O-M-G cheap pick and place machines coming!

[maliqua]

http://hardware.slashdot.org/story/14/07/29/1921200/a-look-at-the-firepick-delta-circuit-board-assembler-video

http://hackaday.io/project/963-%24300-Pick-and-Place-%2F-3D-printer

http://delta.firepick.org/

I for one would love to see dave race it 


*edited for spelling and forgetting the link to the main site*

[EEVblog]

We are developing a really cool robotic machine that is capable of assembling electronic circuit boards (it also 3D prints, and does some other stuff!).

Destined for failure right there. Jack of all trades, master of none.

[miguelvp]

We are developing a really cool robotic machine that is capable of assembling electronic circuit boards (it also 3D prints, and does some other stuff!).

Destined for failure right there. Jack of all trades, master of none.

It doesn't stop there:

http://delta.firepick.org/philosophy/

AUTO TOOL CHANGER

The delta robot configuration gives us plenty of room to install an auto-tool changer.  This makes it a multi-purpose machine.  Here are some of the tools that we are envisioning:

• 3D Printing Hotend
  
• SMT Pick-and-Place vacuum nozzle
  
• Solder paste dispensing syringe: This allows the machine to place solder paste consistently on the SMT pads, which eliminates the need to buy a stencil if you're only making a few boards.
  
• Hot air rework: Imagine being able to rework a BGA or SMT part with full camera vision.  BGA rework machines sometimes cost upwards of $10000.  This hot air tool will also be able to do minor reflow work, if your board only has a few parts.
  
• End mill / drilling attachment for PCB milling and other light-duty jobs.
  
• In-Circuit programmer for Atmega processors (fully-automated Arduino-Compatible booloader programming!).
  
• Pen plotter
  
• Dial indicator for verifying Z leveling.  The built-in auto-Z-leveling routine will be sufficient for most users.  This is more of a verification tool for us designing the machine, but will be available to buy/build for those that may need it.
  
• Laser element for photoplotting photo-sensitive boards.  Sorry, no 80-watt CO2 laser cutting   At least not at this price level.
  

Our Auto-tool changer gives you the ability to hold any four (4) tools at the same time.  The end effector will only hold one tool at a time, but it will be able to switch back and forth between any of the four tools.

That's quite a bit of envisioning, I hope they do focus on the PnP first, just writing the code to be able to handle the Gerbers is going to be a PITA.

[maliqua]

Destined for failure right there. Jack of all trades, master of none.

I don't see that really as a problem other than the tooling the code the machines are almost identical both benefit from precision equally and the stricter requirements of pick and place machines may end up making a better 3d printer than expected. 

but time will tell i'm terrible at placing those small components and don't enjoy it at all so i will be a guinea pig when it becomes generally available  

[miguelvp]

A video of some testing earlier this month:

[EEVblog]

Come on Dave! At least allow some room for benefit of doubt.  Surely whether it succeeds or fails is partly a function of the standard you want to set. Most multipurpose tools are a compromise.

You are missing the point, you don't have to make it a multi-purpose tool.
From that quote it sounds like they want it 3D print and do other things.
FAIL
If you want a PnP machine, then build a PnP, don't try and make it do anything else.

[miguelvp]

More videos from the YT channel

https://www.youtube.com/channel/UCN5wWSFjmIQEShpf8L_wfvw/videos

Looks pretty good to me, other than there is still a lot of work to be done.

As for the PnP software they are using something called OpenPnP with kicad support (still on pre alpha)

http://openpnp.org/

[ovnr]

(Preface: I haven't watched their videos because bleh.)

So yet another P&P machine where they ignore the biggest elephant in the room: Part handling - that is feeding reels, tubes, etc. I mean, the hard part is picking up a part and plopping it down roughly where the code specified it to go? Nope!

Also, if you're populating typical small boards with a non-ridiculous amount of identical components, it'll be faster to just do it manually as opposed to setting up your machine. And god help you if you have more BOM entries than your machine can fit, because you just doubled (and then some) your assembly time!


Yes, sure, it might end up working. How many boards could you have populated in the time it took to design and build a P&P machine that works reliably every time you start it up? Not worth the effort unless you build stacks and stacks of boards all day long, and if you do, fab it out to someone who does it for a living.

[EEVblog]

Yes, sure, it might end up working. How many boards could you have populated in the time it took to design and build a P&P machine that works reliably every time you start it up? Not worth the effort unless you build stacks and stacks of boards all day long, and if you do, fab it out to someone who does it for a living.

