Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


Possibility of a cheaper pick and place machine design?

Views: 9482

Hi (I don't know if this forum is the best place to post ... - Jul 04, 2014 by Andres Pena  

#72370

Possibility of a cheaper pick and place machine design? | 4 July, 2014

Hi

(I don't know if this forum is the best place to post something like this, but I don't know of other places on the internet, so... sorry if I'm in the wrong place)

Might there be an easier way to add components (e.g. computer chips) to a circuit board? I'm mean a slower but maybe more convenient machine.

First, put a small magnet on the circuit board. The small magnet can be moved around by if we pull it with a large magnet underneath the circuit board. The large magnet underneath may be mounted to moving parts, controlled by a machine.

Second, drop a computer chip on the circuit board. Using a video camera, a computer program can locate the chip, and push the chip to it's correct location by moving the small magnet, using the large magnet.

Once everything is in place, you could keep spraying the circuit board with glue until all the chips and other components sinks in.

Note...

Maybe the small magnet should have low friction, because it presses against the circuit board and you don't want it to scratch the circuit board too much. Maybe the small magnet should be shaped like a cross or a hash so it can push components easily. Maybe it can have four "hands" which look like--

. . \

. . . .\____________________

. . . . ____________________|

. . . ./

. . /

(a funnel)

so that it can move many components at a time.

Maybe the inner walls of each "hand" can have high friction, so some components will get stuck and each "hand" can only be unclogged if the magnet moves in a specific angle. The inner wall friction is equivalent to a grab and release ability.

When it is finished with one circuit board, the magnet can push it off the workarea and pull in the next circuit board from a stack. The magnet could climb on and off the circuit board using a ramp with a flap. To pull in the next circuit board, it may hook to a rubberband or string band that's taped to the circuit board. When it runs out of chips/components, it can also pull in more components (in paper trays) from outside the workarea.

The process might be very messy. Most components may fall off the workarea, so many trays of components are needed.

Another problem is that bits of glue may miss the circuit board and hit the workarea instead. Perhaps a sheet of paper should cover the workarea. Also, the magnet may stay clean by taking cover under a tent.

Another problem is that some computer chips and other components may be attracted to the magnets... perhaps in those cases the magnet can still push them from a distance, using a long shovel as a tool.

The magnet might be able to wield another useful tool: the silver ink pen, for drawing circuits on boards. How do we hold the pen upright? Perhaps we can give it "training wheels" like the ones on a bike--except longer, and with tiny wheels.

One possibility: the magnet holds the pen tip deep inside one of its "hands." Also in the "hand" is a small ramp. To draw a line with the pen, the magnet simply pushes the pen; to move the pen without drawing a line, the magnet moves in a special direction to push the pen onto the ramp.

Finally, the most difficult components to add will be push buttons. How do you spray a sea of glue on one without jamming the button? Maybe instead of gluing down the button directly, you could surround it with hood shaped plastic parts (made by 3D printers), and glue those down to secure the button. To prevent any glue from falling on the button itself, maybe you can add a 3D printed umbrella tent to cover it.

Will the suggestion work?

reply »

#72371

Possibility of a cheaper pick and place machine design? | 4 July, 2014

hmm... the picture of the funnel didn't display very well because it's double spaced.

Just out of curiosity I have to ask, does anyone even understand what I wrote?

Does anyone know where on the internet I should actually post something like this?

reply »

#72373

Possibility of a cheaper pick and place machine design? | 6 July, 2014

There are a few Open Source projects for this sort of thing

for instance

http://hackaday.com/2011/11/30/openpnp-working-to-create-an-affordable-and-completely-open-pick-and-place-machine/

Also try my webpage it has some links to these and pther projects.

http://members.iinet.net.au/~sarason/Links.html

regards sarason

reply »

#72375

Possibility of a cheaper pick and place machine design? | 7 July, 2014

If you study how the process is done now, it isn't that different from what you describe. Instead of glue, we use solder paste, and instead of spraying it we screen print it. But not that different. For feeding boards we use conveyors instead of rubber bands, and for placement we use vacuum needle instead of magnets. But none of that is very different.

