Add-ons: feeders

[bobc]

It might be premature to start talking extensions, as no kits have shipped yet! But there are some things which are often talked about, and I think that feeders is one of them.

So I'd like to discuss some ideas on that. My goal is to create a modular feeder which can be easily attached, and possibly also used standalone for manual pick and place. I think that as a side project it is fairly self contained, and wouldn't require extensive changes to the existing LitePlacer design.

Basically there are two ideas, 1) Keep It Simple! and 2) automatic.

Approach #1. KISS!

This is based on the manual reel holders found on ebay etc:

basic_feeder by donotdespisethesnake, on Flickr

Here is how it could look mounted to machine

basic_feeders by donotdespisethesnake, on Flickr

(That shows a ShapeOko, but the principle is the same).

Reels can be mounted at side or front, and trays extend across the work area.
Possible to swap modules if different component profiles are required.
Could also be used for cut strips

Advancement: manual. The user exposes as many components as may be needed.

Cover tape: manual. The user must peel off tape and cut it.

Construction: aluminum or plastic to create channels, plus support for reels. Thin rod for axle.

Set up required: only need to align first usable component

Issues:
- the tape needs to be held firm against the work top (clamped or spring).
- with a lot of components exposed, it is easy for them to jump out if the tape is knocked.

Conclusion.

This about the cheapest way I can think of mounting a reel.
Need to experiment to see if usable.

[bobc]

Approach #2: automated feeder

This idea is based around a combination of the Versatronics feeders and V Besmans auto feeder for his PP4 machine.

Versatronics feeder, 10 x 8mm channels

versatronics_x10 by donotdespisethesnake, on Flickr

V Besmans auto feeeder



PPF Automatic Feeder

The Versatronics feeder makes a self-contained module, and modules can be arranged in different combinations, there are versions for 12mm, 16mm etc. It has a power and serial comms connections, so can be controlled by software command. I don't think it has an advance lever or button.

V Besmans PP4 feeder has a simpler construction, but is more integrated with his machine. Tape advance and cover tape removal are handled by one mechanism, using a cheap servo motor. A cheap microswitch is used to detect tape advance. The mechanism is a lot simpler than using rachets and springs, and requires a lot less high precision parts.

I am still in process of creating a concept diagram, but you can imagine the PP4 feeder as a module of say 10 channels, you get the idea.

Each module would have an Arduino type controller to handle serial comms, and drive up to 10 servos. It could be powered by 12-24V.

An additional feature I would like to add is an advance button on each lane, by means of small tactile push switch. This would allow a PnP machine to use the feeder without requiring to use serial comms. It also allows standalone operation for manual placement, the user just taps the switch to advance.

The Versatronics feeders are all second-hand, since Versatronics stopped trading. They are on ebay at £200 each, which is £20 per lane for 8mm tape. They could probably be used as they are without modification, if you can find the serial protocol used. To fit with LitePlacer, they would need to be recessed into the worktop.

I am hoping that a DIY built feeder module would come in <£10 per lane, or less than £100 for a 10 lane module. A module with more lanes would be more cost efficient, but perhaps less flexible.

[bobc]

Auto feeder concept

I quickly sketched up a concept diagram for auto feeder, derived from V Besmens PPF auto feeder. Not shown are microswitches for detecting tape position, take up spools for cover tape, electronics board, external connectors

auto_feeder by donotdespisethesnake, on Flickr

Hobby servos are used to advance the tape, and also collect cover tape. The microswitch detects the sprocket holes for alignment. The tactile switch allows manual operation or operation via the PnP head. As the force required is a lot lower than for ratchet mechanism, the vacuum tip could be used directly.

Because the tape is not driven by the sprocket holes, this design probably would not work well with cut strips, unless you have quite long strips.

The design is quite flexible, so width etc can be easily altered. The frame could be wood or plastic, perhaps laser cut. The feeder block needs to be more precise, probably milled Alu alloy, possibly one part can be laser cut plastic.

I will have a look for a cheap Arduino type board that can support 10 servos + 20 switch inputs + serial. Might need a custom shield.

I was thinking of using RS485 for comms, as that makes it easy to daisy chain several modules together.

[JuKu]

Automatic feeders would be nice, but obviously, much more challenging. I'll look into those after the kits are shipping in the earliest, I'm afraid.

the simple feeder is something that might work. I like things that are as simple as they can be, but not simpler. I have several those that shown on the photo, and for automatic use, they are too simple:

- Tapes are of variable thickness. If the slot that holds the tape edge is loose enough for paper tape, it is too loose for plastic tape. if the tape is not firmly held, the pickup height becomes arbitrary. Just using the lowest position (value that is good when the thin tape edge rests against the bottom surface of the side groove) slams the tape to the bottom when the tape happened to curve to the topmost position, knocking parts out of their pockets. This is a general problem. => The basic principle needs to be thought over.

