Awesome Clock

[veryevil]

Just seen that and I think its awesome. I cannot find any information on the creator would love to see how it was made.

[PA0PBZ]

http://www.humanssince1982.com/a-million-times

[veryevil]

Ha ha it just appeared on  hackaday!

[Willy.nebula]

That "is" awesome

[TerraHertz]

Wow, beautiful. I'm very jealous that he figured out how to implement this so well. Both the mechanics and software for this are hard.

 I wanted to save local copies of the video. Search vimeo for "a million times" found:
    vimeo. com/60491636 A million times (Time Dubai) by Humans since 1982   The original
    vimeo. com/60333036 Humans Since 1982, a million times    Similar, copied by someone else. Silent
    vimeo. com/60824954 A MILLION TIMES    Another. different view and music

Also Anna von Hausswolff - Mountains Crave   youtube. com/watch?v=QbP--RsNj70]
  Music to the 'a million times' clock

I've wanted to make an analog clock with independently, identically driven hands for ages, but the mechanics is hard in a hobby workshop. Needs two identical fine gears on the sleeved hand-drive shafts, with worm gear drive from tiny stepper motors (such as used in floppy disk drives.) A bit beyond my machining abilities, and so far I didn't find anything suitable commercially.

But the really tricky part is arranging separate index sensors for both gears, with sufficient accuracy to allow the hands to be lined up visually perfectly. Since the eye is so damned good at sensing a slight kink in a line, the index sensor has to be accurate to a tiny fraction of a degree. it's noticeable that some of the hands in that array are a little off sometimes, so I guess he didn't entirely solve that either.

I wonder if he's using a separate small processor per unit? It seems likely. Perhaps with unit IDs that are an X,Y coordinate,  and a command broadcast syntax that allows description of 'waves' across the X,Y space, so individual units figure out for themselves what they need to do.

(Edited to stop the auto-URL insertion of videos. Remove spaces.)

[Precipice]

I've wanted to make an analog clock with independently, identically driven hands for ages, but the mechanics is hard in a hobby workshop. Needs two identical fine gears on the sleeved hand-drive shafts, with worm gear drive from tiny stepper motors (such as used in floppy disk drives.) A bit beyond my machining abilities, and so far I didn't find anything suitable commercially.

Coaxial stepper motors - http://www.jukenswisstech.com/?page_id=105
Although those have a mechanical stop, for their intended application.
You can also get steppers with hollow shafts, which solve some of your problems.

[TerraHertz]

Coaxial stepper motors - http://www.jukenswisstech.com/?page_id=105
Although those have a mechanical stop, for their intended application.
You can also get steppers with hollow shafts, which solve some of your problems.

No, useless for this application, because the step size totally won't be fine enough. For the visual smoothness and ability to gradually shift speed, it has to be a fine increment stepper, driving the hands via a worm gear. With no play - but that's easy enough.
For lowest cost too, standard steppers are a must. In my case I'm thinking of a commercial 'clock-like' application (which is also kind-of art), so mass production is a requirement.

The indexing problem really is the hardest part. Also it has to be adjustment free, and since the hands get pressed onto the round drive shaft/tube, that effectively means you have to sense the actual hands position (at one point), not the drive gear at the rear. Without any visual sign of the index sensor. So, a magnetic induction thing through the plastic clock face, perhaps. But then accuracy is a bear...

[Precipice]

Ditch the local feedback and train it with a webcam?
(Microstepping should get you a bearable smoothness, even with quite crappy steppers)

[TerraHertz]

Ditch the local feedback and train it with a webcam?

Then you can't stand in front of the clock. The indexing ability needs to be there at least at every power-up. Also preferably as a sanity check all the time. Not to mention the huge complexity increment of including camera, image capture and analysis software (and a processor that can handle it.)

(Microstepping should get you a bearable smoothness, even with quite crappy steppers)

Then you can't stop applying power to the steppers when the hands are static. For the intended application, minimal power is a must.

There's nothing wrong with a couple of simple plastic gears on the hands shafts, and small worm gears on the stepper shafts. If they are mass produced.

[Precipice]

You only have to buy the webcam once, rather than fitting the indexing gear onto every instance, which makes it a little more appealing. Also lets you play fun games by drawing outlines of visitors

Making plastic gears isn't too hard - mill one, copy by moulding. Make it from toothed bar stock if you can't face milling teeth, or just buy and modify some gears, then copy them by moulding.
http://www.hpcgears.com/n/products/15.toothed_bar_stock/toothed_bar_stock.php
 The gears you'd need for this wouldn't be too hard, and if you were careful, you should be able to enforce the hands lining up with features on the gears, so you can do your indexing from the back. Just use plastic on plastic bearings, with a bit of plastic grease, for longevity and a bit of smoothness. So now you're down to 2 steppers, a pair of worm gears, a pair of big gears, maybe a bushing to get you through the clockface? Then the hands and indexing stuff. If you were making this in volume, the plastics would be well sub $1, including a frame to hold the steppers and a PCB.

