Open Placer A New Affordable PCB Assembly Machine

[rx8pilot]

So using a pnp it's not only time saving, it's energy saving too.

This is an important point for sure.....I certainly was nearing burnout by making too many boards. There were a number of days that I placed parts for 10+ hours which is soul crushing.

You also need space.

Exactly! If you look in Ebay there are so many nice machines Assembleon, Juki, Siemens or whatever for just 10-20k dollars but holy shit they are BIG and HEAVY! It would be so nice to buy one but living in a single room apartment they for sure aint't going to fit thru the door or in the elevator! You have to have some workshop space or garage at least and then that needs some real money.

This is very true and very much a limitation beyond just the purchase price of a used commercial machine. I had to rent a trailer and pallet jack as well as bring in friends to help with my old/heavy/huge machine. The size and weight of the Quad machine is partly to thank for it's ability move fast and place 01005's, uBGA's, etc - but it has been a major pain.

I think picking loose parts from a tray using vision is entirely doable at minimal cost - cameras are cheap. Tatmay be a good solution to the fiddliness of dealing with short bits of tape

If this became a reality for low-cost machines - it would change the landscape entirely. Cameras and vision software are not exotic at all - seems like it would dramatically lower the cost by eliminating expensive feeders and dramatically lower the setup time by just dumping parts into little trays. It can also overcome some geometry errors in the mechanical system - further lowering the cost.

Cut tapes and mechanical advancing feeders are a pain.

It is a fair price i recon BUT.. what i'd spend to have 300pcs assembled would pay of my own cheap pnp, i dont think one would buy a 10k machine or more to make 300pcs a year but 1k machine to help you do it? easy choice for me

The price of the machine is only a small part of the price of the solution. Your time is - by far - the largest cost.

No doubt you mean well but you are unable to see the market for this thing because all arguments are made of starting a business, pro machine this, that, stop it already. You clearly suffer from Tunnel vision.

I never said there was no market. The posts I made simply voiced my opinion about how a small business like mine would view a machine like this. If the designer is thinking every small business is going to jump on this, it is good for them to know how potential customers view the concept. There are people that can make use of this for a variety of reasons. My belief is that it is a small market - but then again this is not my project or business, simply an opinion.

I have a habit of looking at the time spent trying to work with low-end tools. In some cases it is totally worth it. I got a cheap wire cutting machine from China. Compared to the $8k pro machines it sucks, but it was only $1k and it does most of my wire cutting 'good enough' it is a success story for low cost. I have a dozen other stories about low-cost stuff that simply obliterated my time trying to get it to work. Sometimes my cheap tools are good enough and sometimes they drive me nuts.

I would love to see a small machine pull enough productivity to be justifiable. The definition of 'productivity' is obviously going to be different for everyone. For me, I don't care much about placement speed. I care about ease of setup, number of unique parts, placement reliability, and the ability to do 0402 passives. The first 3 of those are directly related to the time I am hoping to save by setup up a P&P system. In terms of process reliability, I may be better off with 4 independent machines that each hold 20 parts - similar to how a large assembly house is setup. If one machine is having problems - I still have 75% capacity left over. If I have any problem with my single machine, I am immediately at 0% capacity. If there was a small machine that that worked well - I would consider buying 4 of them.

My best guess is that the Quad does not have lots of time left. Parts are becoming harder to find.

[mikeselectricstuff]

I think picking loose parts from a tray using vision is entirely doable at minimal cost - cameras are cheap. Tatmay be a good solution to the fiddliness of dealing with short bits of tape

If this became a reality for low-cost machines - it would change the landscape entirely. Cameras and vision software are not exotic at all - seems like it would dramatically lower the cost by eliminating expensive feeders and dramatically lower the setup time by just dumping parts into little trays. It can also overcome some geometry errors in the mechanical system - further lowering the cost.

I don't think it's a fundamentally hard problem, but it does require tight integration with the P&P system. I vaguely recall Jason (OpenPNP) may have done some experiments with it.
There will be some work in figuring out what materials to use for the trays, and the best way to "tap" it to flip parts over, but the vision is simply finding rectangles- how hard can it be..?
The nice thing is that when it's figured out it is inherently very cheap, and the number of "feeders" is only limited by how much space you have for tray compartments.
You could easily extend the concept to a lot of IC packages - even if you had to lay them out with the right approximate  orientation ( as automatically finding pin 1 would be hard) , the machine should be able to find them easily.
 

[KE5FX]

There will be some work in figuring out what materials to use for the trays, and the best way to "tap" it to flip parts over, but the vision is simply finding rectangles- how hard can it be..?ne should be able to find them easily.