Yes, there is a very narrow window of worthwhile usefulness to these small PnP machine, which is why I don't think they'll ever be successful.
You can't magically make a cheap and simple machine that has decent parts handling usability.

[AlfBaz]

Are you people kidding me!?!?
This is a brilliant idea. PnP, 3D printer, add a router tool and enclose it in an oven.

You can then mill a PCB, mount the components, reflow them and then print an enclosure around them

I smell a kickstarter campaign "Totally Freakin Awesome" 

[miguelvp]

Seems to me they have plenty of space for feeders.

I'm more concerned on rotating the part, not clear if they do that yet, but since they are not using the camera for positioning yet, there is no point.

But they are using openCV for that

[pickle9000]

Just as a general statement, work on the software not the platform. If that is where this is going then I'd like to see what they can come up with.

[Corporate666]

I'm with Dave.  These home-brew PnP machines are totally useless. 

I spent YEARS chasing a low cost and easy to use PnP machine.  The closest I got was buying an older machine and retrofitting it with modern servo drivers and a modern controller and running G-Code with M codes for placements.  It was a total pain in the ass.  It always required constant fiddling to get it to place anything, and it's debatable whether I ever saved any time with it over hand assembling.

I've bought 4 or 5 pick and place machines over the years, and frankly, nothing comes close to a simple, used, good condition commercial PnP machine.   The problems of home-brew PnP's are...

-It's a pain in the ass handling components.  They are small, hard to reliably pick up, hard to reliably center and hard to reliably place. 
-You need something that moves really fast but really accurately, without drifting over time.

Those two problems sound simple - but they aren't... that's why there are no cheap PnP machines on the market, and why "real" PnP machines start in the $150,000 range... because getting it just right is really, really hard.

[miguelvp]

10 years ago a 3D printer would cost you more than $100K, of course better quality than the current under $1K ones. But look at computer vision cost 10 years ago, or xyz table costs, etc.

Actually, look at the whole hobby electronics 10 years ago. Manufacturing cost as well, things have changed a lot on the past decade, why will a PnP be any different than say a CNC machine and it's drop in cost?

I agree it's a tall order, but you just can't dismiss it's feasibility.

[AlfBaz]

10 years ago a 3D printer would cost you more than $100K, of course better quality than the current under $1K ones. But look at computer vision cost 10 years ago, or xyz table costs, etc.

Actually, look at the whole hobby electronics 10 years ago. Manufacturing cost as well, things have changed a lot on the past decade, why will a PnP be any different than say a CNC machine and it's drop in cost?

I agree it's a tall order, but you just can't dismiss it's feasibility.

I tend to agree with this sentiment especially as it's open source.

If this doesn't work out, or it does but it takes forever to set up/maintain, the fact that it's open source allows the next attempt by other parties to have a starting point. This gives the next mob the ability to concentrate overcoming deficiencies found on the previous unit.

There's bound to be a tipping point where interest in this idea will reach a level that more and more outfits will put R&D into it

Having said that, more R&D may very well show that its not feasible.


 

[george graves]

I don't see anything special on a small desktop PNP that can't be home brewed.  Yes, it's going to have limitations like any tool - but to say it's "worthless" isn't correct.  Sorry.

[maliqua]

I still see merit in it for the hobbyist anything smaller than 0805 imho gets hard to manage by hand its doable but its not just quick easy work at least not for many of us, and doing really compact 0402 size is essentially out of the question for most of us. I for one don't have the dexterity.

To me it opens up a level of miniaturization currently not available to people of my skill level

I'm not saying it will work, but if the goals they set for  it are achieved in the final product at the price they mention. I think it will for me at least, be worth the money.

[ovnr]

Yes, it's going to have limitations like any tool - but to say it's "worthless" isn't correct.  Sorry.

I think of any tool that's supposed to save you time and effort, but which fails to do either, as worthless. Sorry.

[miguelvp]

Yes, it's going to have limitations like any tool - but to say it's "worthless" isn't correct.  Sorry.

I think of any tool that's supposed to save you time and effort, but which fails to do either, as worthless. Sorry.

I guess it's doomed already because you'all made up your minds, run for the hills! 

[Precipice]

Just as a general statement, work on the software not the platform. If that is where this is going then I'd like to see what they can come up with.