The amazing thing about the existing pick and place machines is not how expensive they are, but how CHEAP they are. A decent pick and place machine bought new today would cost something like a hundred thousand dollars. That machine should be able to place about 20,000 parts per hour. if it is running 24/7 and placing a bit more than half the time that is 8 million parts per month. In a year that is about a hundred million parts. These machines can easily last ten years. So this $100,000 machine may place a billion parts before it is worn out.

A typical consumer gadget has a hundred parts or so. The cost of the machine to place those hundred parts, is only one cent. Not one cent per part. One cent total. That's pretty cheap.

What's not cheap is the cost of making scrap boards. The operator who knows how to run the machine is not cheap. The components are not cheap. Repairs are not cheap. Rent for the building isn't cheap. Programming new products into the machine isn't cheap.

The machine itself, though? They almost couldn't be cheaper if they were free. Despite costing more than my house.

reply »

#72377

Possibility of a cheaper pick and place machine design? | 7 July, 2014

Thanks for the info because I honestly have no darn clue about the whole industry.

Does anyone know why it is expensive to have an operator? If it has conveyors can't it operate on its own?

Why are the boards expensive? Why wouldn't any rigid piece of plastic work (excuse my ignorance)?

reply »

#72378

Possibility of a cheaper pick and place machine design? | 7 July, 2014

@sarason Should I ask them if they're interested in this? I'm not sure if I should bother them, because I already posted on this forum and emailed some group with no reply so far.

reply »

#72379

Possibility of a cheaper pick and place machine design? | 7 July, 2014

The problem about covering it with glue (without using solder) is that the components might dislodge by a miniscule amount underneath the glue. So maybe the contact points will disconnect.

I'm not sure if this will be a real problem, but if so it would be a big one. Maybe some types of glue can prevent the problem by pulling the components down using elasticity. Maybe a 3D printed shell can clamp down on all the components, although that would raise the costs greatly.

reply »

#72382

Possibility of a cheaper pick and place machine design? | 8 July, 2014

The boards are expensive because they are your product... covered in circuitry. A 2% scrap rate costs twice as much as a 1% scrap rate.

Say the parts on your consumer electronics board cost $5. A machine that makes 99% good boards and 1% bad boards has a defect cost of 5c per board. Remember that the machine itself only costs one cent per board at $100,000 purchase price. So even with a good machine making 99% good boards, the scrap cost over the lifetime is $500,000.

A really cheap machine making the same boards might have a 5% defect rate. Say the machine only cost $20,000. That seems like a better deal, until you realize that a 5% scrap rate would be equivalent to a scrap cost of two and a half million dollars. Penny wise and dollar foolish.

If the boards are expensive enough you'd rework them rather than scrapping them, but rework is also expensive.

It can operate on its own... except when it needs to be set up, or loaded, or isn't placing well, or a reel runs out, or changes have to be made. If you figure that an operator costs $30,000 per year, and the machine lasts 10 years, that's $300,000 in operator costs versus 100,000 in machine costs. If the machine runs better so an operator can run four of them, that's only $75,000 in operator costs over the lifespan instead.

reply »

#72389

Possibility of a cheaper pick and place machine design? | 9 July, 2014

We had a Tim at Daycounter engineering ask about nozzle styles for placement machines they intended to build. I do not know if they have brought anything to completion yet. Here is his contact info if anyone wants to inquire: Tim Daycounter Engineering 13940 S. 4000 W. Bluffdale, UT 84065.

reply »

#72403

Possibility of a cheaper pick and place machine design? | 10 July, 2014

Hi,

Nice idea but it requires care electromagnetic induction. Thus, if a product sensitive electromagnetic field then the solution will cause shorts in a PCB and even disconnect signals.

Best Regards, Alexei

reply »

#72406

Possibility of a cheaper pick and place machine design? | 10 July, 2014

@Comatose You answered all my question about the board and operator costs, and I'm getting a good idea of how the numbers look like.

reply »

#72407

Possibility of a cheaper pick and place machine design? | 10 July, 2014

Smaller companies many want to rent a machine, but may not necessarily use it 24/7, and a smaller machine may require less expertise to use. If many small machine can work fast enough, it can operate for less time and reduce operating costs.