1. Pick the part up from the tray
2. Move arm to well-lit area where it can safely...
3. ...drop the part.
4. Examine dropped part with camera
5. Part oriented correctly?
6. If not, pick it back up, go to step 3 and ...
7. Else pick part back up, remembering its rotation so you can...
8. ... place it correctly.

There, I saved you guys all that R&D time.    You're welcome. 

Now for the love of God will somebody please stop talking and do this?

[rx8pilot]

There will be some work in figuring out what materials to use for the trays, and the best way to "tap" it to flip parts over, but the vision is simply finding rectangles- how hard can it be..?ne should be able to find them easily.

1. Pick the part up from the tray
2. Move arm to well-lit area where it can safely...
3. ...drop the part.
4. Examine dropped part with camera
5. Part oriented correctly?
6. If not, pick it back up, go to step 3 and ...
7. Else pick part back up, remembering its rotation so you can...
8. ... place it correctly.

There, I saved you guys all that R&D time.    You're welcome. 

Now for the love of God will somebody please stop talking and do this?

You can only pick the part if you know where it is, to do that you need to image the tray. If you image the tray, you should know the orientation.

Why not just use a down facing camera on the carriage? Image the parts and pickup one time.

[sokoloff]

Mike and KE5FX are talking about how to flip a part "right side up" that was dumped into a tray upside down.

[rx8pilot]

Mike and KE5FX are talking about how to flip a part "right side up" that was dumped into a tray upside down.

Vibration - A number of companies already do this although it is for massively expensive machines.

[rx8pilot]

This is insane.....

https://youtu.be/EFP0RqktaR4

[grimmjaw]

Interesting discussion.

I'm currently on this conundrum.I have need to assemble two boards with 92 unique part n 415 total parts on each board.
Tried assemble by hand, took me almost 8 hours (0603 part, QFN, some leadless package) with some issue.

Got a quote from PCBA companies , cost about 1000 USD(PCB+PCBA)..
Wondering if buying a low cost P&P , will actually be cheaper in long run (need 2-3 board to be assemble per month, different design)
3-4 hours setup time is okay comparing to 8 hours hand place the component.

[rx8pilot]

Interesting discussion.

I'm currently on this conundrum.I have need to assemble two boards with 92 unique part n 415 total parts on each board.
Tried assemble by hand, took me almost 8 hours (0603 part, QFN, some leadless package) with some issue.

Got a quote from PCBA companies , cost about 1000 USD(PCB+PCBA)..
Wondering if buying a low cost P&P , will actually be cheaper in long run (need 2-3 board to be assemble per month, different design)
3-4 hours setup time is okay comparing to 8 hours hand place the component.

This is very similar to what drove me to build a P&P system. I had a low volume of PCB's with a lot of unique parts and a lot of placements. Almost the exact same numbers. The high unique part count makes the small desktop machines not look so good. I gathered up my credit cards and chose a used machine that can hold 100+ unique parts. I set it up and was able to run a handful of PCBs and the variations any time I needed them. It was very scary at the time, but it has really turned out to be a great decision overall.

If I had not found the big machine, I may have tried getting a number of small machines to accomplish the same task. For anyone wanting P&P, just understand that the machine is only a part of the puzzle. After my machine was fully functional, it still took a a lot of effort and a lot of pieces and parts to make it worthwhile.

[KE5FX]

There will be some work in figuring out what materials to use for the trays, and the best way to "tap" it to flip parts over, but the vision is simply finding rectangles- how hard can it be..?ne should be able to find them easily.

1. Pick the part up from the tray
2. Move arm to well-lit area where it can safely...
3. ...drop the part.
4. Examine dropped part with camera
5. Part oriented correctly?
6. If not, pick it back up, go to step 3 and ...
7. Else pick part back up, remembering its rotation so you can...
8. ... place it correctly.

There, I saved you guys all that R&D time.    You're welcome. 

Now for the love of God will somebody please stop talking and do this?

You can only pick the part if you know where it is, to do that you need to image the tray. If you image the tray, you should know the orientation.

Why not just use a down facing camera on the carriage? Image the parts and pickup one time.

Because the "tray" looks like this:



The trouble is, everyone working on P&P machines is under the mistaken impression that their customers are trying to compete with FoxConn and Jabil.  This isn't the case, but the widespread perception to the contrary drives the market in a direction that's useless to me.  Take all day, I don't care, as long as I can do something else while it happens.

[girts]

Mike and KE5FX are talking about how to flip a part "right side up" that was dumped into a tray upside down.

Another way:
https://www.fritsch-smt.de/en/manual-pick-place/features/flipping-station/

Because the "tray" looks like this:

You need vacuum cleaner to pick up your electrolytic capacitors...

[Kjelt]

Tray picking has some other challenges as well: dust, hairs other contaminations.
The order of the trays for each setup is also costing time.

