SimplePnP: Your personal open hardware PCB assembly machine

[lutkeveld]

Pretty excited about the cost of pick and place machines coming down.
In combination with a small reflow oven, this is great for proto and small batch manufacturing.

https://www.crowdsupply.com/citrus-cnc/simplepnp

[mikeselectricstuff]

Yet another waste-of-time cheap P&P machine.
It will take longer to set up than hand place for most use cases.

[ottoragam]

Hi Mike, creator here.

Yet another waste-of-time cheap P&P machine.
It will take longer to set up than hand place for most use cases.

That is true in various cases, but in my experience, there are some assembly jobs that fall between outsourcing and hand placing. The machine is not for everyone of course, but I think there are people that could benefit from it, and it is better for something like this to exist. You can even assemble one yourself if you don't want to buy it, I made the design open source.

Besides, setting one up is not hard, think about strip feeders and not extra fine parts.

[fcb]

Good luck ottoragram.

It's ALL about the FEEDERS.  Solve the low-cost feeder conundrum and you're there.

I started a low-cost P&P project some years back, got a basic unit picking and placing with vision, even had some minor success picking from a small pile of parts (we did it with a pager motor vibrator on a small plate).  But whichever way I looked at the financial feasibility of bringing one to market I couldn't design a low enough cost feeder that worked.

I have had several Versatronics RV/RV1S machines and they have fairly good feeders, now I have a TWS Quadra machine with a metric ton of feeders.  I couldn't imagine a simpler feeder than the TWS.  You'd also be surprised by just how much force you really need to strip the cover tape off.

Start with the feeders!  They are the secret to success with P&P.  If I ever restarted the project I would probably spend 80% of the development time/budget on the feeders.

By the way.. did I mention it is all about the FEEDERS.

[ottoragam]

It's ALL about the FEEDERS.  Solve the low-cost feeder conundrum and you're there.

Completely agreed. I'm already trying. It's HARD!




There are quite a lot constrains for the feeders: compactness, ease of manufacture (so anyone can fabricate it without special processes if the design is open), reliability and price. I was aiming for $30 USD for one unit.

[fcb]

I'll let you into a secret, especially as I never got round to patenting it.

The best feeder I ever designed used Nitinol wire and springs to do the 'pull' on the tape, and a second piece of Nitinol to pull the transfer pin.  No electronics in the actual feeder, sort of mix of Agilis and Versatronics/TWS. I didn't get round to doing the covertape puller, but probably would have made it either a gang piece or a ratchet driven by a Nitinol pawl.  Design was 10mm pitch with 8mm tape.

It couldn't do much more than 1 action per second, but fast enough for what I was wanting to do.

[ottoragam]

Hmmm, interesting idea. It never crossed my mind to make a Nitinol actuator. The tape feeding part of my setup works well enough already, and, as you found already, the cover tape peeler is the complicated part.

I don't think my feeders will be that thin tho. Making them from printed plastic makes them not very stiff. Maybe I'll have to migrate to metal.

[fcb]

Printed plastic is not going hack-it, the wear on the parts is huge.  Mine was designed to be made from 0.7mm stainless steel, that was laser cut and folded, the carriage was milled from HDPE and slid in grooves cut in the steel, although I suspect I would probably have modified the design to reduce wear.

I used springs to return the Nitinol back to it's home posistion, and prevent damaging the wire when it hit the retraction 'end-stop'.  The amount of force that hair-thin Nitinol can generate is shocking.

[ottoragam]

My choice to use 3d printed parts was due to process accessibility. Even if the material wears out, it is easy just to print another piece. I did make some prototypes out of steel, but it is a PITA to bend accurately if one doesn't have access to a press brake. I'll probably end up doing some front and back plates from sheet metal and sandwiching plastic inserts in between too.

The amount of force that hair-thin Nitinol can generate is shocking.

I honestly have to do some research on the Nitinol. I'm completely inexperienced with the alloy.

[mikeselectricstuff]

It's ALL about the FEEDERS.  Solve the low-cost feeder conundrum and you're there.

Start with the feeders!  They are the secret to success with P&P.  If I ever restarted the project I would probably spend 80% of the development time/budget on the feeders.

By the way.. did I mention it is all about the FEEDERS.

This 1000%. Anyone can knock together an X-Y gantry.
I think the most scaleable approach could be to use a camera to puck loose parts from trays using a head-mounted camera (e.g. the fid cam)  Cameras are cheap, vision software is free. The number of trays is limited only by the X/Y area, and you could use a single global shaker.
You probably do still need some proper tape feeders for parts in large qtys like 100n caps, but if you can do all the lower-qty parts loose, this could potentially save a lot of cost.

