Board feeder/conveyor

[mr_darker]

Errybody is working on component feeders, but I don't see anyone working on board feeders. I just started looking at making panels since liteplacer is in the mail, and they have like... 10mm strips top and bottom, and 4mm holes to grab the board with that was recommended (off an EEVblog video). There oughtta be a stationary bottom rail (w/ conveyor belt thing) that can hold the bottom end of the panel, and a top rail that slides, is adjustable as a one time setup kind of thing, so that the machine can hold a board between the two rails and pull in/out one after another. And maybe supports or something for flimsy boards once placing parts.

Then, could put more effort in, and instead of using a toaster oven with something like xtoaster, just make a box that the board goes into via conveyor, probably needs actuated/sliding doors (or just the whole top/lid moves up and down) to keep the temperature even inside.

I bought the CYBRES SP2421 because based on my few experiences using stencils, it's a bitch (i choose the smallest parts I can get and put them as close as is allowed), and I need to do so 100 times for my first project? I think it'll be worth it to have a frame you just set up once. But this thing could easily just be attatched to the conveyor line right before the pick and place part. Make a thing that loads syringes in, or however else it might be able to self-apply paste, then have it apply constant pressure as it squeegees it all across the stencil. Something like that, then have the PNP inspect the solderpaste on the board prior to putting parts together.

General idea is just automate it all, set it up, load in reels of parts, load in your pcb panels, and just push a button. Main thing is the panel rail/holder, everything would work around that. Apparently the holes to grab the board with are usually adjustable around the board, but so it's simpler, it's probably better to adjust the board around the machine since a panel is already adjusting for a machine. In order to apply a stencil, the top of the rails & pegs to grab the board need to be flush with, or lower than the top of the pcb, so maybe make the pegs as tall as the thinnest common pcb. Rails on the sides would just hug those 10mm strips on top and bottom of the board, keeping the board on those pegs. For moving the conveyor belt and stuff, this PNP is so slow that you can probably find some way to power the movement from the printhead (whatever the pnp equivalent of a 3d printer printhead is), lose almost no time (comparatively) and less electronics. This thing really oughtta have a solenoid/peg for pulling/advancing reels, and maybe for moving something like a board conveyor.


Just some ideas. I have zero 3d modeling skills, but will try to get someting like this going, all just manual at first. It'd be nice to have it all automated. Figure it all out and then make it the standard, or maybe an official Upgraded model option. Sell kits for existing users. My goal is small scale mass production. Like 100. I don't want to stand around this thing all day constantly setting it up again and feeding it at a ton of intervals. That's just for my use, some other guy wants to start selling them for his own use, and that'd probably bring in a few thousand boards needing to be printed. Even if they're panelized, that's still a lot.

Also make it work on linux, I'm just crossing my fingers that this program works on wine.

[mrandt]

Hello and welcome to the forum,

I understand your desire to automate everything.Basically what you describe is exactly how commercial PCBA lines work; however these cost a fortune and have to be dialed-in well to work fully automated... Your plans sound quite ambitious.

I seriously doubt that we will be able to automate end to end process including stencil printing + reflow anytime soon; and I'd also question if that is really necessary:

From my experience I'd say I spend about 90% time setting up parts, configuring locations and testing a job before I have my PnP populate a board. You could cut down time if you have fully automatic feeders. I'd argue that there was no point in discussing a conveyor system if you don't have auto feeders.

Paste printing, loading board into machine and putting populated boards into the reflow oven are neglibile in terms of effort when building prototypes.

Different story with large scale manufacturing; no question. But once I get to those quantities, it is usually easier and more economical to outsource to a professional PCBA shop - or invest into a "professional" assembly line. Still, if you want conveyor, you will face a significant investment.

Neoden4 seems about the cheapest machine to feature a conveyor; however I have not seen it being integrated into a fully automated assembly line. Their conveyor seems to be mostly used to populate oversized boards (e.g. LED strips) and make loading and positioning of the PCB easier.

Automatic paste printers cost a fortune and a suitable reflow ovens with multiple heat zones and conveyor are rather large devices... Many smaller shops seem to use semi-automated or fully manual paste printers and oven move panels to reflow oven by hand.

