Hacker News new | past | comments | ask | show | jobs | submit login

There is a 'default' bucket (#7) to the side of the machine that catches everything that it does not pick out in that particular run.

The simpler solution was to just stagger the tubes a bit, that way all I needed to do was drill a few new holes. I see what you mean though, that's also a clever way but the 'feed' to the air valves is a single 6 way manifold that they bolt straight on to. It would be very hard to put a buffer between the manifold and the valves.




I see. So right now, if two pieces are too close together on the belt, they just go into the default bucket?

What is the limiting factor determining how many buckets you can operate with?

And I may have missed this detail, but you must perform a multi stage sort since you just have a few buckets? What is the level of the final sort, and what becomes of the parts at that time?


> So right now, if two pieces are too close together on the belt, they just go into the default bucket?

Yep. But this fortunately does not happen often.

> What is the limiting factor determining how many buckets you can operate with?

Belt length, measurement precision during the stitching, cumulative errors in the math determining how far the belt has moved.

> And I may have missed this detail, but you must perform a multi stage sort since you just have a few buckets?

Yes.

> What is the level of the final sort, and what becomes of the parts at that time?

A single part category in multiple colors, or and that's the next level, all (or at least most) of the parts required to make 6 sets + remainder.

That final stage is a bit tricky, it also interfaces with a database that keeps track of the lot id of that set and what parts are still missing. I have most of the pieces for that second idea ready now but they are not yet tied together.


A simpler way to maintain steady pressure with simultaneous puffs would be to use a oversized manifold, and put the valves on the end of the copper air lines, perhaps mounted on a board or something. Maybe you could gang up a number of puffers for larger parts.

And what about doubling the bins by placing them on both sides of the belt, with puffers blowing in both directions?

Or a laser trip sensor to get a more precise belt location measurement?

Fun project!


> A simpler way to maintain steady pressure with simultaneous puffs would be to use a oversized manifold, and put the valves on the end of the copper air lines, perhaps mounted on a board or something. Maybe you could gang up a number of puffers for larger parts.

That's an option. But the problem is solved so that part is ok now.

> And what about doubling the bins by placing them on both sides of the belt, with puffers blowing in both directions?

It does that, you can see it in the video if you look closely and also in one of the pictures. There are two bins at the back and four in the front (that's the easiest way to fit them right now).

> Or a laser trip sensor to get a more precise belt location measurement?

Or a gray code printed on the belt! That would give you absolute positioning.

> Fun project!

Absolutely!


A laser per bucket could be integrated with no change to existing software. Just AND the valve signal with the laser detection signal.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: