Assembly line feeding system

[John Boozer]

What I'm trying to do is inspired by the feeding mechanism of
http://ricquin.net/lego/laurens/bricksorter4/index.htm

I'm trying to feed 1x1 plates into an assembly line.  I've built several hoppers
to hold a few hundred pieces each.  I've got a chute design so that the pieces
slide down without room to turn or flip.  I'm just having trouble getting the
pieces from the hopper into the chute.

My plan was to use a tank tread to dump a few pieces at a time into some system
to get the piece oriented to slide into the chute.  But so far the stuff I've
come up has a problem with pieces getting stuck.

Anybody have any ideas/suggestions?  Anybody know of anybody doing something
similar?  Or maybe an example of someone doing something like this in a real
world factory?

My chute design at the moment has a technic brick stacked on a plate stacked on
a technic brick for the side walls.  Technic beams form the ceiling and floor,
They're connected to the walls via the technic bricks.  2x3 plates are used on
top and on bottom to make sure that walls are just the right distance apart,
otherwise stuff gets stuck.  The tolerances are pretty picky.  The details are
bound to change, but if you can decipher what I said, that's the size of the
target I'm aiming for.

John Boozer

[Matthias Paul Scholz]

Maybe a look into actual GBC designs and their approaches to transportation
might prove useful (?).
(Ok, it's for LEGO balls instead of plates)

http://www.greatballcontraption.com/

-----Ursprüngliche Nachricht-----
Von: news-gateway@... [mailto:news-gateway@...] Im Auftrag von
John Boozer
Gesendet: Freitag, 27. Juni 2008 20:46
An: lugnet.robotics@...
Betreff: Assembly line feeding system

I'm trying to feed 1x1 plates into an assembly line.  
[...]
I'm just having trouble getting the
pieces from the hopper into the chute.

My plan was to use a tank tread to dump a few pieces at a time into some
system
to get the piece oriented to slide into the chute.  But so far the stuff
I've
come up has a problem with pieces getting stuck.

Anybody have any ideas/suggestions?  Anybody know of anybody doing something
similar?  

[Jordan Bradford-2]

GBC module creators tend to independently reinvent the same hopper solutions
used in factories. Some of them shake, many of them have an agitator spinning
inside it to loosen the pieces, but the simplest just use gravity and a change
of direction.

The third one requires some explanation, but it's very reliable.

GBC modules that jam when too many balls arrive are always trying to funnel many
balls into a narrow output chute. If two balls arrive side by side, they block
each other from entering the chute, and any subsequent balls pile up in the
hopper and eventually overflow.

The solution is to change the direction the balls are heading once they enter
the hopper.

Make your hopper a wide ramp that slopes toward a second, narrow
one-ball-at-a-time ramp that is placed 90 degrees (perpendicular) to the wide
ramp. The incoming balls will roll en masse into the narrow chute on its long
side; they will no longer be trying to roll into the narrow chute one at a time
from its upper end. This arrangement can still overflow its hopper if the lower
end of the chute isn't dealing with the balls fast enough, but it won't jam --
the first ramp is too wide.

Moving 1x1 plates with just gravity won't be as easy as soccer balls unless the
ramps are pretty steep, but I think you can pull it off. You could still use a
tank tread as the narrow "chute"; you just need to have the hopper dump the
plates onto it sideways instead of from the end.

[Philippe Hurbain]

> The solution is to change the direction the balls are heading once they enter
> the hopper.
>
> Make your hopper a wide ramp that slopes toward a second, narrow
> one-ball-at-a-time ramp that is placed 90 degrees (perpendicular) to the wide
> ramp. The incoming balls will roll en masse into the narrow chute on its long
> side; they will no longer be trying to roll into the narrow chute one at a time
> from its upper end. This arrangement can still overflow its hopper if the lower
> end of the chute isn't dealing with the balls fast enough, but it won't jam --
> the first ramp is too wide.
>

Hello Jordan,

Thanks a lot for the tip! This is an amazinly simple and effective way to
serialize a ball flow. According to my experiments, there is a little thing to
add to your explanations: the narrow ramp level must be sufficiently lower than
the base of the wide ramp, or ball arcs can still build up.

Philo

[Jordan Bradford-2]

In lugnet.robotics, Philippe Hurbain wrote:
> Thanks a lot for the tip! This is an amazinly simple and effective way to
> serialize a ball flow. According to my experiments, there is a little thing to
> add to your explanations: the narrow ramp level must be sufficiently lower than
> the base of the wide ramp, or ball arcs can still build up.
>
> Philo

Thank Rafe Donahue, actually. I learned about it from a math/statistics-based
presentation about GBC modules he gave at BrickWorld 2007. If he didn't discover
the technique himself, he at least documented it. :)

I haven't seen a lot of modules use this technique, at least at BrickFest 2005,
2006 and Brickworld 2007, 2008. The modules that cause balls to change direction
don't usually do it in their input hoppers, which is where the technique does
the most good. Therefore, I think Rafe probably did discover this by himself.

I hope more people learn about this technique, because jammed input hoppers tend
to cause problems with the rest of a module, especially when they have things
like chain lifts.