LitePlacer.com forums

Small video for a simple machine ( I did own that type )

https://www.youtube.com/watch?v=3kkf6bktIHA

I think the principle is similar to what you suggest

jarekk wrote:I think the principle is similar to what you suggest

Hey Jarekk, I think the principle shown in the video is more the optical measurement and offset / rotation correction of parts - like discussed e.g. here:
https://github.com/jkuusama/LitePlacer-DEV/issues/34

I am suggesting something different:

With John's solution you could mount a number of cut strip feeders to your table. You could combine lanes to support 8mm, 12mm or others in any order.

The big advantage is, that each and every feeder will be in a predfined position. The machine's software roughly knows where to look for feeders.

If we then include a small QR code, the camera could capture it, decode it and would instantly know how wide a given feeder is - and could thus figure out where the first part should be picked up.

If we also had fiducial markers on the feeder, the machine could instantly determine the exact feeder position and aligment - and from that deduce the pickup locations. Finicky optical sprocket hole detection would no longer be necessary.

The QR code could also include a unique feeder id; which would allow for the software to "recognize" a feeder which it has seen before and ask you if it should reduced stored configuration (pocket pitch, part type, part designator, etc.). So if I had a bunch of feeder modules, I could recycle them for commonly used components without the need to configure them over and over again.

This combination would greatly reduce setup time for a given job:

1. put component strips in feeder modules

2. mount feeder modules onto table

3. start "feeder configuration wizard"

a. camera scans all potential feeder locations, detects and reads QR code for each

b. per location, it already knows the feeder module / tape width

c. camera scans for feeder fiducials (optional, if alignment is fixed and known - depends on mounting location) and calibrates feeder position and alignment

d. wizard detects if a feeder has been "seen before" and allows user to reuse existing config for that lane

e. wizard asks user to configure part pitch, part type, part orientation, designator etc. for the other lanes

Compare this "Auto Setup" to glueing parts to table surface, navigating to sprocket holes of each tape strip, fighting opctial detection, fiddling with orientation and part pitch, hoping that everything works and - worst of all - repeating this tedious procedure each and every time you want to build a board...


@John: The grid is a nice idea. But I am not sure that I like the spacing between 8mm feeders... I will try to think of something.

The more I think about it, the less I think we even need to have that grid in the rails.

Assuming we use QR codes, maybe fiducials and auto setup in software.

If we simply stack the feeders next to each other without gaps, machine could simply figure out the location of the next feeder based on the width of the previous (decoded from QR code)...

IMHO that would be the most flexible solution.

mrandt wrote:The more I think about it, the less I think we even need to have that grid in the rails.

Assuming we use QR codes, maybe fiducials and auto setup in software.

Can you provide more information about why you don't like the alignment slots?

I'm liking the slots, as then I don't have to rely on QR codes or fiducials. In my case, I'll be using OpenPnP on my own machine. Currently my machine does not have a down-looking camera that I can use for this type of alignment. Instead, I'm using two cameras spaced 90 degrees apart that look at the pickup head. So in my case having the alignment pins is actually a good thing, as it means I can put the feeders back in the exact same position.

If your concern is about not getting the maximum number of feeders, there are some other options. Let's assume that 8 mm feeders are by far the most common, with 12 mm in second place, and all other widths a distant third. If I reduce the spacing between slots to 3.7 mm (instead of 4 mm), that reduces changes the gap between 8 mm feeders from 2 mm to 0.8 mm, and it will make the gap between 12 mm feeders just 0.5 mm. So that would mostly remove the issue of gaps for the common feeders. For the larger feeders, the gap will often be larger. But I think that's a reasonable compromise.

Here is more information about my machine:

https://groups.google.com/forum/#!topic ... Edizt44R5A
https://groups.google.com/forum/#!topic ... mf2CoZMfyI

JohnSL wrote:Can you provide more information about why you don't like the alignment slots?
[...]
If your concern is about not getting the maximum number of feeders...

You are spot on - I want the option to fit as many feeder lanes as possible.

But that is probably not an issue at all, as we could have two versions: one featuring rails with a grid and the other with straight rails

I think that QR code and auto setup are still valuable features to reduce setup time.

It's been a while since my last update, so I wanted to give you an idea of what I've been up to. I've been iterating through a number of different versions. One of my goals is to have a single set of injection molds that can be used for all the different tape widths. My idea is to use spacers of different lengths for the different tape widths. There are a total of 5 spacers. For the 8mm tape, these spacers are $0.38 each, so five of them totals $1.90. When I added up all of the hardware costs (spacers, springs, screws), the total hardware cost came to $2.55. The injection molded parts would add to this cost as well (I don't know how much yet). And, of course, we have to cover the cost of making the mold itself, and also some profit for myself and Juha. The typical advice I've seen on hardware is that the price should be about 5 times COGS in order to be profitable. So if I were to apply this to these holders, this means the price would have to be $12.75 just for the hardware. So the entire holder would be even more.

