State of the clock after clockathon #2

[Mathieu Glachant]

First up, thank you Makerbot and thank you Bre for setting this up and hosting it. The people and the tools gathered in the Botcave could have solved any design challenge, and had a blast doing it.

Ben brought two printed clocks this time, and with a series of alternate parts printed over the course of the afternoon, we made some significant progress. Specifically, by 6PM, we had a clock prototype running smoothly with a power source able to run it for a day with a reasonable weight drop distance, AND with concentric shafts coming thru to the face of the clock to support both a minute hand and an hour hand. It also seemed very forgiving of gear sloppiness of various kinds, all of the sudden.

The main breakthrough came when Ben sggested we move the drum to the gear between the hour and minute gears. It solved both our power issue and reduced radial loads on the shaft bearing the hands.

There are still some rough edges to sand smooth, both in the code base and in the design, but for the first time, it really felt like we had overcome the major roadblocks. Left to do to solidify this new baseline:
1. Adjust the gearing to account for the fact that the escapement beats 30 times per rev (oops!).
2. Write clean code for the current non-ratcheting drum version of that intermediate gear.
3. Scale the clock up to use more of the available build area of a ToM.

Next design enhancements, in no particular order or priority:
1. Design a ratcheting version of the intermediate gear drum, preferably one that does not move the gears both before and after it in the train, and that has a smaller drum radius (less drop, but more weight). Alternatively, design a drum that sits on its own separate shaft and transmits power to the gear train from the side.
2. Adjust design to keep vitamin costs low, avoiding expensive or hard to source bearings in favor of skate bearings, or other alternatives.
3. Explore ways to make the printed concentric shafts wear resistant (brass tube inserts? Delryn?)
4. Make it pretty! Design a case that shows off the moving parts, particularly the escapement.
5. Scale it up even more to fit on a Replicator! Because size matters. :-)
6. ... Anything else you guys can think of?

[brep...@gmail.com]

Great list. So awesome to have people over and work with you all. This clock is really coming along!

Bre

Sent from my HTC smartphone on the Now Network from Sprint!

[Mathieu Glachant]

Here's some bad video of the clock ticking away...

http://www.youtube.com/watch?v=AH4Tg2tIgfg

[ruste...@prototribe.net]

That looks awesome! Congratulations!

Where are the files for that version?

Might I humbly offer the parametric ratchet lib for the ratcheting drum work if needed. Or a pointer to the code and i'll take a hack at it. =)
http://www.thingiverse.com/thing:8752

Also, i'm working on a lib to help carve up/create inter-lockable panels to more easily overcome size limitations of the build platforms.


Let me know if anyone wants an advance copy (OpenCSG rendering runs away with memory and hangs my machine if enough tiles are rendered at the same time. It can be used to both generate individual panels or generate a negative space block to carve up an existing object. Here's the sample usage:

panelsbysize(width, height, depth, panelwidth, panelheight, keysperside, gap, x,1);

This generates a grid of panels of panelwidth and panelheight centered across an area of width and height with a spacing of gap. The last two parameters specify which panel to actually draw in the grid.

Regards,
Rustedrobot

[Mathieu Glachant]

That's a really smart addition to the toolset, particularly the negative space cookie cutter version!

I had to do some truly dastardly things to the code on the fly during the Clockathon, so some cleanup is required, but I hope to be able to publish it soon. (Ben, I forgot to grab a copy of that folder off your laptop... Could you email me a copy?)

I was thinking that a regular pinion wheel where the little cylindrical spacer between the large gear and the pinion gear was replaced by the ratchet's inner toothed crown, with an outer ring with the ratchet arms bearing the actual drum, would be just the thing for the new ratcheting drum. The outer ring could either be slipped over the pinion, or if there is danger of it falling off, perhaps the two gears should snap together and lock the rachet ring between them?

On Sunday, February 5, 2012,  <ruste...@prototribe.net> wrote:
> That looks awesome! Congratulations!
>
> Where are the files for that version?
>
> Might I humbly offer the parametric ratchet lib for the ratcheting drum work if needed. Or a pointer to the code and i'll take a hack at it. =)
> http://www.thingiverse.com/thing:8752
>
> Also, i'm working on a lib to help carve up/create inter-lockable panels to more easily overcome size limitations of the build platforms.