[TerraHertz]

You only have to buy the webcam once, rather than fitting the indexing gear onto every instance, which makes it a little more appealing. Also lets you play fun games by drawing outlines of visitors

You're thinking 'one-off installation with many clocks, in a gallery', just like the Million Times video. But the application I'm thinking of is almost identical to a traditional clock. Needs to be mountable anywhere, by anyone, work for years, draw very little current, and be mass produced cheaply. There just isn't going to be any video camera. Let alone ability to mount and wire one up.
One solution to the indexing problem is just to have a second set of hands, inside the body. Then you only need one optical interrupter sensor, since each hand can be indexed separately while the other one is out of the way. Have both hands pass through the one sensor slot. Use a jig to fit the hands to the shafts in exact alignment front to back. It's a little bit annoying that the inside hands have to be mounted on the gears, and be profiled to have clearance for the worm gears, steppers, etc.

Making plastic gears isn't too hard - mill one, copy by moulding. Make it from toothed bar stock if you can't face milling teeth, or just buy and modify some gears, then copy them by moulding.
http://www.hpcgears.com/n/products/15.toothed_bar_stock/toothed_bar_stock.php
 The gears you'd need for this wouldn't be too hard, and if you were careful, you should be able to enforce the hands lining up with features on the gears, so you can do your indexing from the back. Just use plastic on plastic bearings, with a bit of plastic grease, for longevity and a bit of smoothness. So now you're down to 2 steppers, a pair of worm gears, a pair of big gears, maybe a bushing to get you through the clockface? Then the hands and indexing stuff. If you were making this in volume, the plastics would be well sub $1, including a frame to hold the steppers and a PCB.

The toothed bar stock is news to me, thanks. Didn't know such a thing existed. The rest- yes, agreed. Apart from injection tooling costs.

This project is one of many on my list to pursue once my workshop is completed. Hence not currently pursuing things like low cost suppliers of gears. Oh, and a 3D printer is on the tools list, for things like prototype frames to hold bushings, etc.

[Precipice]

One thing I often ponder, but very rarely seen done, is having a constantly rotating shaft, with clutches to engage or disengage for rotation.
You'd need proper position sensing (index pulse and quadrature) for each hand, but only a solenoid per, rather than a stepper. Two solenoids, if you wanted bidirectional. (I'm imagining you having a splined shaft with a load of spur gears, and pressing, or not, them against the gear with the hands attached.)

But little steppers are so cheap...
http://www.aliexpress.com/item/NEW-10pcs-Japan-Nidec-Mini-stepper-motor-micro-stepper-motor-2-phase-4-wire-D6mm-stepping/1612405669.html
or
http://www.aliexpress.com/item/10pcs-New-DC-3v-2-phase-4-wire-micro-step-motor-20-mm-Mini-stepper-motor/1799485746.html
if you need a bit more force.

[Mr Smiley]

I have several radio controlled analogue wall clocks, and they control the hands separately 

The minute hand only steps every 15 seconds, it's not a smooth transition, but a clunk for each 15 second step, the second hand moves independently from the minute hand when initially putting in the battery. On power up, it moves the second hand to 12, then moves the minute hand to 12 and the hour hand to either 3, 6 or 9, whichever is the closer in the clockwise direction. Then it sits there waiting to lock onto the minute marker and off it goes.

There unmarked so make is unknown 

 

[PointyOintment]

(Microstepping should get you a bearable smoothness, even with quite crappy steppers)

Then you can't stop applying power to the steppers when the hands are static. For the intended application, minimal power is a must.

The hands don't need to stop on a microstep. The only positions that really matter are 0°, 90°, 180°, and 270°. The angles for the clocks that aren't participating in displaying numbers don't need to position their hands at particularly precise angles.

[TerraHertz]

The hands don't need to stop on a microstep. The only positions that really matter are 0°, 90°, 180°, and 270°. The angles for the clocks that aren't participating in displaying numbers don't need to position their hands at particularly precise angles.

You're still thinking 'clock'. Take another look at the videos. When there are 'force field' type arrangements, they certainly do need precise angle positioning. Any that are even slightly off will stand out like dogs balls.
Same thing with any 'organic' style motion - the human eye and brain are very good at spotting discrepancies. Something that is slowing down to a stop (especially with 'S curve' deceleration) can't just suddenly glitch a little bit at the end to the nearest unit step position of a 1.8 degree or whatever stepper motor.
Anyway, enough of the details. The application is a cross between utility (clock, sometimes) and art. So the requirements are complex. Not worth going into further.