LitePlacer.com forums

I've been doing quite a bit of tweaking, 3D printing tests, and then more tweaking. The 3D printing is a little fussy as I need 0.1mm layer heights, and my printer doesn't have automatic bed leveling. But fortunately they're good enough for me to do some testing and discover issues. Here is the latest version of the design:

mrandt wrote:Third option I could think of would be to have a "snap on clip" (like for the sprocket hole index pin) on both sides of the feeder; this clip could also hold fiducial markers. So front clip would have a tab for QR code and fiducial, back clip only fiducial.

I've added the clip on the right side that has a pin like the clip on the left, as well as a flat area for a QR code. Because it has a pin, it will align a long tape between the sides. This QR clip currently isn't something that is easy to injection mold, so I'm still thinking about the best way to build this. One idea I have is to make this clip out of two parts that can be glued together with a solvent weld. Both parts would be very easy to injection mold separately.

I've also been doing some work on the alignment/hold down system. Previously I had alignment pins and tabs on the bottom of the hold-down bar. However, you couldn't see where they were when trying to put the feeders in place. So I changed to having the "pins" on the top, as you can see here:
And here is a bottom view:
I'm pretty happy with where this is at. I'm getting really close to being ready to send it off to the mold maker for a quote. Currently I'm thinking of having the hold-down bars 3D printed to reduce mold making costs. After all, two hold-down bars will hold 10 8mm feeders, so the production volume would be a lot lower on the hold-down bars.

mrandt wrote:I'd vote for black side walls, spacers and clips - just because it looks better. Only the sliding support rails should be green

I'll need to see what the mold costs will be. I want the inner parts to be a medium color so there is contrast with both black and white tapes. Having the outer parts be a different color, like black, would look really cool. But would require a second mold. I'll have to check with the mold maker to find out what the cost difference would be between one and two molds.

mrandt wrote:When I designed my feeder, I had some trouble with the tolerances of the rails which are meant to hold down the tape.

Currently I'm getting a very good grip, so I'm not sure I need the cutouts. On the other hand, it would certainly increase the grip, so I'm might give that a try. One advantage of injection molding is you get much better tolerances than with 3D printing. I also have my 3D printer dialed in really well (except for the first layer), so I'm getting good results for my test parts.

-- John

Hi John,

the latest design is looking really good - two thumbs up

Just for clarification - can the clip at the back (right) slide a bit back and forth to accomodate for sprocket hole spacing tolerances in tape (see picture)?
The QR code and / or fiducial holders could be a 3D printed (or molded) "add on". Not everyone will need it; me for example, I'd be happy to print my own.

I'd say make the clip with a flat surface and wide / thick enough, so one can glue / cold weld the top part onto it. This is something that the end customer could easily do.

It might be a smart idea to include a small cutout (1,5mm wide) exactly opposite of the index pin - so it is easy to align the top part correctly; especially if that contains any fiducials.
The new design for the alignment / hold down rails is actually very easy to manufacture now. Just separate the top part from the bottom.

I'd simply laser cut the the top plate from acrylic or aluminuim; CNC milling would be another option. Then, when attaching it to the table top, put spacers underneath the rail and you're good to go. What is the current spacing between surface and holding rail?

It is simple and cheap to outsource laser cutting. Juha has all the steel plates laser cut; I am pretty sure he could also get a good deal on aluminium. 1,5mm should be strong enough.

This way, you could also offer different lengths (like 6 modules, 12 modules, 18 modules) without much extra effort.

Thanks and best regards
Malte

mrandt wrote:Just for clarification - can the clip at the back (right) slide a bit back and forth to accomodate for sprocket hole spacing tolerances in tape (see picture)?

Yes, I forgot to mention that the QR tab/pin does slide to allow for small variations in the hole spacing, as you suggested I do. Right now it can slide about +- 1.25 mm from the expected center location of the hole in the tape.

mrandt wrote:It might be a smart idea to include a small cutout (1,5mm wide) exactly opposite of the index pin - so it is easy to align the top part correctly; especially if that contains any fiducials.

That's a good idea. If I made the QR board as a separate part, it would be really easy to add a circular indentation that you could fill with a little white paint in order to make it more visible. Thus, it could act as a fiducial. I'll work that into the design.

mrandt wrote:I'd simply laser cut the the top plate from acrylic or aluminuim; CNC milling would be another option. Then, when attaching it to the table top, put spacers underneath the rail and you're good to go. What is the current spacing between surface and holding rail?

The current spacing is 2mm. Going with laser cutting makes sense. Juha, if you're reading this, what do you think of that idea? And what would be a good thickness for the sheets? Two pieces aligned with the hold-down screws in the slots would seem to work well. I'll draft up a version of that design.

There is one last thing that I'm thinking about. Right now the maximum part height, even for 24 mm wide tapes, is 10 mm with this design. The distance from the bottom to the top of the tape is currently 20 mm. I'm thinking of increasing this height to 22 mm, which would then allow 12 mm high tapes, which is the maximum allowed for 24 mm wide tapes. Any objections to adding 2 mm to the height?