I know Juha wants to keep the price low on these. Thinking about this for a while, I realized I can eliminate the cost of the spacers for the 8mm version by including spacers to the injection molded parts, like this:
Then for thicker tapes, we add spacers. Here is what it would look like for 12 mm wide tapes:
Using this approach, we'll be able to sell the 8 mm tapes at a lower price than all the other tape widths. Since this is probably the most common width, it means we can keep the overall cost lower.

You may have noticed that I'm still experimenting with the best way to add the spring. I have some other ideas that I need to try out. I've ordered some springs and spacers so I can start to experiment with 3D printed prototypes. Right now for my prototypes I'm using rubber bands. I really like how rubber bands are working, and I'm slightly tempted to use them instead of a spring. The advantage of rubber bands is that they're really cheap. The disadvantage is that they break down over time, and therefore would have to be replace. So my current plan is to go with an actual spring. The question, then, is how to anchor the spring to the two assemblies. The diagrams above show an injection molded part (which isn't quite right yet). But I have some other ideas as well that I'll test out once I get the springs.

By the way, I'm using my FFF printer along with supports to make the prototypes. They're not as rigid as injection molded parts, and the surface finish is inferior. However, they're close enough so I can try out different designs before making a mold.

I also have to work out including a pin to index the tape each time.

Hello John,

thanks for the update. The new version (with the spacers for 8mm included) looks good.

I'd definetely go for a spring; we want these feeders to be reliable and avoid fiddling with rubber bands every now and then...

The cost for fasteners (screws, nuts, spring etc.) seems rather steep to me. Where do you intend to source these from? Maybe it is worhwhile to investigate a bulk-buy from manufacturer or other distributors?

For the spacers: Would injection molded spacers work? It might be possible to include the spacers in one of the molds; so they would come as a side product for very little cost (basically only a bit extra plastic raw material).

That being said; I totally understand that you need to cover the cost and make a profit - but IMHO something in the range of EUR 20 per feeder lane would be too expensive. Let's assume, I need about 40 feeders. That would already cost about EUR 800 !?! Much too expensive for my taste.

Compare these to second-hand automatic feeders, which sell at prices starting at EUR 50... If I were to spend this kind of money, I'd go automatic.

Judging by the use case I'd say that I would pay about 5 - 10 EUR per feeder lane to get these ready made. If price would be significantly higher, I'd rather continue using my own feeders and if I did not have those I'd mill or print something myself.

I thought the idea of injection molding was to lower the price by mass manufacturing and not create an unaffordable product?!

But I'll wait and see what happens.

I don't think $20 for feeders is a good idea. My target is $10 retail per feeder. When I calculated the costs of the hardware I knew I had a problem and needed to change the design. My post was to let people know where I'm at, show the problem I'm currently working on, and get feedback or ideas.

I used pricing from McMaster, which I'm sure isn't the best price. I haven't spent time yet to find better sources, so I don't know how much I could get these prices down. However, this exercise motivated me to rethink the design.

After this post, I had another idea that I've been working on that will either reduce the number of spacers to 2 (down from 5) or eliminate them entirely. Here is where I'm at with the redesign:
You can see it's very much a work in progress. But the key thing is that the slider section is held to the outside section with screws, thus eliminating two of the spacers. Likewise, I'm planning on changing how the springs are attached to eliminate another spacer. You can see a screw poking out without a nut--this will hold one end of the spring.

I'm sure this design also won't be the final design. One thing I've learned from previous projects is that the DFM process is all about iteration. I'm not smart enough to come up with the best design on my first try, so I go through multiple iterations.

Speaking of sourcing, does anyone have suggestions for sources for springs and spacers? Here is the space I'm using for the 8 mm wide tapes: https://www.mcmaster.com/#95947A002

For springs, I'm using extension springs. I've done a little bit of looking for other sources, and so far they've all been more expensive than McMaster-Carr, which doesn't make any sense to me.

JohnSL wrote: Speaking of sourcing, does anyone have suggestions for sources for springs and spacers? Here is the space I'm using for the 8 mm wide tapes: https://www.mcmaster.com/#95947A002

For springs, I'm using extension springs. I've done a little bit of looking for other sources, and so far they've all been more expensive than McMaster-Carr, which doesn't make any sense to me.

For both spacers and springs I'd recommend finding a Chinese vendor on eBay or Aliexpress that has something close to what you want and then ask if they do custom. Most likely they do. A lot of these vendors are just representatives for machine shops, or more commonly, one guy with a spring machine. You can get these type of mass produced product for fractions of a penny that way.

Here is an example I found searching Aliexpress for "custom spring": https://www.aliexpress.com/item/steel-e ... bc86406e18

The price you see there is almost always basically a random number. Ask them how much for 100 and the price will be much, much lower.

Jason

I don't have anything productive to add here. I just want to say that I really like the direction this is going!