keep synchronized to national and regional AC distribution grids, so as to be able to flexibly sell and buy excess power.

So, buy an inexpensive electric clock, pull the AC clock motor and hand drive transmission, and have a sprocket drum made whose diameter will move 1.5mm per minute. Drive that drum from the minute hand drive output. Hint: This drum will likely have a diameter about that of the original clock face, particularly if you buy a clock whose minute hack marks are exactly 1.5mm apart. :-) Hint #2: There will be precious little torque available on that drum, which is good for preserving the integrity of your film for many, many cycles, but should be a warning to you not to introduce significant drag with your film loop (guides, other rollers, etc.)
posted by paulsc at 2:01 PM on April 23, 2007

Also, this clock is a battery operated quartz movement that has enough torque for bigger hands- still maybe not enough for you. I have bought clocks from them before and can recommend them.
posted by MtDewd at 2:07 PM on April 23, 2007

A stepper is, well, stepped. At such a small rate of motion the steps are going to be obviously non continuous if that's a concern unless you run it thru a gear head too. As well steppers will occasionally miss a step so they aren't 100% accurate.

Unless you need a lot of torque the easiest thing would be to salvage a motor out of an electric clock. The hour hand is geared to sweep once an hour so you'd need a drum for your tape 90mm in circumference or about 28 mm in diametre to get a surface speed of 1.5mm per minute. Plus they use the 60Hz line frequency to keep very accurate time.
posted by Mitheral at 2:53 PM on April 23, 2007

Or ya, what paulsc said. must preview more.
posted by Mitheral at 2:56 PM on April 23, 2007

Miteral's got it, if your belt is really small. I thought we were talking about something really huge here...
posted by phrontist at 3:34 PM on April 23, 2007

Thanks a lot, everybody!

The clock in mention is actually 1m wide, so the loop will be significantly large (in comparison to a small electric clock, that is). I realize my 1.5mm / min requirement is pretty hard -- I'd settle for a motor turning the film, say, 1.5cm once every 10 minutes. Should I still go for an AC motor in this case? Or should I work with a stepper motor to drive this every once in a while? How accurate are stepper motors?
posted by provolot at 3:54 PM on April 23, 2007

provolot: No, I would not use a stepper motor for this. If it's a meter wide, you're almost certainly going to have to go with the feedback/microcontroller method with a beefy motor and gearbox described above.

These motor controllers are pretty great. If you aren't a propellorhead, check out the Ardunio microcontroller, which is intended for art applications. Looks like someone has written some example rotary encoder code.
posted by phrontist at 4:05 PM on April 23, 2007

I realize my 1.5mm / min requirement is pretty hard -- I'd settle for a motor turning the film, say, 1.5cm once every 10 minutes.

Uh, what? That's the exact same speed...
posted by phrontist at 4:11 PM on April 23, 2007

phrontist, thanks! I've heard a lot of great things about Ardunio lately -- I guess it's time to try it out.

And I meant that it'd be okay if a motor only turned, say, 1 second every 10 minutes, and during that one second, the motor would move the film 1.5 cm. That'd be different than continuous movement, and be easier, no?
posted by provolot at 4:17 PM on April 23, 2007

If it was me and the motor isn't a visual thing I'd go with a solenoid, a timing chip (maybe something as simple as a 555) and a ratcheting mechanism. The solenoid is activated by the timer and pulls the strip one notch forward using a spring loaded hook.

Excuse the Crude Diagram, I don't have CAD here. You can see that at rest the hook is setting in the hole in the strip. When the soleniod activates it pulls the strip down one increment. The hook is then released and it overrides the strip to the next hole where it waits for the next increment. You could activate two solenoids at the same time to pull on each side of the strip. You also might want to have the hooks activate a sprocket which moves the strip (ala sprocket fed printers) to save on wear and tear on the strip.

You could also activate the solenoid with a switch activated by a multi lobe cam mounted on an electric clock and avoid transistors.

PS: Someone used to make film strip imitation digital clocks. Similar to the Dansette on this page but instead of the little flaps a strip of film was used for the numbers. Two for the minutes and one for the hours. Very Rube Goldbergest.
posted by Mitheral at 4:37 PM on April 23, 2007

Okay, I grok now. That does make things easier. One of the problems of using a rotary encoder is that it's almost impossible to have a non-slipping belt, and the gearbox will also have slip, and these mechanical issues quickly lead to an unacceptable loss of precision. So you've got an input to this system (motor speed) and an output you're trying to control (the absolute position of the belt). Life is made easier in direct relation to how directly you can measure your output sooo...

I would pick up one of these lego reflectivity ("light") sensors and hook it up to the adruino board. On the edge of your belt (on the part that faces inward if you want you don't want to mar your display surface) paint lines (something with good contrast to your background color) every 1.5 cm. So the belt starts off with the sensor sensing a line, after 10 minutes (minus the previously ascertained time required to move the belt 1.5cm) the board turns the motor on (no speed control required anymore) until it senses the next line. Lather, rinse repeat.
posted by phrontist at 4:39 PM on April 23, 2007

On Preview: I think my method would be easier than Mitheral's, and allows you more control over aesthetics, whcih I'm guessing is important here. Really, it's the same idea, it's just mechanical complication versus software complication.
posted by phrontist at 4:41 PM on April 23, 2007

Actually, it's about as easy software-wise in either case.
posted by phrontist at 4:42 PM on April 23, 2007

after 10 minutes (minus the previously ascertained time required to move the belt 1.5cm) the board turns the motor on

Oh, and just to be clear the time required to move the belt 1.5 cm could be tweaked through mechanical design to be close to 10 minutes, giving the illusion of continuous movement (this will be slow).
posted by phrontist at 4:51 PM on April 23, 2007

I say make it easy and more rube goldbergish---faster motor, more belts, on ever-increasingly-big cogs.

And if it's for an art installation, I'd use a PWM so that I could make time...you know, *fly*.
posted by TomMelee at 5:08 PM on April 23, 2007

TomMelee: But that's really hard to make accurate.
posted by phrontist at 5:22 PM on April 23, 2007

It might be worth looking around for a heliostat as a starting point. A heliostat is designed to track the sun as it moves through the sky, so the motor and the reduction gearing are already in place to solve a simiilar problem.
posted by Killick at 5:24 PM on April 23, 2007

Thanks everybody! and thanks phrontist!

I got a lot more responses than I thought I would, and this is great! I think I'll use phrontist's solution, as it does make sense to measure the plastic film directly.

Also, if/when I finish the project, I'll update this thread. Thanks again, AskMe!
posted by provolot at 7:01 PM on April 23, 2007