Yep - there have been plenty of machines that get to the 'ooh, we can place a few components' stage, and then the enormity of the software effort and getting working feeders kicks in.
This one impresses me the most - I wish people would join in, rather than starting yet again from scratch, but hey ho, the early stages of the project always seem to be the most fun and fastest moving.
http://www.vbesmens.de/en/pick-and-place.html
sort of maybe getting supplied by
http://www.placecomponents.com/

[EEVblog]

Those two problems sound simple - but they aren't... that's why there are no cheap PnP machines on the market, and why "real" PnP machines start in the $150,000 range...

Not really. You can get a useful base machine for $10-20K. The ones of the likes that Adafruit and the smaller players like to use are in the order of $50K IIRC.
And of course there is a big 2nd hand market in these things.
Even Mike has a usable PnP machine in his shop, and I think it was about $5K 2nd hand?

[tridentsx]

In this case its exactly what has happened, firepick is a fork and extension of open pnp which has been around for a few years. The firepick fork mainly has added better visioning utilizing open CV.
Give it another couple of years and a few forks and I think the s/w will be in usable shape. There are at least 4 other designs that uses open pnp s/w base that are functional machines with one working in a small production environment.
Time will tell where this will lead, I have followed firepick since inception and I am surprised how far they have gotten. I always thought the two guys were crazy with their goals of a sub $500 machine that can place ~ 1000 cph.

I think open pnp for example could be used to upgrade older systems that doesn't have sophisticated computer visioning.

[Mechatrommer]

what my thought is... 3 in 1, 3d printer, pnp and cnc machine. i just dont have the time to do it on my own. someone briliant and patient may do it. once was thought impossible now is possible, ignore the naysayers, engineering is about solving problem. but remember to get the fruit, hardwork is needed and patient, real hardwork and real imagination no kidding.... talking about master of none... i just recently bought/upgrade to 1, infact without this model i dont need the rest (the norm) of the androidiotic world, my 7 years old nokia 7610 did the job flawless, now handed down to kiddo for symbian gaming only they love it so much. but with this master of none thing i dont feel like to bring my heavy 40d around again...

[EEVblog]

Actually, look at the whole hobby electronics 10 years ago. Manufacturing cost as well, things have changed a lot on the past decade, why will a PnP be any different than say a CNC machine and it's drop in cost?
I agree it's a tall order, but you just can't dismiss it's feasibility.

You can, because you can't compare a PnP machine with a 3D printer.
A 3D printer squirts melted plastic through a nozzle - that' it.
A PnP machine has to place a countless variety of parts from reels, peel back the reel tape, rotate them, optically detect and check them, use vaccum to suck up tiny parts, and do it at a speed that makes it usable.
The reel handling is the show stopper, it's not easy to get this right, which is why commercial reel handlers cost big dollars, and can even cost more then PnP machine itself if you get enough of them.
It's all huge trade-off between being able to handle enough reels to make a reasonable board in a single pass, handle the reels, get the opticals right, learn to program it and be on-hand to massage the machine when it fails (even the best ones do, and need full time attention) etc.
You can't magically do all this for a low cost.
You can calculate and test how much time you save vs either hand placing yourself, or simply giving it to a shop that has a better machine and better people who know how to use it and can man it 24-7.
And when you look at it, it's a very narrow windows, and almost certainly always will be.

[EEVblog]

I guess it's doomed already because you'all made up your minds, run for the hills! 

It ultimately comes down to how you value your time.
If you have endless time that in your view costs you nothing, then fine, use a DIY PnP machine, you can make it work.
It might take 5 passes of your board, tweak it after every 10th part, fiddle with the tape, change the reels, set it up and trial it etc.
But ultimately you will never ever be more efficient than a professional level machine and someone who works cheap who's full time job it is to use and maintain these machines. Or simply placing the parts yourself by hand.

Remember, the only time the PnP machine saves you is hand placing the parts. You still have to manually stencil your board, and then manually reflow it.
So it comes down to a matter of trading off time to set up a hand placing system or set up the PnP machine.
And then trading off hand placing time to watching the PnP machine and fixing the inevitable things that go wrong. Even the pro ones aren't really set and forget.

There are certainly some instances were a home made PnP machine could be very useful. Say for example you had a board with 1 hundred 0805 resistor and caps and one big chip. Well the PnP could be useful for the huge number of parts, and the one awkward chip could be placed by hand.
But if you have a complex board with 30 different parts of all different physical sorts, well, good luck with that, you are probably wasting your time even trying to get it working.

It's just like making your own PCB's, there is a very narrow window of justification when you can get a professional PCB made for $5/sqin