(I'm more thinking along the lines of how the machine is designed than quality/functionality, so the failure rate should not be too high.)

reply »

#72408

Possibility of a cheaper pick and place machine design? | 10 July, 2014

@Alexei Shkolnik Ah, I hope this doesn't become a problem. But if it does perhaps we can redesign the whole system? Push the components with a pusher that's connected from above rather than pulled by a magnet below.

Perhaps the pusher can still climb on and off the circuit board by lowering itself when it's off the circuit board. Or maybe the pusher can be made of two parts, the upper level and the lower level. The lower level detaches in order for the pusher to move above circuit boards. The pusher can detach its lower level by locking it to a station and moving away.

reply »

#72431

Possibility of a cheaper pick and place machine design? | 16 July, 2014

My gut feeling is that I should do more than just post it here... the suggestion will ultimately be forgotten.

Does anyone know where on the internet I should actually post something like this?

reply »

#72541

Possibility of a cheaper pick and place machine design? | 7 August, 2014

Why use magnets? I found a KS project of a pcb prototyping machine that uses vacuum for PnP. Here is a video https://www.kickstarter.com/projects/botfactory/squink-the-personal-electronic-circuit-factory/posts/929661

reply »

#72659

Possibility of a cheaper pick and place machine design? | 29 August, 2014

I'm very sorry I did not see this reply earlier. It's been almost a month and I hope you get to read this!

I really like your idea of a *personal* electronic circuit factory. You should try getting schools to buy it, because students will love the opportunity to design their only electronic contraptions and quickly test them.

But anyways, let me get to this question about why magnets might be better.

First, just to make sure everyone knows what I'm rambling about: the magnets push the components into place. They should not attract the components, just push them.

The advantage of pushing, I think, is that we only need to move horizontally in a plane, so there are less moving parts. The machine will take up less space and less maintenance. At least I think it would...

So one operator should be able to build more boards at a time.

The disadvantage is that I don't have the resources to test this idea, so I don't even know if it'll work. Will the magnets damage the components? Can we make the software? Will there be hidden quality disadvantages? (As one mathematical person wrote above, a tiny quality disadvantage will outweigh any cost advantage)

reply »

#72661

Possibility of a cheaper pick and place machine design? | 31 August, 2014

Andres, I think if you were to understand the process a little better you would realise the many many problems you would encounter building even very simple things using your method. Even 30 minutes watching Youtube videos of current machines might help. You need to get your parts on the board befoe you push them - how? You need to lift parts to get them out of pockets in the tape they come in. You need to lift parts, over the pasted pads you are working on. Devices these days don't just have solderable contacts on the edges. BGA's and QFN's have them underneath. Some people ven use these machine to place connectors and LEDs that fit through the board. Even manual pick and place machine lift the device, a person with tweezers lifts the device. Can people make cheaper machines... yes of course they can. But they do that by making them more flimsy, slower and with crappy software. Someone with tweezers can built a board at about 200 cph unless its very simple. With a manual pick and place, maybe 400cph. With a manual pick and place with computer assist perhaps 6-800 (its amazing when building by hand how much time you spend looking for where the part belongs. A cheap Chinese tabletop pick and place can have a go at these tasks @ 1500cph+ which is about the same as very basic entry level machines manage but the Chinese one does it badly for 1/10 the price. I work in a small company building small simple boards in small batches. Even so that means I build things with 50 parts numbers, 400 parts to a board and place up to 50K parts in a day. Some people have machines that can do that in under an hour. To do that not only does the machine need to be accurate and consistent, but it needs feeders that supply it with the parts continuously. 90% of people with surface mount machines probably spent more on the feeders that do this than the core machine itself.

reply »

#72736

Possibility of a cheaper pick and place machine design? | 10 September, 2014

Sorry 10 days passed before I replied. This happened and that happened and I completely forgot about this forum.

But here's my explanation:

1. How to get parts on the board?

This is a very good point, and I totally understand why you believe I haven't thought about this. Somewhere in the walls of text above, I briefly mentioned this problem already, but my solution was badly designed + badly explained so let me tweak and rewrite the solution.

