LitePlacer.com forums

I'd like to start a conversation about tape strip holders. I have been working on 3D printed holders and have just released the files: https://www.youmagine.com/designs/smd-t ... -16mm-24mm. You can also find the discussion on the OpenPnP forum here: https://groups.google.com/forum/#!topic ... PgTmROUzns.
A couple of weeks ago, Juha reached out to me with interest in making an injection-molded version of this strip holder. I've made my own molds and parts, for example: http://trains.socha.com/2012/10/making- ... t-iii.html. This is done with a desktop CNC to mill the molds, and a desktop injection molding machine. So I offered to design the injection molded version. Here is what I have so far:
I'd like to get some input before I get too far in the design. One suggestion that Juha had was to make the design modular, so you can have short segments for infrequently used parts and longer segments for frequently used parts (Juha, please correct anything I got wrong). My current design has the following characteristics:

• Top of the tape is 1 inch from the bed (no matter what tape thickness)
• Each segment is 70 mm wide
• Segments can be connected together (I'm still working on this part) to form longer segments
• The sliding clamps apply pressure to the bottom of the tape, so the tops of tapes are all at the same height regardless of tape thickness
• There is an internal spring that pushes the inner parts sideways, which then apply uniform pressure from below, firmly hold the tape in place
• Supports a maximum tape height of 12 mm, which therefore works for all tapes from 8 mm wide to 24 mm wide
• The same parts are used for different tape widths, but with different spacers. That means we don't need a new injection mold for each tape width
• Will be made with some ESD friendly plastic

Here are some things I'm still working on:

• The spring mechanism. I feel I'm really close, and need to order some parts for testing
• Holding down to the machine. I haven't solved this issue yet. On my 3D printed version, I'm using magnets, and then connecting a bunch together with threaded rod because the magnets are not strong enough to hold a single feeder in place

To get the discussion going, here are some questions I have:

• What is a good segment width? Is 70 mm a good width for each segment?
• What is a good height from the hold-down surface to the top of the tape? Is the current 1 inch OK?
• Does anyone have any good ideas for how to hold the feeders in place?

Thanks,
-- John

Hey John,

I have been following the discussion on GGroups from the beginning - thanks for also posting here on LitePlacer forum.

You may have seen my cut strip feeders before:
http://malte-randt.de/strip-feeders-and ... y-for-smd/

I've downloaded your STLs from YM and just started printing the 12mm version to give it a test

What I like about your design is that strips seem easier to load. With my feeder, the screws which hold down the tape need adjustment from time to time.

Also, I need different versions for paper and embossed plastic tape - which is not necessary with yours.

When I designed my feeder blocks, one important aspect where the tray plates which allow to configure one tray for a given job with all the parts, be able to refill outside the machine and then replace the tray on the machine. This is a modularity I miss in your design.

On the other hand, if reloading is easier and feeders can be swapped quickly, this might no longer be necessary.

Only peeling off the cover tape is probably fiddly with any strip feeder.

Last not least, your design might be easier to mass manufacture. My feeder blocks are easy to mill on a CNC, but yours are much better suited for injection molding.

That being said, this is not about which design is better. Instead I'd like to offer to join forces when working on a productized strip feeder.

Can you share more views of your new design? Or are you willing to share the source files? What CAD are you using? Please send me a PM if you don't want to share these publicly (yet).

I thought about mounting the feeders:

Magnets are relatively expensive, might not be strong enough and could interfere with parts.

"Lego like" mechanisms are difficult to make, as they require small tolerances. They also wear out.

Thus I came to the conclusion, that a purely mechanical approach might work best.

Tried to draw something in CAD already:
The idea is that feeders are locked into place using two simple rails (e.g. milled from aluminium, or laser cut from two sheets). One is fixed, the other can slide back and forth when unscrewed. Instead of normal screws, I would probably use a small turning knob but did not have a model readily available in my CAD.

This would allow to quickly replace feeders and support any arbritrary combination of feeder widths.