My idea (dream) about this is a two stage setup. One larger machine that stores all components dustproof in trays, and can then retrieve a single component and put it on a conveyor belt which moves the part to the p&p machine. A sort of automated smt storage wharehouse device, which ofcourse also can load parts into the correct tray. Nice project if only I had time.

[mikeselectricstuff]

Tray picking has some other challenges as well: dust, hairs other contaminations.
The order of the trays for each setup is also costing time.

I don't think this is a big issue. You would load the trays per job, not have a big bunch of permanent stock trays. You'd empty parts out of tape into the trays. You may want to use an intermediate scoop or something to minimise the chances of parts falling into the wrong tray, or bouncing out if one into another.
A decent graphical UI should make things quick to set up.
 

[Kjelt]

I don't think this is a big issue. You would load the trays per job, not have a big bunch of permanent stock trays. You'd empty parts out of tape into the trays. You may want to use an intermediate scoop or something to minimise the chances of parts falling into the wrong tray, or bouncing out if one into another.
A decent graphical UI should make things quick to set up. 

Agreed on that. Loading the trays ain't the biggest issue, it is unloading and putting them back into storage/stock that is the issue.
Look at the tray I use below, I now use one tray per pcb and keep them filled/stickered.
But what would be great is if all for instance 0603/0402 resistors are preloaded into such a tray and that you can store the entire tray back in a dustfree environment.
Now keeping/mounting all those trays (R0603/ R0402 / C0805 / C0603 etc. etc.) on the P&P machine would make it very big since all trays have to fit.
I am also searching to find the right combo between storage/stock and usage 

[Kjelt]

Thinking further if there was a seperate stock pick head this head could have a small acryllic fence.
This fence would only be open to the single "traycell"with the right component size and covering neighbouring traycells so components could not "jump" or "fall" into neighbouring cells.

[mikeselectricstuff]

I don't think this is a big issue. You would load the trays per job, not have a big bunch of permanent stock trays. You'd empty parts out of tape into the trays. You may want to use an intermediate scoop or something to minimise the chances of parts falling into the wrong tray, or bouncing out if one into another.
A decent graphical UI should make things quick to set up. 

Agreed on that. Loading the trays ain't the biggest issue, it is unloading and putting them back into storage/stock that is the issue.

For Rs and Cs, they are so cheap you can probably just dump the leftovers most of the time.
Alternatively you make the tray compartments removable so you can lift each one out and empty it into a bag or whatever. or use small lidded boxes as the trays, so you just put in the ones you need, and afterwards take them out and put the lids on.

Maybe something like these, but square.

https://www.ebay.co.uk/itm/20x-28mm-Plastic-Coin-Cases-Storage-Capsules-for-50p-Coins-UK-Seller/122434954404?epid=882898687&hash=item1c81b124a4:g:gfAAAOSwWxNYy8x7

Another option may be to have a machine function to pick all remaining parts from a tray and place them in  fixed locations ( storage/decanting tray)

[KL27x]

I have happened to notice that every one of the tiny SMD components I have ever used can be picked up with a magnet. That is one way to clean out a connected array of trays.

I'm sure you can come up with a better tool for your specific tray sizes, but the ghetto version is to just wrap a strong magnet in a plastic film, dip it in the tray to lift out the parts, and then unwrap the plastic film over a container which you can now pour back into a storage container.

[Kjelt]

If that is true an electromagnet would be ideal.   

[KE5FX]

Tray picking has some other challenges as well: dust, hairs other contaminations.

I don't see that as a problem at the 0402 level and up, because (once again, sigh) I'm not Foxconn, and no ISO9000 auditors are inspecting my prototypes and pilot builds.  Close the trays when not in use, or simply cover them with a tarp.

The order of the trays for each setup is also costing time.

The gantry should have enough reach to cover many dozens of trays.  Set them up once, refill them when necessary, otherwise forget them.

My biggest concern is that whatever picks up the parts had better be perfect, or it will occasionally drop them inadvertently.  If there's even a small possibility of dropping a 1M resistor into a neighboring compartment full of 100 ohm resistors, that would be annoying.  The whole compartment would have to be tossed out.

[jmelson]

My biggest concern is that whatever picks up the parts had better be perfect, or it will occasionally drop them inadvertently.  If there's even a small possibility of dropping a 1M resistor into a neighboring compartment full of 100 ohm resistors, that would be annoying.  The whole compartment would have to be tossed out.

An EXCELLENT point!  My P&P machine is pretty reliable, and tries to dump all parts with weak vacuum into the dump bin.  But, still, when I get done with a modest build run, there are a few parts strewn here and there on the machine.  With tape feeders, it is VERY unlikely any dropped parts would end up in the right spot to be picked up.  I can really see it happening with loose part trays.