The parts (computer chips, transistors, etc) are scooped into trays. Maybe paper trays. The trays are strung together with string into a chain. The small magnet pushes one tray off the platform and pulls the next tray in. Maybe the magnet can flip or tilt the trays somehow to take the parts off the trays. It's messy I know.

2. BGAs and QFNs

I'm terribly sorry, but my technical English skills are bad and I don't know what question you're asking about BGAs and QFNs. What problem do you think they will pose to my design?

3. Can people make cheaper machines... yes of course they can. But they do that by making them more flimsy, slower and with crappy software.

I may have misinterpreted you here, but let me guess: are you're saying that the reason some machines are cheaper is because the software is bad? If that's the case, it's irrelevant. My design makes the machine cheaper by simplifying the mechanics, not making the software worse.

However, another interpretation of you're sentence, is that you think it's hard to make the software for mydesign. If that's what you're saying, I agree absolutely: the software programming will be a bottleneck, and a big one to worry about. My design indeed requires more intelligent software than a suction based machine.

4. Some people even use these machines to place connects and LEDs that fit through the board.

I don't know how that works, but it sounds plausible.

5. "manual pick and place machine"

Are you talking about something like this? http://www.manncorp.com/smt/prod-179/smt-place-2000-manual-pick-and-place.html Or just plain tweezers?

6. Cheap "Chinese" tabletop pick and place

Are you talking about something like http://www.youtube.com/watch?v=yRxcYO0nuD8

A commenter above named Comatose talked about how when it comes to saving money, and good quality pick and place matters a lot.

7. ...

Anyways, I did think of another problem with the my design: in the video (http://www.youtube.com/watch?v=yRxcYO0nuD8), the demonstrator mentioned solder paste on the board. The magnet pusher might scratch the solder paste while pushing the components, causing damage. Perhaps using magnetism to move the pusher is a mistake, and we should use a hanging pusher controlled from above.

reply »

#72739

Possibility of a cheaper pick and place machine design? | 11 September, 2014

1 - Too messy, parts already come in nice easy to use machine compatible packaging, tapes, reels, tubes and trays. All you are doing is adding an extra step and complication. A key production aim is to reduce the number of steps. 2 - BGA - a chip with solder balls underneath it that make all the connections. QFN - a chip with flat pads on the underside, typically all around the edge of the device and then a large central one. You identify the key issue here in your last point - the paste on the board that solders these in place. They have to come from above. 3 - Cheaper machines are cheaper because EVERYTHING about them is bad/considerably worse than the more expensive ones. Rigidity, build quality, software, mechanical design, cameras the list goes on. 4 - plenty of videos on youtube of that. 5 - yes pretty much, some can have cameras, laser pointers and a screen to help the user locate the position required. 6 - Yep, its bad. Maybe that's OK if you really are one man in a shed, and you don't mind adjusting the placement afterwards of every part. In reality even one man in his shed has better things to do. That is why we have customers that are one man in his shed.

I don't really understand how you think you can make it simpler. You still need to build a cartesian robot for your solution. To be accurate it ideally needs some form of location feedback from something like an encoder. There is nothing complicated or expensive about a vacuum. For a basic low end machine you can do a pretty good job with a $200 pump and skip the need for a compressor. You should also consider what else makes machines expensive: Distributor markup - These guys have to sell it and make money and support you afterwards. Inherent value - These machines make product, that makes them mission critical, which means they are worth money. not simply the value of the sum of their parts, but real value to your business because without it you cannot operate. This allows manufacturers to charge for that value and we will pay, because we need it. Intellectual Property - these machines are full of bespoke parts and hi-tech sensors and algorithms, they cost money. Volume - this isn't an iPhone, there is no significant economy of scale.

reply »

#72741

Possibility of a cheaper pick and place machine design? | 11 September, 2014

Not to squash anybody's ingenuity, but most of the problems that you're trying to solve have been solved, and fine tuned, over the last 30 years of SMT. (Surface Mount Technology)

I'm all for reinventing the wheel, but I suggest taking some time to try and grasp a concept of the "wheel" before trying to reinvent it.

But I must say, I admire your passion.

reply »

PCB machines