Pick and place machine kit, with vision

[JuKu]

Ever since I couldn't find one when I needed one, I have discussed about the feasibility of an affordable pick and place machine on this forum every now and then; especially one that would be aimed for prototype building. Sometimes the discussions have turned to polite arguing (hi Mike!), and at the end of the day,  the conclusion has been that we don’t know if there is room between fully manual methods and real production P&P machines, unless someone tries to find out.

I put my money where my mouth is, and built it. Building a finished machine is far too big effort to be worthwhile for one hacker only, so I took the extra effort and made it available for all: http://www.liteplacer.com.

On my site you'll find lots of info, software (with source), shop for a kit, support forum and more.

Please let me know your thoughts!

-Juha

[elgonzo]

I must say that i am blown away by the amount of information and documentation about your project.     
And while i am not in market for a PnP machine, i will certainly visit your site and reading some of the stuff when i need to kill some idle time.

[mikeselectricstuff]

Very nicely done, but of rather limited usefulness, and  questionable cost-effectiveness.

It's no faster than manual placement, even before you count setup time,  and without proper feeders, of rather limited use.

If you build SMD prototype boards more than once a year, I believe you need one.

that is just ridiculous. Once a week, maybe, in some cases.

 

[sleemanj]

I think you could save a lot of seeking time there, every time it picks apart it appears to seek diagonally to a home position, pauses half a second, and seeks horizontally to the part, this seems unnecessary, you should just seek diagonally to the next part.

[JuKu]

I think you could save a lot of seeking time there, every time it picks apart it appears to seek diagonally to a home position, pauses half a second, and seeks horizontally to the part, this seems unnecessary, you should just seek diagonally to the next part.

That is the camera measuring the exact location of the part, then taking needle there. Measuring each part means that the user doesn't need to be precise when placing the tape strips. That would be time consuming; this way approximate tape location is fine, the machine can see the exact location.

[EEVblog]

Very nice, but I have to agree with Mike.
This looks no faster than hand placing, and therefore of limited usefulness with those all important feeders.

[sleemanj]

That is the camera measuring the exact location of the part, then taking needle there.

I see, a reasonable solution.

But perhaps you could have a tape setup routine which looks at the first part on the tape, and the last part (or just X parts along depending on your necessary accuracy), and from that you should be able to extrapolate positions for parts on a that straight tape of arbitrary length, once you have the positions, no need to look at each part before pickup.

[timb]

Something like this could be very useful for people with vision or ambulatory issues that physically can't do SMD work. Partial Blindness, Non-Stereo Vision, Arthritis, Handicaps or even just age.

Hell, after a bit of caffeine I can't even do precision placement reliably by hand.

Don't listen to the nay-sayers, keep up the great work! I might even look into building one.


Sent from my Smartphone

[rx8pilot]

I do a TON of manual assembly and I see this as useful. Here is why...

Manual assembly takes a lot of time to setup just as a machine does. If I am doing 2-3 PCB's, I will simply spread out the components on the bench and get going. This means that for every component, you have to find the tape and the then find all of the instances on the PCB. If the PCB has 20 components, not a problem. If the PCB is like the ones I have, there are 200+ per side and many are so close together that silkscreening is not helpful. When I know that I need to make a small batch of the PCB's - say 25 pcs or so - I organize it much better. I print color coded lists, labels, and PCB references and organize the components on special color coded trays in the order they get placed. This takes a lot of the thinking part out of the chore and leaves the physical placing. No delays trying to remember where the pads are because it is all color coded and sequenced. I also designed and CNC'd special PCB holders and some other fixtures to help out.

My hands are very steady and so I am able to place fine pitch components with no problem. The delays are my limits of focus as with most people. There is only so long that I can place components at "full speed". I can have bursts of speed, but it is not sustainable and the average speed over a day of placing is very slow. I designed and built a "manual" PnP but it was the same speed at best compared to my freehand methods.