Jon

[mikeselectricstuff]

Tray picking has some other challenges as well: dust, hairs other contaminations.

I don't see that as a problem at the 0402 level and up, because (once again, sigh) I'm not Foxconn, and no ISO9000 auditors are inspecting my prototypes and pilot builds.  Close the trays when not in use, or simply cover them with a tarp.

The order of the trays for each setup is also costing time.

The gantry should have enough reach to cover many dozens of trays.  Set them up once, refill them when necessary, otherwise forget them.

My biggest concern is that whatever picks up the parts had better be perfect, or it will occasionally drop them inadvertently.  If there's even a small possibility of dropping a 1M resistor into a neighboring compartment full of 100 ohm resistors, that would be annoying.  The whole compartment would have to be tossed out.

If you have vacuum sensor you could at least detect this and stop. However once it's picked, it's unlikely to drop, especially as it's already used vision to make sure it's picked centrally 

[rx8pilot]

So back to the Open Placer.....

I looked at the feeders and have some practical comments. My opinions are not random - they are based on actual experience of assembling boards manually and with P&P as well as designing and CNC machining my own cut tape holders over the past few years.

 It appears there are small dovetails that wedge the tapes in place. While this is a clever and low-cost design, it will have a critical flaw  - stability. Tapes are all thicknesses and have variations in width. This scheme will nearly guarantee that the tape will be very tight or too loose for many tapes. When the tape is too tight and the user is struggling with it, it is VERY easy to bump all the parts out of the other tapes. Any user will have to load a single lane while not disturbing the others and this is not easy after the cover tape is removed.

When a tape is even slightly loose, the pickup process makes the other parts jump out of their pockets. Hand placing a bunch of parts back into the tape is not productive.

The other major issue is that they are fixed width. The ones I see in the pictures have 10x 8mm lanes. This is great if you have 10 8mm tapes, but the idea quickly becomes a problem if you have 11 8mm tapes. You have to waste a lot of very valuable space with 9 empty lanes. In my experience - many of my boards have a bunch of 8mm (not in multiples of 10), some 12mm, some 16mm, and 1-2 24mm. The feeders really need to be individual lanes to accommodate a practical number of parts.

My first cut tape holders for manual assembly and P&P were fixed groups and the limitation was immediately obvious. I then started making adjustable cut tape trays that are adjusted in 4mm increments and had clamps to hold any tape rigidly. I rarely have parts jumping out and I can mix any width tape either 6" or 12" on the holder. This was a dramatic improvement in how I use my P&P because I can get 20+ unique parts on a single cut tape tray. I have trays for production boards and open ones for prototypes. My machine has room for two trays so I can swap 40 unique parts in a few minutes with any combination of parts. The clamps on top allow tapes to be swapped easily without disturbing the adjacent tapes.

Stability and flexibility of the feeders are a crucial element of success. I have spent a lot of time working this particular problem - cut tape holders for manual and P&P processes. It is not trivial.

On the Open Placer Indigogo - you can see one of the test videos. The system picks up a few parts, and then one jumps out of the tape. These are the things that force an operator to watch the machine and correct issues all the time.

The images show some small parts being placed but the consistency is predictably not so good. You can get away with some misalignment, but at some point, your passives will tombstone. The user either fixes this before or after the reflow - taking time, a steady hand, a microscope. The smaller the part, the more sensitive they are.

So - feeders are crucial. A poor feeder setup can send an otherwise capable machine to the bottom of a river. If you are thinking that you will simply push the GO! button, walk away, and come back to an assembled PCB - you need good feeders.


For the double sided PCB's - It is a fairly simple setup. For irregular shaped double-sided, it requires a bit more consideration. I designed special trays for double-sided single PCB's that are irregular shape. I add a photo of that setup as well.

[ar__systems]

I'm making a similar concept of strip feeder, but slightly different. You appear to be using 4mm walls separating your tapes. In order to maximize the space usage, I'm using only 1mm walls between the tapes. I originally intended to have it machined but the design was never ready for that, so in the meantime I made one entirely 3d printed. Some improvement is required, but it already works quite well. Due to the high tape density, I can place 30 of 8mm tapes on 1 jedec tray size feeder.

[rx8pilot]

The edges of the tapes are separated by 3mm to make room for the clamp screws.

The design has some rotten flaws. While I have sketched out some nice improvements, I cannot find the time to machine a new system.

I would have to start selling them if I put in much more effort. Can you post your design?

Short and misplld from my mobile......

[ar__systems]

I'll make a pic when I go to the office. I find I need to make the width of the lane slightly larger that 8mm, perhaps 8.15. Due to tapes being not exactly 8.00, it is difficult to slide the tapes into the lanes.