some thoughts about pick & place

Since I dont know where to put it, I put it in general...

I was looking at the conversation about SMD, and I was wondering why we need a pick & place tool for electronics, if we can use the same system for bolts & nuts.

The machine would just pick up a bolt from a small production line, put it in the plastic and you got a bolt in plastic. I think if you have some sort of changable toolhead you can even get a screwdriver to do the work for you.

*future mode on*

4 or 5 repraps linked together. Every reprap has his own toolhead. A conveyor belt or some sort of positioning system is used for movement of parts. In the end the conveyor belt puts the partly finished parts on a tray and a mechanical arm or operator just needs to put the parts together, a job that takes less than 5 minutes. The operator just needs to feed the machine with new parts, while the machines do the job.

*future mode off*

Also thought about a mechanical arm for putting pieces together, but that is a bit too much I think for the versions we have now (but it can be done, i'm sure. We need some plastic bits and some servo's for controlling the arm).

It felt a bit like the matrix... If my future mode comes true, then we need a robot for the mailman, opening packages and giving every reprap the stuff that it needs and I will just relax and be the powersource.

What you guys think?

[www.archive.org]

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Forget putting RepRaps together, I want to have it assemble LEGO.

That's a very interesting thought: building captured nuts, washers or even bolts into parts could be very handy when it comes to assembling things, especially for complicated structures.

One thought on captured bolts: if you put a bolt in and it sticks up above the work, won't the extruder head hit it as it tries to fill the surrounding area?

Pick and place is either trivial or incredibly hard, depending on the parameters.
If you have known objects in known positions and orientations being moved to other known positions and orientations the programing is simple and the important variables are the tolerances of your robot and final product. This type of machine is great for mass production but of little use for reprap since we are making individual items or short production runs. The effort involved in loading 15 different electrical components into the proper pick up sites in the proper order and amount so that a robot can place them into a circuit board for soldering is the same as or greater than just putting them in the circuit board yourself, without even considering the software necessary to tell the robot where they are and where they go. Alternatively you can build a robot with access to an automated parts warehouse containing thousands of types of resistors, capacitors, IC's, nuts, bolts, etc. but again that sort of defeats the purpose of desktop manufacturing doesn't it?

On the other hand a pick and place head able to select the needed items from lightly sorted collections (such as a box of assorted resistors) would be incredibly useful. It would also be stronger AI than is available outside of university labs and supercomputers; and resistors are easy.

Not that I am opposed to pick and place functions, I just think we need to be very clear about what we want them to do. A head which places captive nuts at the proper time and location would be a boon to the production of Darwin brackets. On the other hand putting a resistor in a circuit board involves not only getting the right resistor, but also bending the leads properly, orienting the bent leads to the correct holes, placing the resistor correctly with the proper force, bending the leads again to hold it in place and responding correctly to any failures in this process. And resistors are easy. Diodes and capacitors add a whole new layer of complexity as they are polarized. I find myself drawn to the idea of printing resistors by using precise lengths of graphite loaded epoxy.

I believe that the goal of RepRap is not to recreate the factory with it's collections of machines, conveyors, and drone-like human tenders. I believe the goal is to allow people to create the implements of their daily lives to suit themselves, either locally or individually, from raw materials with minimal training, drudgery or capital.
Pick and place certainly has a place in that idea but it is in competition with other functions for the available developmental resources of this group, (as are all other functions, such as paste/ceramic extruders, milling heads and 3D plotters) and I think its utility would be relatively low for the foreseeable future.

BDolge: Great distinctions about the "simple" vs "complex" forms about pick-and-place.

As I was reading your post, I had a brainstorm. Perhaps this isn't a new idea, and has long been assumed, but at least the idea was new to me.

I was thinking about the SMT machines we have here at work with the reels and feeders and whatnot, and how much of a pain it is to reload them and customize them for every part.

Then I thought, what if part of the RepRap build process was to print a custom "tray" for each circuit board that could be filled by the user with all of the SMT components? This would take the place of the feeder reels of a traditional SMT machine, and would not only assist in letting the machine know exactly where every part is, but would also help the user with inventory and orientation.



Much the way that the pulleys forms were built to assist in a manual build process, these plastic trays could be made individually for each board to be built, and could make loading the machine much easier.

I realize that it's still tedious, and you still have polarity issues for LEDs, caps and whatnot, but the raised edges of the tray could help with ensuring that they sit reasonably accurately, and are located precisely in the places that the RepRap would expect.

I realize this doesn't solve all of the problems, but I think that it helps make the idea of general-purpose pick-and-place to be a little more feasible and reasonable in my mind.

--clint

Edited 2 time(s). Last edit at 01/22/2009 11:53AM by HanClinto.