> </mail/u/0/s/?view=att&th=1354f6cd19ac3e89&attid=0.1&disp=emb&zw></mail/u/0/s/?view=att&th=1354f6cd19ac3e89&attid=0.2&disp=emb&zw>

>
> Let me know if anyone wants an advance copy (OpenCSG rendering runs away with memory and hangs my machine if enough tiles are rendered at the same time. It can be used to both generate individual panels or generate a negative space block to carve up an existing object. Here's the sample usage:
>
> panelsbysize(width, height, depth, panelwidth, panelheight, keysperside, gap, x,1);
>
> This generates a grid of panels of panelwidth and panelheight centered across an area of width and height with a spacing of gap. The last two parameters specify which panel to actually draw in the grid.
>
> Regards,
> Rustedrobot
>
>
> On 02/05/2012 10:25 AM, Mathieu Glachant wrote:
>
> Here's some bad video of the clock ticking away...
>

> Clockathon #2 - The OpenSCAD clock

[Benjamin Fraser]

Saturday was really great, and thanks to everybody for making it so. I feel pretty lucky to have this thing sitting in my living room ticking away. It's definitely measuring time, just don't know how much yet :)

1) Over the course of today I've come up with a minimal case design which will expose almost all of the drive train assembly and escapement. Having the drum mounted on the lower shaft really makes this possible. I'll work this up in sketchup and share in a few days.

2) I've attached the working files from the clockathon. I know mathieu only has limited access to OpenSCAD this week but from the sounds of it, it's probably best to stay away from the clock lib until he cleans up the chainsaw masacre that happened to his code on Sat. 

3) Bigger! Of course bigger is better, but right now the fact that the gears are about 80% of the build area allows for the frame sides to be built in one piece. Scaling up on the ToM will require the frame to elements to be broken up a lot. Something to think about. But of course if you have a replicator......

4) Chimes, Cookoos, etc. I don't think we're jumping too far ahead to think of this now. As we hone in on a basic functional cloak that anyone can print and run I think it's a good idea to design for the future. Specifically to allow for a simple hourly notifications. Driving home I was thinking about what your could attach to the hour gear which might do the trick. Bre and others have suggested just mounting a little nub or tooth of some kind which can be accessed to push a marble or make another event occur. I'm a little concerned that this and similar strategy might cause a spike in power consumption by the grear train which could slow or stall the clock. I propose that instead we mount an oblong cam on the hour gear between the large gear and the pinion. This cam could have a gentle rise and fall, peaking once an hour. Mounted above this cam would be a vertical shaft with a bearing riding on the cam. This shaft would rise and fall once an hour. We could, of course, adjust the slope of the cam near the high point to still allow for a definitive moment where the action should occur. 

My thought here is that we could (as part of this release of the clock) create the interface where future functions could be placed. It may also create an opportunity for the thingiverse community to be creative, knowing that they could make whatever they want to attach to this shaft. We could also build available mounting points into frame so that accessories can be easily be added to the top of the clock. Personally I'm drawn to the steel marble run with bell at the bottom.

5) Rustedrobot has boldly gone where I thought we could never go: tiny teeth. I can not believe how small those teeth are! I brought home the parts and hope to get it together sometime later this week. This is definitely the future, especially as we peruse the Flavor Flav version. :)

[Mathieu Glachant]

Designing a stable interface on the hour gear for people to extend the clock is a good idea... Like a physical API, basically...

Your comment on the small teeth made me think... How small could we scale the clock and have it still work? Using pins or toothpicks as shafts, and printing the gears quarter sized and one mm thick, would it still keep time? :-) could we get it to the point where you print the clock in one single plate filling the build area, or two, and in a couple hours?

[ruste...@prototribe.net]

From some of my experiments, the teeth size is nearing whats possible with a 0.4mm nozzle. If I had to guess it may be possible to push it down to maybe 80% of current size, but that may be stretching it. Interestingly if we apply the small tooth size to your gear arrangements, a pretty darn small clock seems possible, easily driven via a motor and possibly via wound spring (though i've heard they can be dangerous).