[Mechatrommer]

so whats your recommended/best feeder? i want to reverse engineer (seriously) kudos to SimplePNP creator (ottoragam?), one break through is using that Juki nozzle, thats one hell of an idea imho, we can search for the best nozzle in the world that can take hours, days, weeks, months or years, but one suggestion like this can wrap all the possibly wasted time into a single minute. 3d printed part is ok as long as you brace weak points with stronger (metallic) parts ie hybrid materials/tech, just dont use 3d printed material that disintegrates itself within few years.

[mikeselectricstuff]

My choice to use 3d printed parts was due to process accessibility. Even if the material wears out, it is easy just to print another piece. I did make some prototypes out of steel, but it is a PITA to bend accurately if one doesn't have access to a press brake. I'll probably end up doing some front and back plates from sheet metal and sandwiching plastic inserts in between too.

The amount of force that hair-thin Nitinol can generate is shocking.

I honestly have to do some research on the Nitinol. I'm completely inexperienced with the alloy.

I can't see 3D printing cutting it.
For cheap, accessible manufactured parts I'd be looking at using PCB material and maybe stainless stencil material, both of which can be cheaply and accurately cut & drilled in small volumes. And where necessary, off-the-shelf metal parts like pins, rod etc.

[fcb]

You probably could little boxes (probably green or blue, to give contrast or the equivalent of 'green-screen' chroma-keying) - these would all be sat on a single shelf.  The whole shelf would vibrate on command.

If the boxes were say 40x40mm and quite shallow, with maybe slightly sloped sides and a barcode/description plate, then the hobbyist/prototype set-up time would be really fast as these would double as the prototype stores.

In operation, you would probably:
1. Posistion the downloooking camera over a tray
2. Take a snapshot
3. Identify which parts are the right way up and the orientation (diodes, SOT23, polarised caps - perhaps even read the text).
4. Pick a part if correctly identified X/Y/Theta or buzz the shelf/tray if nothing suitable identified, goto step 2.

[ottoragam]

I think the most scaleable approach could be to use a camera to puck loose parts from trays using a head-mounted camera (e.g. the fid cam)  Cameras are cheap, vision software is free. The number of trays is limited only by the X/Y area, and you could use a single global shaker.
You probably do still need some proper tape feeders for parts in large qtys like 100n caps, but if you can do all the lower-qty parts loose, this could potentially save a lot of cost.

OpenPnP supports tray feeders and vibration feeders already, and with the bottom camera one can correct the linear and angular offset of the component before placing it. I still need to test these kind of feeder setups, just need a bit more time.

[ottoragam]

one break through is using that Juki nozzle, thats one hell of an idea imho

Yeah, I feel commercial nozzles are the way to go. And any other nozzle brand can be used, as long as there is an available nozzle holder for the nema8 stepper.

[ottoragam]

Just for reference, a working 3D printed automatic feeder design:

https://www.docs.mgrl.de/maschine:pickandplace:feeder:0816feeder

[ademuri]

I know this is a little old, but...

As someone who hasn't used PnP machines before, could you elaborate on when using something like this would save time? I'm a hobbyist who builds a lot of one-off boards, and a fair number of small volume (say, 5-10 board) designs. Right now I place parts manually using a tweezers. It's fine, but if I could e.g. take half as long to assemble those (without spending all of my time tuning/maintaining the machine), it would be worth buying a $800 machine.

I read through the how-to for the LitePlacer, so I have a little idea of what I would need to do for a machine like this. What sort of things end up taking the most time? How annoying is the work, versus manual placing? How much fiddling/tuning is required?

[mikeselectricstuff]

I know this is a little old, but...

As someone who hasn't used PnP machines before, could you elaborate on when using something like this would save time? I'm a hobbyist who builds a lot of one-off boards, and a fair number of small volume (say, 5-10 board) designs. Right now I place parts manually using a tweezers. It's fine, but if I could e.g. take half as long to assemble those (without spending all of my time tuning/maintaining the machine), it would be worth buying a $800 machine.

The first thing to do is to get a foot-pedal-operated vacuum pickup - this saves a lot of time for hand placing, especially where you have a number of the same parts, and more so when they polarised.
An $800 machine is unlikely to save you much time unless you have a fairly large number of parts (at least a few hundred) but only a small number of different ones.