Don't get me wrong; I do not intend to discourage you - but for automation my #1 priority would be auto feeders and easier parts setup.

Cheers
Malte

[JuKu]

Hi Malte,

Do you have any ideas how to make parts setup easier (maybe in another thread)?

[mr_darker]

I understand your desire to automate everything.Basically what you describe is exactly how commercial PCBA lines work; however these cost a fortune and have to be dialed-in well to work fully automated... Your plans sound quite ambitious.

Only costs a fortune if you don't do it yourself (the only way things are done right, eventually). There is not a thing in the world that isn't overpriced, even this liteplacer, at overpriced DIY prices, it's $100 of rails, let's say $50 per motor (total at $300), screw/washer/fastener costs are negligable uness buying them individually wrapped, the little solenoid and vacuum pump I would guess cost 50 bucks tops (total at $350), bearings can get expensive if high quality so I'll just price them as luxury skateboard bearings @ $50 for 8 which brings that price up to around 300 as a rough guess (total at $650), tinyG (total at $800) if you have a lasercutter or access to one... all those custom parts are negigable in materials cost, so a DIY kit price of about 1k (still at overpriced individual parts cost) being sold at 2k. I'm not complaining about the price btw. The remaining 1k of price would be royalties for his design and efforts, something you don't have to do (that time saved is worth 1k to me), also the custom software he put a ton of time and effort into, plus digging into community efforts and distributing improvments as upgrade kits, that and plain old incentive/profit for him to do all this make up the extra 1k, either that or he's got some cost cutting to do on his end. Nothing is too expensive unless it's materials cost (raw material, like aluminium and iron) that is too expensive, question is if it's too much work for you to make it affordable. As for quality and precision, we got OpenCV or whatever visual software this is running on. I really think the cameras should all be moved so that the part is completely in sight while being placed (in sight of two cameras so there is a view on all sides). Not sure how the software works yet but (i think) it should only use coordinates to get into the general area of a board and then use the camera to make real time adjustments as it does things, like I should be able to yank it out of place as it's placing a part (ignoring the fact that I can't do it that fast), and it should stop because it sees things not matching up, and move back into place before continuing. Or set a part down and see it is out of alignment, and pick it back up to try again, or abort the process on that board if it can see a short circuit. Add that kind of software superiority and how rickety your machine is really only affects how long it takes the machine to get it right.