My other remarks:

• I don't see any benefit in modular lane length. If I need more parts than I can fit in one lane, I would simply put the same part in two lanes. If I need less, I could always load a shorter strip to a longer lane.
• I think that an index pin for sprocket holes is required, not sure if this is in your new design already. It should be on the far end of the feeder. If I use shorter strips, they would all align at the same side. Example: If I put feeders on the West side of my PnP table and orient them parallel to Y axis, all strips should align to the right.
• 70mm lane length seems a bit too short to me. At 4mm part pitch, that would only allow me to fit 17 parts per lane - that's too few for most PCB and gets worse with wider tapes. I have looked at existing standards and found that JEDEC trays are 136mm wide. Why not use that? I don't know the tradeoff between size and manufacturability though.
• 25mm height requires quite a bit of Z-clearance; would still fit my machine (and LitePlacer) but might be an issue with some DIY builds, especially in openPNP land. On "open frame" table constructions it likely will not be a problem as part of the feeder could be "below surface".
  Still, why do they need to be so tall? Most parts I use are much smaller, so I'd question whether we need to make all feeders taller just to accomodate a few exotic parts.
• I suggest to include some sort of fiducial marker on both ends of the feeder and a place (tab) where we can put a small QR code. This would allow for much simplified auto setup in software.

Looking forward to your replies and other people's feedback!

Best regards
Malte

Thanks Malte,

Yes, I have seen your feeders. I liked the way you used pins to hold modules in place, but wasn't sure how to make that work with injection molded parts. Your idea of clamps gives me an idea that I'll draw in CAD later today. I think I can make clamps that can be injection molded, and have stops to help align the feeders.

By the way, here is the design that inspired me: http://www.liteplacer.com/phpBB/viewtop ... =312#p2232

I originally set the length of the feeder to 70 mm because I couldn't fit something 140 mm wide in my injection molding machine. It's a desktop machine, so I can't make parts that large. However, I'm currently planning on working with a company I've used in the past, and they can very easily handle this size. More importantly, they can use any plastic, whereas I'm limited to plastics I can get in pre-colored pellets. This is an important consideration if we want the plastic to be ESD friendly (not sure that's the correct term).

I'm using SolidWorks 2011 for the CAD design. In any event, here is a view that shows what's inside:
Comments:

• I'll change the length of this part. I can change it to 134 mm or some other number
• I want to support 12 mm height tapes because I'm using 60V 100 uF capacitors in my design that are close to this height
• I should be able to reduce the height and still support 12 mm tape. Making the parts longer should help with this.
• I'll work on incorporating at least one index pin. I think there will be cases where you might want to have two tapes in a single holder, so I'm going to try to include a pin at each end

I have a couple of questions for you:

• Can you provide me a link to the JEDEC spec? I searched for JEDEC trays and the length I found was 315 mm, which is way to long for our smaller machines
• What color should the parts be? I don't want to use black because it would be too hard to see the holes in black plastic tape
• How large would the tab need to be for a QR code? And does it really need a fiducial marker if it can use the holes in the tape?

Hey John,

funny you have posted a link to Manfred's design; I suddenly remembered where I had seen the toothed mechanism before

Can you provide me a link to the JEDEC spec? I searched for JEDEC trays and the length I found was 315 mm, which is way to long for our smaller machines

JEDEC specs are non-public, you have to be registered to download:
http://www.jedec.org/sites/default/file ... e4-10D.pdf

But everything you really need to know is that the outline dimensions of all JEDEC matrix trays are 12.7 x 5.35 inches (322.6 x 136mm). Low profile trays have a thickness of 0.25-inch (6.35mm).

I thought it would make sense that the feeders occupy the same space as one JEDEC tray so when laying out the table, one could put feeders or a standard tray into a given field. Also I thought that 136mm would allow about 34 parts per lane (at 4mm pitch) and that should be sufficient for smaller prototypes.

I want to support 12 mm height tapes because I'm using 60V 100 uF capacitors in my design that are close to this height

I understand. I also thought that electrolytic caps are about the tallest parts I have. However, more than 90% of my other parts are smaller.

My goal is to have all action (pick, measure with cam, place) happen on the same Z-level (height). So if the feeder is 25mm high from table surface, I need to put PCB at same level - so Z clearance from table top has to be 25 + tallest component height which could become a problem.