- In particular, those holders that you can buy from internet around $25 per tray have too loose slots sideways as well. (They all seem to come from the same place.) Those are great for manual work, but don't work with this machine. => We need to source or design something else

- Some component tapes have rather thin pocket bottoms. The force that is needed to get a good grip on some parts is uncomfortably close to force that just pushes small resistor deeper in their pockets, getting them to stuck in the tape. I don't think we want to put in any fine-tuning requirement to the machine => Whatever feeder solution there is, a bottom part support saves a lot of headaches later.

- Smaller components are very, very light. Sometimes I get the feeling that all that is required to get them to jump out of their pocket is an angry look. => When peeling the cover tape, the tape must be held firmly. (Double side tape handles this nicely)

As said, I won't work (at least not much) on feeders before the kits ship. At the moment I've been thinking of a 3D printable holder:

Black is the body. Red parts move up and down, and there is some kind of spring mechanism holding down the tape (green) and pushing the moving side plates sideways (forces in blue arrow directions). I think this would work at least with manual feeding and peeling. The resistor tapes actually have flat cross-section, so when peeling, there is enough holding power to keep parts in their pockets, at least if you use a finger to put in some extra. This might work with automatic advancing (very little speed penalty if advancing several parts with one go) if there is a reliable way to peel off the tape.

I've been wanting an excuse to buy a 3D printer anyway. This idea is enough to have one before summer.

[bobc]

Thanks for your feedback ! I will try to incorporate them into my design. I am planning to develop this independently so that you can concentrate on Liteplacer.

To help with tape alignment, I will try some plastic spaces with a slight spring to them.

I've made several refinements to the autofeeder design, and I think it will be quite workable and cheap. I think I can get to < £10 per lane, some bulk purchasing will help. Arduinos and small motors are cheap and plentiful on ebay.

I will try a prototype using 3D printed parts, otherwise I will try milling some POM. I am also ordering components to start construction.

I will also publish the OpenSCAD files on github over the weekend.

[pjotr]

This project looks great! I have checked out your software and you've put a lot of hard work into it, it seems very user friendly. Great job!

I make few boards that have lots of different parts, not lots of boards with a few parts, so reel feeders are not as useful as the method you already use-- taping down little bits of cut tape. What would be nice is to be able to save the relative tape locations for a project as a file so if I populate a pcb I can prepare the all tapes on a separate board specific to the project that has the rulers permanently affixed, place the PCB and the board on the machine, peel off the tape covers, locate just one fiducial and go. This would let me make a few boards at a time, and not need to keep reels of parts in stock. Am I right that it finds the part from tape every time by looking for the tape hole? How far off can you be from the location it thinks the hole is initially to still find the right part? <1/2 hole spacing?

-pjtr

[JuKu]

What would be nice is to be able to save the relative tape locations for a project as a file so if I populate a pcb I can prepare the all tapes on a separate board specific to the project that has the rulers permanently affixed, place the PCB and the board on the machine, peel off the tape covers, locate just one fiducial and go.

You can save the project file already. I call it a job file, but it is the same thing. You still need to have more than one fiducial on the board (at least three, four preferred).

This would let me make a few boards at a time, and not need to keep reels of parts in stock. Am I right that it finds the part from tape every time by looking for the tape hole? How far off can you be from the location it thinks the hole is initially to still find the right part? <1/2 hole spacing?

Right. At the moment*, it looks the tape hole for each part. And you guessed correctly, it needs to be a bit less than 2mm (1/2 hole pitch) from where the machine thinks it is, otherwise it might pick a wrong hole.

*: This was not the most intelligent choice. It is the most accurate, but the machine would work significantly faster and sufficiently accurate (for most parts at least, likely for 0402s, too), if it would look for just the first and last hole for a part strip and extrapolate the rest. You can place the tapes into sufficiently straight line easily enough. I will keep the current method, but I'm going to add this faster method, too.

[JuKu]

Some food for thought: one big problem in handling tapes (feeders in particular, but in generallly too) is how easily parts jump out of their pockets after the cover is off. Here is a video how a super high end machine solves the problem. One head out of twelve, each picking up about ten components per second - you do the math...
https://www.youtube.com/watch?v=275BCu9lQJQ

It has some kind of a lever holding the exposed part in place while the tape is moving. This is a feature worth of copying, I would say.

[Spikee]

To be honest it should not be super hard to make a simple electronical feeder to work.

One component exposed -> advance with a wheel or so the tape till the next aligment hole (small holes at the side -> pick another component.


But than with a light bridge to accuratly detect the holes. Worst case one can use one usb camera per feeder to align the parts.
As these usb webcams / cameras with another lens are so cheap now this should not be super hard to achieve.

[bobc]

The design you posted is from here http://tim.cexx.org/?p=803 It's experimental, not sure if it ever saw practical use. It looks rather problematic to me, I can see the tape jamming a lot. It also doesn't deal with cover tape, which is an issue.

I think that would work ok, if you only want one or two feeders. What if you want 10, or 20? In that case, cost and size becomes an issue.

Btw, optical sensing of holes doesn't really work for transparent tapes.

If you read above, the design I am working should be effective, compact and cheap, and deals with the cover tape. Hopefully it will also not rely on too much precision cutting. I am building a prototype with 3D printer parts.

I will be posting the design to github for anyone to use at https://github.com/bobc/SmtAutoFeeder