I tried printing a half dollar sized gear initially and the end result was unusable as a gear, but not bad for tire treads for a wheel on a small car. =)

I like the cam idea, except cut off the second half. That way a spring loaded (possibly even just gravity driven) arm can be slowly pulled back over the course of an hour and when the cliff on the cam is reached, the arm swings back freely to its original position, or, even better a little futrher to strike a bell of some sort. A single cam cycle, attached to the minute gear could provide an hourly chime and I'd imaging the minute gear should still have a decent bit of power behind it. Alternately or in addition, the hour hand could have 1 or even 4 ramps (against a different bell?) to chime midnight or every three hours, etc...

[Mathieu Glachant]

I wonder how much travel is needed and whether we cold fit those cams between large gear and pinion? Great idea tho. :-)

[Benjamin Fraser]

You're absolutely right (Rustedrobot), slow and steady rise and sudden drop makes the most sense. And yes, there should be plenty of power on the minute hand.

Physical API: Yes! Exactly. I'll think about that as I make the minimal frame and try and incorporate a number of possible mounting points for future extensions.

I was looking at Rustedrobot's prints just now. It looks like the whole thing could fit within a replicator's platform. Between the tiny teeth and the integrated bearing concept we might have the very beginnings of a one shot printed clock with dissolvable support media. There is obviously a LOT to work out before something like that, but it is a very noble goal. In fact the collected intellectual assets accrued during such a goal would be a massive benefit to the community and open the doors to an unimaginable collection of desktop-printed mechanical wonders.

[Mathieu Glachant]

So many places to take this after we get a basic clock going. :-) I agree the one shot no assembly clock is a good long term goal!

Quick confirmation: resizing the clock at this stage will create problems for your current case design, correct? I should leave it t the current 80% for now?

[Benjamin Fraser]

Yes, I'd like to leave it at 80% for now. At least until the next clockathon. Partly for the case design, and also so that we don't introduce too many changes before solving these last few problems.

[Mathieu Glachant]

Ok. Side note, I'm moving the pinradius parameter to the parameters script rather the clockbuilder one, makes it easier to adjust.

[ruste...@prototribe.net]

Could be not much travel at all. Depends on where the fulcrum of the armature thats being displaced is and how much travel you need to accomplish whatever its supposed to do. I can easily see 5mm being workable under the right circumstances.

Also, I need to work out a bug in cutting the teeth, but I may have a solution for the drum attachment. Slanting the ratche faces can provide a way to capture the drum on the gear. The arms themselves will be trapped under the angle of the ratchet teeth! I'm hoping to have an updated version of the ratchet lib available within the next day or so.

[Mathieu Glachant]

Ooh, nice! I think I see what you mean and that could definitely work.

[Benjamin Fraser]

I like the idea of slanting the faces, would there be any danger of the drum sliding off during a ratchet action when the arms are sitting on top of the teeth? For example, if somebody winds the clock by pulling on the counter weight, but pulls at an outward angle instead of straight down.

As for the cam, I was thinking in the 5mm to 10mm range for total rise/fall. If it can handle a 10mm rise with a very small amount of force from the shaft I think people would find it much more useable in terms of making their alarms etc.

[ruste...@prototribe.net]

If enough force was applied in the right direction then it is conceivable to pop the drum off, although you also run the risk of pulling the line off the drum itself as well if you're pulling with a sharp enough angle. There are are a few ways to mitigate this:

• Position a guide hole for the counterweight line to prevent lateral tugging on the drum
• Somehow add structure around the drum to prevent it from lifting too far

The bug i'm fixing is related to the cutting of the teeth in the parametric version so we'll be able to try all sorts of angles. I'm thinking 10-15deg will be plenty.

[ruste...@prototribe.net]

The bug is fixed (re-learning Trig is fun!) There is an updated version of the ratchet lib available that adds support for the new parameter "slant" which will slant the angle of the ratchet teeth and arm faces. This should provide a basic interconnect for the two components with any number of arms beyond three.



http://www.thingiverse.com/thing:8752

[Mathieu Glachant]

Nice!

On Monday, February 6, 2012,  <ruste...@prototribe.net> wrote:
> The bug is fixed (re-learning Trig is fun!) There is an updated version of the ratchet lib available that adds support for the new parameter "slant" which will slant the angle of the ratchet teeth and arm faces. This should provide a basic interconnect for the two components with any number of arms beyond three.
>

> </mail/u/0/s/?view=att&th=13555d014fefc847&attid=0.1&disp=emb&zw>

[Benjamin Fraser]

Wow! Nice work. I'm super excited to try that out!