We could always change the table or lift the complete machine, but most people who already have built a LitePlacer or other machine might probably not appreciate that

So if the feeders were more about 15-20 mm high that would better suit my needs. For those few tall parts; I could still put them to the table with double sided tape or make an extra feeder which fits their height but is not as fancy.

I'll work on incorporating at least one index pin. I think there will be cases where you might want to have two tapes in a single holder, so I'm going to try to include a pin at each end

Pins at both ends make perfect sense.

What color should the parts be? I don't want to use black because it would be too hard to see the holes in black plastic tape

A big advantage of having feeders in predefined locations (be it a snap-in mechanism, clamps or pins) is that feeder setup would no longer have to rely on detecting the sprocket holes by camera.

That is why I would like to have fiducial markers on each end of the feeder. Those could be any symbol with a good contrast that we can detect by camera - circles, crosses, triangles... Could also be stickers or printed marks.

Software could then be enhanced to measure the fiducial positions of each feeder and calculate the location of the first and direction of tape part relative to that.

Given that the tapes have a tight fit and the index pins would also us to align the sprocket holes in a repeatable fashion, there is no longer a need to measure sprocket holes.

This would also allows for a more rigid top rail; as we could partially cover the sprocket holes without issues.

If you wanted to support sprocket hole detection, I'd suggest light blue or green. But that might also limit your choices (and the price) in material, especially if it should be ESD safe.

I think black will be by far the cheapest ESD-safe plastic. Usually they just add carbon / graphite to make it electrically conductive...

How large would the tab need to be for a QR code? And does it really need a fiducial marker if it can use the holes in the tape?

I think that depends on the resolution of the printed code; which would probably be a sticker. Camera resoltion will be quite good to see parts.

I'd like to encode feeder width and length (if we have different lengths at all) and a unique ID into the code - so not too much data.

A Micro QR code M3 or M5 should be sufficient:
http://www.qrcode.com/en/codes/microqr.html

For now, I'd plan about 8x8 mm. If we find a way to print codes with good enough resolution, they could be even smaller.

Cheers
Malte

mrandt wrote:But everything you really need to know is that the outline dimensions of all JEDEC matrix trays are 12.7 x 5.35 inches (322.6 x 136mm). Low profile trays have a thickness of 0.25-inch (6.35mm).

Currently I have it at 134 mm, which provides a little buffer.

mrandt wrote:So if the feeders were more about 15-20 mm high that would better suit my needs. For those few tall parts; I could still put them to the table with double sided tape or make an extra feeder which fits their height but is not as fancy.

I've reduced the height to 20mm and made some other changes, which gives me a clearance of about 10.5 mm. I think this will work for my capacitors, but I need to do some testing. However, having a special feeder just for these parts is also another possible option, as you suggested. In that case, since it would be a specialty holder, it might be OK for it to be a 3D printed part instead of injection molded.

Pins at both ends make perfect sense.

I still need to figure out the best way to incorporate pins. I saw one commercial strip holder that used spring wire, which might be an option (that holder was $75 for each row--yikes).

That is why I would like to have fiducial markers on each end of the feeder. Those could be any symbol with a good contrast that we can detect by camera - circles, crosses, triangles... Could also be stickers or printed marks.

If you wanted to support sprocket hole detection, I'd suggest light blue or green. But that might also limit your choices (and the price) in material, especially if it should be ESD safe.

I think black will be by far the cheapest ESD-safe plastic. Usually they just add carbon / graphite to make it electrically conductive...

I'm not sure I can get enough contrast with a fiducial that was part of an injection molded part. I think it's probably better to rely on the contrast between the tape and the holder. So I agree that a color like blue or green would be good. I'll check to see what colors are available that support ESD. I suspect the cost of the plastic will be small compared with the cost of injection molding, and the cost of the mold divided by the number of parts we make.

A Micro QR code M3 or M5 should be sufficient: http://www.qrcode.com/en/codes/microqr.html

OK. I'll think about how I can incorporate that into the design.