With feeders which everyone else is working on (so I'm not focusing on it) setting up parts should take less time, also with a board feeder, it'd know where your board is all the time and only need you to set up the conveyor. If you have both, you just load the reels and adjust the rail, load the board magazine and the board file, maybe part files too. Global coordinates just taken from the conveyor/board position, local coordinates taken from visuals on the part and the board. Rather than have a huge mechanical thing for feeders, just stick the reels on a bar (to hold & let rotate them), make a little slot that's attatched to the table so the tape is held down and affixed to a general spot. The tape on top gets peeled back, taped or someting to a little extender, and affixed back onto the reel except on the underside of the tape , thus peeling it off as the reel unwindes as well as bringing waste tape&film to the same place, probably just a hole that leads to a garbage can under the table. Like I said, solenoid that plunges down into the tape holes, and advances the reel, seen it on one pnp machine. Just tie a weight on a string to each reel or something to get a light force trying to wind it backwards, and on the little thing attatched to the table, just have a little gear or something with the teeth sized for whatever tape you're using, and have it rotate on a one way bearing. That way the pnp just moves to the tape, locates it, goes down and sticks the solenoid in the reel holes, advances it until a part is in range, then grabs it. Get the visual software checking everything as it goes, and it's no problem. There's an idea, affix it to the front of the machine since there's no rail there, and a ton of open space for tons of reels. The back rail is needed for the conveyor. Edit: The film might not continue on its own once the initial taped point is freed, so maybe have it all go under the table and roll back up on a reel for collecting waste, this garbage reel would provide tension in the other end too using a string and weight or something fancy.
Conveyor would be based off the back, probably attatch to it so the rail can extend bast the outside edges pulling boards in and spitting them out. Just have little rollers sit on top of the board to hold it down (is this necessary holding a board down?) and obviously a surface for it to slide on at the edge of the board. Have the same exact thing on the other side but affixed in place by screwing onto the the outside of the makerslides and pressing down against the table, maybe give it two points of contact on each rail so it isn't just a flimsy extension, this would be something you just put your board between on one end to get the spacing before screwing it down, and repeat on the other end. Bam, there's a conveyor, just have ye' ol solenoid drop down into a hole in your pcb panels near the front and pull the board along, could even have it do multiple panels at once this way since the work area is so long. Have a giant pez dispenser as the board magazine xD Just have a little ram attatched to the pnp head that will do the "head tilting" thing and push a board out enough that it can grab the hole and pull it out.
As for paste dispensing, the rail is alreay spaced out to fit the board. This could prettymuch just be the cybres sp2421 sideways sitting on the conveyor rails, probably needs a little more stuff to make it stable and adjustable without needing new parts for each stencil. Maybe there is a lever that the solenoid (the general purpose mover thing) can actuate in either direction to flip the stencil up and down. ALl the lever parts would need to be made, or something else. So the thing rams the pez dispenser board feeder, grabs the single board that pops out, pulls it into place, pulls the lever (or whatever way it is actuated) slowly to flip the stencil down. checks alighnment which should only be in one direction if you set the rails and stuff up correctly. Then.... maybe the PNP head has a solder paste extruder on the other side of the makerslide that the pnp head is affixed to, independant operation from the pnp head, basically apply paste like a 3d printer does, I've seen some paste applicators that use this method without a stencil using super liquidy paste, but I wouln't trust it. Just use the needles to extrude into the stencil, hopefully mess free, maybe just work at high speed like an inkjet printer and just use the stencil to fine tune all the details, maybe removable nossels for different sized stencils. Something like that. Stencil application is the hardest part.
As the boards go out, they can just go on one of those conveyor toasters, modified to be fed by the pnp (perhaps a hole in the front AND back of each pcb panel), the thing can easily have the speed adjusted to whatever time is required by the reflow profile (liteplacer can easily make sure only one panel goes through at a time), replace the heater wires with quartz heaters, pop in a reflow controller, have something all the finished panels are collected into, basket or something with a slide so they don't just fall.

Paste printing, loading board into machine and putting populated boards into the reflow oven are neglibile in terms of effort when building prototypes.

But if you have only 100 boards or so of each model you make, I'd say it's pretty good. Also can kinda cut out a fabrication factory if you don't have enough sales to justify one. There's really no reason not to if you can.

Different story with large scale manufacturing; no question. But once I get to those quantities, it is usually easier and more economical to outsource to a professional PCBA shop - or invest into a "professional" assembly line. Still, if you want conveyor, you will face a significant investment.

Investing in an assembly line isn't an option for me, barely got approved for a loan to buy this machine. My first project is intended to raise funds so I can afford to make better things. Can't skip ahead to having equipment or money, the plan is to build it all myself near reasonable prices, part of why I got liteplacer.

[mrandt]

As I said, you sound ambitious. It is fair enough to try and build as much as possible and I wish you good luck & much success.

With regards to LitePlacer pricing, I think the kit prices are more than fair. I built my machine based on Juha's plans before he started selling the kit. I do not have access to a laser cutter so I had to have the plates made by a professional shop. Also, sourcing all the parts and bearings (many from Misumi / Japan) was not cheap either. I believe I paid almost the same amount just for the parts, since I bought small quantity; still I had to invest many hours to find all parts, place orders etc.
Considering all the time Juha has to spend on the project, I doubt that there is much profit for him - at least not monetary...

For the conveyor have a look at Neoden 4. I think it might be possible to come up with some similar design using pulleys and rubber-wheels.

Still, considering the whole process, I would focus on other areas of improvement first. Stencil printing and reflow would not be my primary concerns.

I can easily print solder paste onto hunders of boards by hand; especially when using panels. Also, putting panels into the reflow oven one by one is not a bummer.

If you were to build a conveyor, I would make it in such a way that you can place multiple panels with solder paste applied and then populate them one by one.

Preparing components is the biggest at the moment. I agree many people are working on feeders, but that has been going on for a couple years now and a modestly priced, solid and reliable Open Source feeder has yet to be invented. Most people seem to fall back to commercial feeders at some point... So unless this is solved for you, the rest of the automation does not really gain you anything.