Supporting the shafts between the gears?

[Mathieu Glachant]

So we have reports that the M3 shaft is flexing pretty badly due to the weight hanging off the drum. Ben is switching to M5 shafts, which will probably help, but perhaps all we need is to support the shaft just behind the hour gear, with an extra piece of frame and a bearing.

Looking at the clock configuration, I think we could turn it on its side (the two gear stacks side by side, not one over the other) and have the frame made of a series of vertical walls at each successive gear position, rising to the height of the shafts. We could carve out space for the gear present at that slice, and have a holder for a ball bearing to hold up the other shaft.

No flexing of the shafts, and it would look really nice with the upper half if each gear fully exposed!

[Mathieu - Syvwlch]

There's a negative space option in the clockbuilder script, so this
might actually not be that hard to design... I'll take a crack at it
after work today.

[ruste...@prototribe.net]

This also has the additional benefit of ensuring the gears can't rub
against each other! I was planning on moving in that direction once I
had my prototype working and have even started it with the escapement
wheel since the screw can't travel all the way through due to another
gear being in the way. This also can afford additional flexibility in
gear arrangement.

[Mathieu Glachant]

Beside, we need a printed frame/case for this thing! :-)

[Benjamin Fraser]

This all sounds good. Sorry for not being in the discussions much today, pretty busy over here. Finally got my bot back online and am now printing the first of the new gears for the 5mm version.

I agree about the need to have a printed frame. I've always resorted to a quick wood jig because it's hard to get machine time for it (my ToM runs almost 24/7 these days) and it's super quick to whip one up in my shop. Need more printers :)

I'm looking forward to this 5mm version. I hope it will answer a lot of questions and eliminate a number of problems. Once tested I think we'll have a lot of information about how to proceed.

Is the farm printing Rustedrobot's design? I'd love to see it.

That's all for now...

[Mathieu Glachant]

No problem generating the negative space version of Test Jig 4, with
6mm shafts (so we can use the bearings to support them). However, when
I import the resulting STL and difference from a cube, I can F5 anf
F6, get no error message about non-manifold shapes, but when I try to
export as an STL, OpenSCAD crashes and the file is generated empty (0
bytes).

Might just be something about the server I'm running this on (Ubuntu
11.10 64 bit) and it's not a complicated script to run, so I've
attached it to this post, along with an image of what I can see on my
screen before I try to export the STL.

Anyone who wants a go is of course welcome to try with other software,
starting from the negative space STL included here. I will also upload
this to Thingiverse.

[Mathieu Glachant]

http://www.thingiverse.com/thing:16973

[Tony Buser]

Crashing for me too on OSX.  The problem appears that OpenSCAD doesn't like the jig stl file. (it crashes if I just import it and try to render to stl)

When this happens I resort to Blender's boolean operations.  It's slow and makes bad stl, but netfabb usually fixes it.

I've attached the frame I made in blender.

--
NOTICE: This email may contain information that is confidential or attorney-client privileged and may constitute inside information or trade secrets. The contents of this email are intended only for the recipient(s) listed above. If you are not the intended recipient, you are directed not to read, disclose, distribute or otherwise use this transmission. If you have received this email in error, please notify the sender immediately and delete the transmission. Delivery of this message is not intended to waive any applicable privileges.

[Mathieu Glachant]

Thanks! The STL looks great in meshlab. :-)

[Benjamin Fraser]

Thanks Tony, This looks pretty good. The only issue I see it that there is no allowance for the escapement and pendulum to swing. I could simply cut away the extra plastic if necessary. I'm still printing gears right now so not sure if I'll be able to get this in before tomorrow. But it's possible.

Another thought, It's really helpful to have some windows from the tops and sides to be able to see what's happening. Or more likely, what's not...

Thanks!

<frame.stl>

[Mathieu Glachant]

Yeah, it's definitely a proof if concept at this point. I only just realized that if all the gears are on bearings, there is no reason to put the shaft on bearings too... So I should have used 3mm shafts, not 6mm.

It needs extra work to make allowances for the string off the drum, the pendulum as you pointed out, and the shape could be further slimmed down, I agree.

[ruste...@prototribe.net]

Looks like it may be a cgal bug. This is what I get when trying the render.

openscad
Xlib:  extension "RANDR" missing on display ":0.0".
 sorry, this triangulation does not deal with
 intersecting constraints
CGAL error: assertion violation!
Expression : false
File       : /usr/include/CGAL/Constrained_triangulation_2.h
Line       : 620
Explanation:
Refer to the bug-reporting instructions at http://www.cgal.org/bug_report.html
Aborted (core dumped)

[Mathieu Glachant]

Good thing Tony found a work around, :-)

On Friday, February 3, 2012,  <ruste...@prototribe.net> wrote:

[Benjamin Fraser]

I'm not going to pretend I know what any of that means. :)

[Benjamin Fraser]

I'm looking at these gears which are printing right now with the larger hole for 5mm bearings. The bigger hole is MUCH better quality than the previous smaller ones. Much cleaner, more circular and a much better snap fit. I think there are going to be a lot of benefits in terms of keeping the gears properly registered when under strain. Especially with gears 1,2 and 3.

Also, I've been letting these cool a full 20 minutes (as opposed to about 10 before) before pulling them off and they seem to have less wobble, I'll test that later.

[Mathieu Glachant]

Hum... Having slept on it, I now remember that last time I was trying to generate frames using negative space STLs of the whole clock, I had to clean up the file before importing it. I mostly used a web service, the name of which escapes me at the moment.

Sorry, slipped my mind yesterday!

[Mathieu Glachant]

Can't wait to see the results!

[Mathieu Glachant]

Netfabb!

http://cloud.netfabb.com/