What I like about the Liteplacer is that it will place PCB's at the same rate without any thought. Even if that rate is slow, it does not require me to think or do after it is setup. I can be printing the next PCB's or tending the batch oven or testing the PCB's while it places components. For me, a 3000CPH machine would be pointless since I cannot print, reflow, test the PCB's that fast anyway.

The feeders will be important for sure if you want to make more than a couple of PCB's. Fortunately, I transitioned into EE from ME. I spend the better part of the last 10 years designing and CNC machining in my own shop. I could easily modify the table and create some simple feeders. My PCB's are usually two-sides so the table needs to be changed. I am inclined to get a kit and modify it with upgraded mechanics to suit my needs . For the price, if it does not work, I would not shed a tear.

[zapta]

Do you have a picture of the entire machine?  What are the overall dimensions? What is required in addition to the kit (e.g. vacuum pump?, power supply?)?

Looks very impressive.

[Kjelt]

Great job! For the hobbieist that makes small series of pcb's it looks like a really good machine. 
You must have a lot of spare time to do this project.

[JuKu]

Do you have a picture of the entire machine? 

See http://www.liteplacer.com/plan-your-work-area/.

What are the overall dimensions?

About 91 x 54 cm board area is required. Depending on the cable management solution, roughly 100 x 65cm overall.

What is required in addition to the kit (e.g. vacuum pump?, power supply?)?

Vacuum pump is included. The items not included in the kit are those that a typical kit builder is likely to have already or where adding them would not add value to the product:
•Power supply
•Cables and wires
•Connectors, termination blocks, ferrules and other cabling accessories
•Drag chains or other cable management
•Enclosure for electronics
•Reset switch and mount
•Work table, screws to secure to machine to it
•Computer

Looks very impressive.

Thanks!

[JuKu]

You must have a lot of spare time to do this project.

It is winter time, golf courses are closed.

[rx8pilot]

I think I will commit to one of the first kits.

What could go wrong? I have an electronics lab and a CNC machine shop. I have done motion control design and have a pile of Maxon servos, 8 axis Trio motion controllers and drives. I have never coded in C# but I can learn as I am well versed in C. We shall see if it can place .5mm pitch QFN's.

First thing is to get 2 sided PCB capability for me. Second will be figuring out feeders that I can swap quickly to change jobs out.

[timb]

I'm wondering if you could add solder paste dispensing to this? Perhaps replace the part sucker head with a custom syringe and injection needle. Just use same airline as the suction head but switch to a compressed air source instead of the vacuum pump.


Sent from my Tablet

[JuKu]

I'm wondering if you could add solder paste dispensing to this? Perhaps replace the part sucker head with a custom syringe and injection needle. Just use same airline as the suction head but switch to a compressed air source instead of the vacuum pump.

Unlikely to happen, at least very soon, for a few reasons:
- Using a stencil gives much more accurate paste amount, and that is crucial for reliable results.
- A stencil costs about $10; the price of solder paste add-on buys quite a few stencils.
- there are more crucial development directions, as mentioned on this thread: Speed and use of feeders. For speed I already have several directions to pursue, and too many people ask about feeders although personally I'm not that convinced that they are necessary for a prototype machine. Laying tape on table is easy, fast and reliable.

The software is out already, and the mechanical files will be very soon. I am certainly not stopping you pursuing that direction!

[Kjelt]

Real feeders seem quite difficult to make esp. to make them reliable.
Selling some aluminium plates, each with lanes to hold different tape sizes would already help a lot since you only have to teach the machine once the position of the start of each lane.
You then number the lanes and you only have to set the lane# in the software for the correct component fetching.

[JuKu]

Real feeders seem quite difficult to make esp. to make them reliable.

Especially at this price point. Besides, feeders imply production environment, which is NOT the point of my machine.

Selling some aluminium plates, each with lanes to hold different tape sizes would already help a lot since you only have to teach the machine once the position of the start of each lane.
You then number the lanes and you only have to set the lane# in the software for the correct component fetching.