I spent a few hours this afternoon updating the design. Here is what I have now:

Please don't worry about hole recognition! I'm currently testing a software version where it is possible to define part locations directly. You show the first and last part on a tape at setup, and the software doesn't do any optical measurements. If the feeder has two locating pins, it will go at the same position, so we need to do this only once per feeder.

Maltes system for fixing the feeders to the table looks good. Especially, if there are locating pins on the feeders, the feeders stay in place when you loose the holding bar, allowing changing only one.

JohnSL wrote:I still need to figure out the best way to incorporate pins. I saw one commercial strip holder that used spring wire, which might be an option (that holder was $75 for each row--yikes).

I guess you are talking about this one:
http://www.nozzles4smt.com/STRIPFEEDERS ... c_135.html

They are ridiculously expensive... Might not be an issue if you operate a 1.5 M$ machine, but for DIY home / semi-pro users they are out of the question.

However, a spring loaded mechanism might work. You could make the rail supporting the side of the tape a bit wider and include a pin there.

Here is a quick sketch to illustrate:
The only problem I see with having the pin in the lower support rail is repeatability - depending on tape thickness the lower rail will be shifted a bit.

The other option would be to include the pin in the top rail - I think you had that in one of your earlier designs.

Assuming that we can cover the sprocket holes, as software either measures tape location from fiducials on feeder or is taught first and last component center, this should work also?

JuKu wrote:I'm currently testing a software version where it is possible to define part locations directly. You show the first and last part on a tape at setup, and the software doesn't do any optical measurements.

Better late than never:
https://github.com/jkuusama/LitePlacer-DEV/issues/29

*scnr*

BTW: This way to setup a feeder is also supported in OpenPNP.

John, don't worry too much about fiducials then. If they were needed, we could still place stickers on the feeders.

mrandt wrote:I guess you are talking about this one:
http://www.nozzles4smt.com/STRIPFEEDERS ... c_135.html

They are ridiculously expensive

Yes, that's the one.

mrandt wrote:The other option would be to include the pin in the top rail - I think you had that in one of your earlier designs.

The tricky part is designing it into the part so it's easy to make the injection molds (to keep the cost down). I want the molds to be straight pulls without any slides, so that puts limits on what I can easily do with the part. Right now I'm focusing on the main mechanism and then plan to deal with the alignment pin once I have the main assembly in good shape. I do want the alignment pins to come down from the top for the reasons you pointed out.

JuKu wrote:Maltes system for fixing the feeders to the table looks good. Especially, if there are locating pins on the feeders, the feeders stay in place when you loose the holding bar, allowing changing only one.

Yes, I agree. The most recent photo I posted has a version of his design. I was thinking that I would have an end stop that would set the position for one feeder, and then the other feeders would be pushed against that end feeder to control their position. However, I like the idea of fixed positions. I have an idea on how to implement this, and I need to check to see how this works for a mix of different tape widths.

don't see any benefit in modular lane length. If I need more parts than I can fit in one lane, I would simply put the same part in two lanes.

The software needs to understand this. For example, lane 1 has 15 parts left, lane 2 is full. Then, the job takes first 15 from lane 1 and continues from lane 2. Ok, I can add this.

Here is a system I've come up with to position the modules precisely when you add them back in:
This is an image from the bottom, and is using a modified version of Malte's rail suggestion.

You can see that one side of the feeder has longer tabs than the other. The long tab will fit into a slot in the rail, allowing precise positioning. The feeder slots in the rail are spaced 4 mm apart, and the feeders are currently 6 mm wider than the tape--an 8 mm tape will be 14 mm wide. This means the 8 mm feeders will have a 2 mm gap between them, so you don't get the absolute maximum density. On the other had, you can mount different combinations of tape widths, all aligned on 4 mm spaced positions. And you should be able to put feeders back into position with high repeatability.

I need to do a little more work on this design, but I think it has a lot of promise. The one thing that will be very important is to have a close fit between the screws and the screw slots for holding the rail down so it won't be able to shift sideways, and therefore out of alignment.

I currently have the rails 160 mm long, which means they can hold ten 8 mm wide feeders with a pair of rails. Of course, you can always add more rails and feeders.