I was thinking the same sort of operation for the feed tapes though I would extend the above board with a guide to butt the board being build up against; that way positions are relative.

Though pick and place is great for repeative; but the only boards we make mulitples of are the stepper.

On components: buying a whole reel of bits is going to be expensive! Many distributors (I use Digikey) will supply 'cut tape' - i.e. a short length of tape with, say, 10 components on it. These short segments could be stuck down onto a base board to avoid having to develop tape dispensers.

I am not sure that pick/place is absolutely necessary unless we're doing large runs. I think that it's more important to get paste printing in place. I'm not sure whether oven or heat wand is the way to go for soldering.

I dont think this will ever be a "Core" function for darwin, but the "pick & place" tool is listed as something that we can maybe use in future development, and I am willing to test if thats really possible or that we can scrap it as "not for homebrew".

There are some things I really like to try/find out, if Mendel does get to support multiple heads like a CNC milling machine:
- Grabbing an electronic component and releasing it
- Methods to do this (either with vacuum or clamps)
- Bending leads

About the sorter problem of BDolge:
You have a round spinning disk inside a small tube where all the resistors go in, and a opening at the bottom where the correctly placed resistors go trough. The robot checks the color, picks it up and places it in the right place (or puts it back in the tube if incorrectly). Total AI? None. Commonly used for realligment of bolts.

About the polarity issues of HanClinto:
If you have a non-changable head then the position that the SMT is picked up will be the same that the SMT has when dropped. If you have a certain polarity, all you need to do is just putting the right polarity on the right side of the board.

Small edit on soldering in ovens:
You need to get at least 183C of heat to make sure that solder melts... You even need to get a higher level of heat to make sure that the RoHA-compliant solder melts... I am currently waiting to experiment this (the oven in our dorm needs to get broken, i hate the smell of burned pizza) but at this moment I am not in possession of a reprap...

2nd small edit on large runs: Guys like bushing from hackmii started to solder their own stuff in their kitchen, when orders of 50-60 per day came in and they couldnt handle the pressure...

Edited 2 time(s). Last edit at 01/22/2009 03:00PM by Mr. Seeker.

@freds: Great idea about having a slot on the tray for the board to slide into or to rest in -- that's a really good idea!

Also, I hope that we'll see more than just a single RepRap produced with a single printer -- at the bare minimum, we should see each RepRap producing three copies of itself -- one for spare parts, and two to give away. I realize that three boards isn't a /huge/ run, but it could be nice to have the RepRap be able to populate a board for you. This would of course be an optional step in the RepRap process -- a person who preferred to do it by hand could skip this part, but for those who wanted the assistance for volume or accuracy (BGA anyone?) then they could of course use it.

@Mr. Seeker: Sorry for being unclear -- I meant the polarity for the user to load the unit. If the user puts the caps into the tray backwards, it will just be picked-and-placed backwards -- the RepRap in my design wouldn't know that it needed to be turned around.

What I've been looking up specs to draw is to put a row of posts along the side of the build area - It would look rather like a big row of lego - with the same pitch and hole dimensions as standard tape for SMT. What I don't know is the diameter of the actual holes themselves - I do know that they are on a 4mm pitch. The parts already come coarsely aligned and positioned in that tape, and small amounts are sold as a strip of it cut from a roll. After the user fits the tape onto the specified posts, a bar would come down and pinch them down to constrain the tape (hopefully this provides enough constraint to let the machine index and grab them okay). If you want to be more slick, let the user set positions of what they've added, so they can leave the resistors in the machine

Then use a vacuum wand to place the parts, pushing them against the side of a box on the build platform to position them close enough for soldering to handle the rest. Very little intelligence to handle such a system, and it's compatible with and uses the same tech as industry.

Reading the past work that adrian did to make a simple robot that had RP'd circuits in it, they extruded the traces, press-fitted the chip, and soldered in the (non SMT) diodes. Seems clumsy to me - certainly better that having to mail off for a pre-printed board. Still doing all the soldering by hand. Hardly ideal for the "makes anything" printer!

I've thought that the "final" version would be able to embed the SMT components (both chips and simple resistors/diodes) in the plastic, with traces in true 3d. Soldering in place would be done at the same time as the traces themselves are extruded.

I'd assumed that you'd keep it loaded with a set of standard components, resistors, crystals, caps, processors, and so forth. Then for each project, you'd load a few specific vitamins (IR diode for the "TV-B-GONE", etc) and hit "print" and not worry too much about the common stuff. After all, there are a few things that almost any project is going to use - resistors, capacitors, etc. and these would be kept loaded as a matter of course.

---

--
I'm building it with Baling Wire