That is how the software works now. The optical recognition of the tape hole means that the "teaching" is only pointing out the area to look at. I build in the flexibility to be able to have tapes in any orientation; you can feed them in from left, right, front or back. The price is that you need to tell the machine that.

I actually experimented aluminium plate/reel holders, that you can buy rather cheaply, such as http://www.goodluckbuy.com/smt-smd-feeder-for-diy-prototype-pick-up-place-5-way.html. Turned out those are more trouble than worth in this application:
- The tape location is not accurate enough in horizontal direction => part position is not accurately found
- The bottom of some SMD tapes is not that strong => Without bottom support, the pickup operation needs to be very sensitive in order not to push a small part through the tape. Having a feather light touch brings in loads of other problems
- The tape location is not accurate enough horizontally, the slot is loose => If a paper tape is twisted so that the tape is as high as the slot allows, parts tend to be pushed through the tape or the pickup operation does not reach the bottom parts when the tape is twisted the other way
- Peeling the cover tape needs to be done very, very carefully if the tape is not held firmly.

The K.I.S.S. principle applies: There is nothing wrong with double sided tape, other than it does not feel as high tech sexy. :-)

[rx8pilot]

I was thinking about making a few plates that can be prepped offline and loaded into the machine. I have done the tape trick for manual assembly and it drives me nuts. I machined a few tape holder blocks to slide various tape sizes into. This allows prepping the assembly job on a table where I have boxes of components laid out.

I don't see any need for real feeders that hold reels, but similar to what DDM Novastar does on the low-end machines. Simply tape lanes that you manually advance a few inches at a time.

I do agree with the paste dispenser. A stencil is worth the price generally. I have a Zephyrtronics dispenser that does a good job but I only use it when there is no alternative.

[IconicPCB]

Juku,

Great effort.

I have a small prototyping machine which has both smart feeders and the capability to process trays, cut tape and loose bulk components.

Good for up to 1600 components per hour.

Placing 200 to 400 components per hour is not that easy, Your machine should be a great boon.

Keep up the good work and dont listen to the wankers ( aussie slang for nay sayers ).

[JuKu]

I was thinking about making a few plates that can be prepped offline and loaded into the machine. I have done the tape trick for manual assembly and it drives me nuts.

I'm thinking of a 3D printable design of a tape holder that pushes the tape down from the edges (getting vertical position constant and supporting the parts from the bottom) and spring loads the tape both sideways (getting accurate position sideways). The latter is not that important, the former is.

[rx8pilot]

I machined pieces like that. 3D printing is very rough for this kind of application IMHO. Its an edge guide so the parts, like you said, are all at the same Z level and precisely guided for X/Y predictability. Very simple to make and use for sure.

My previous career was ME and CNC machining, so I have have bad habit of fixing anything with a machined part  I would machine my clothes if I had more time.

[IconicPCB]

A variety of cut tape feeders in use.

[rx8pilot]

That is a lot of tape feeders. What machine is that?

[Corporate666]

Excellent effort, and I don't want to be overly critical, but I agree with Mike and Dave - I really don't see the point.

Hand placing would be significantly faster (especially when you include setup time), and since the machine doesn't support feeders or auto loading/unloading of boards, it's not like you can let it run all day on it's own.  The placement accuracy also seems quite poor... seems there is 0.010" and/or several degrees of error on many of the placements. 

I don't think it comes close to buying a used machine (I always recommend a Quad 4C) for the $5-7k range which will likely come with a couple dozen feeders, be orders of magnitude faster and more accurate and have factory support if necessary.  I know Mike has another brand of machine he's used that also seems quite capable, and perhaps even less $$.

Picking and placing is not a difficult task.  What makes it hard is performing the task quickly and accurately, as well as the feeding of components.  This device doesn't address any of the difficult aspects of PnP'ing.

But still, a significant achievement in difficulty and you should be proud of your work.