[quote="JohnSL]The current spacing is 2mm. Going with laser cutting makes sense. Juha, if you're reading this, what do you think of that idea? And what would be a good thickness for the sheets?[/quote]
Of course I'm reading this. And looks very good. Laser cutting makes sense, and sheet thickness doesn't matter that much, as long as it is stock material, which means even half mm's. There is some advantage if both sheets are the same thickness. Also, thin steel is cheaper that thick, but I'm nitpicking.

20mm vs 22mm height: No objections.

JuKu wrote:Laser cutting makes sense, and sheet thickness doesn't matter that much, as long as it is stock material, which means even half mm's. There is some advantage if both sheets are the same thickness. Also, thin steel is cheaper that thick, but I'm nitpicking.

OK, I'll continue down the route of laser cut parts, then. I don't have access to a metal cutting laser, so I won't be able to test tolerances. Is that something you can do?

There are two designs I have in mind. One design is to have spring pins to hold the two pieces together, which would look something like this:
Both parts are 2 mm thick, and you can see that they both have oval slots. That allows you to loosen the screws in order to add or remove feeders. The other option I thought about, but haven't drawn up, is to have a circular hole in the bottom plate and an oval slot in the top plate. That way only the top plate would move. Any thoughts on the design?

Finally, were you thinking of using steel for these or aluminum?

I did some work on a two-part QR plate with fiducial. Here is the new design:
Both parts can now be injection molded for better precision, and then glued together. I extended the bottom alignment pin out the top so that it also acts as an alignment pin for the plate, and the hole can also act as a fiducial mark. You could paint the pin white before gluing the two parts together. If both parts are molded out of black, you'd have a nice white circle inside a black hole. I think I might need to make the top pin a little longer so that it extends all the way above the plate in order to reduce the impact of shadows from uneven lighting. A 12 mm square micro QR code can fit to the right of this pin without covering the pin.

The top pin is not flat on the top because of injection molding limitations. I need it to be curved to make it easier to cut the mold. My goal with the mold is to avoid using EDM and very small cutters, as I think avoiding both will help keep the mold making costs down.

JohnSL wrote:Both parts are 2 mm thick, and you can see that they both have oval slots. That allows you to loosen the screws in order to add or remove feeders. The other option I thought about, but haven't drawn up, is to have a circular hole in the bottom plate and an oval slot in the top plate. That way only the top plate would move. Any thoughts on the design?

Finally, were you thinking of using steel for these or aluminum?

How about: Put the indexing on the bottom plate. That way, the holders will always sit precisely in the same place, the software does not need to measure the location and the placement is faster. The bottom plate then has holes for screws to fix it on the table. The top plate has openings for these screw heads, so we don't need countersunking in production. The top plate has oval holes so it moves; the bottom plate also has tapped holes for thumb screws that hold the top plate down. And therefore, the plates are steel, so the threads in the bottom plate are solid.

Edit: I talked to my plate supplier: For M4 thread, they recommended 3mm plate thickness.

I have no objections to 22mm total height either - I will need to change my machine anyways to support more Z-clearance and higher pickup / placement level.

I disagree to putting fiducials (only) on the rails for the following reasons:

- They will be at a much lower Z-level, so they might be out of focus and hard to recognize

- Because of the different Z-level and projection error, any measurements taken by camera will be off by orders of magnitude; assuming camera is calibrated for pickup Z-level

- Just having fiducials on the rails does not tell the machine where the individual feeders are; you can only calculate where they might be

- You will still need the QR code on the feeder itself to identify specific feeders when setup has changed

- Having additional the fiducials on the rails does not make sense either, as the rails will be in a fixed position anyways


I like the index pin "poking through" the top plate though. May I suggest to just extend it even further, so it goes a bit beyond the plate surface? It will be very easy to cut flush with an exacto knife once the QR code plate is mounted / glued in place.

In case you manage to mold two different colors materials, the clip should be made from the "medium colour" plastic (green, blue) and the plate from black. That way you would get a very precise and good contrast fiducial without fiddly paint jobs

I can only confirm maltes arguments.

Having worked on video processing for the last couple of weeks I found it extremely important to have every object used for position recognition (zero mark, fiducial, tray fiducial, tapes, feeders and trays) on almost the same Z height. All cameras I am using (Juhas 10mm cam , logitech 270) have a large diverging view field. 5mm hight difference already lead to unusable measurements.

Talking of fiducials: try to place them as far away as possible from eventual other circles of the same color (black on white / white on black) like tape holes because other circles can be recognized within the fiducial view port. One of my PCB has some vias close to one of the fiducials which led to some wrong fiducial measurements.

mrandt wrote:I disagree to putting fiducials (only) on the rails for the following reasons

Manfred and Malte, I believe Juha is simply addressing the design of the hold-down bars, not the fiducials. The idea is that if the hold-down bars are at well-known locations and hold the feeders in place at precise locations, this is another approach to having the machine know where the feeders are instead of fiducials. So it provides more options for the user.

JuKu wrote: I talked to my plate supplier: For M4 thread, they recommended 3mm plate thickness.

I like the idea of a thumb screw. My current design is for M3 screws. Is there a reason you're thinking of M4 instead of M3? Or would M3 work just as well.

I'll work on changing the design again tonight.