$1000 pick place

[mikeselectricstuff]

 you will still need a descent reflow oven and stencil printer.  Which aren't cheap.  Otherwise, you have a neat PnP bottlenecked by a toaster oven and messy stenciling station.

Oven and stencilling will rarely be a bottleneck, even with a moderately fast P&P unless the number of parts is low.
Reflow+cooldown time will be typically of the order of 4-6 minutes, so until your P&P time gets below maybe 8-10 mins per panel  you don't need to look at converyor oven - a second toaster oven perhaps.   
there is nothing inherently expensive in a decent stencil printer, though this is still a gap in the market - the Eurocircuits one is pretty good, but over-engineered and expensive.

[Corporate666]

Actually, the software side isn't the biggest headache and probably not the most time consuming bit in making a low-cost machine.  I suspect that given a couple of days, most of us could program a TM220A style machine, it is basically driving motors to a simple list of co-ordinates.

In my experience, the two challenges are 1. vision centering of a component and 2. accurate feedback of present head position.  The vision centering is complex and people often underestimate how important a good image is.  Accuracy of present position is usually taken care-of by building a very rigid structure, which ends up weighing alot and you get into this sort of performance death spiral.

As far as the vision goes, I rolled my own over several months, rather than using available libraries.  My approach to accuracy is to use light-weight parts and a novel realtime 2D triangulation system for the head. You get speed and instant out-of-the-box calibration, from a light-weight low cost machine.

I spent YEARS pursuing the cheap, easy, reliable PnP machine and what you said above exactly mirrors my experiences.  There really just isn't a cheap way to do it... speed means rigidity and that means expensive.  Inaccurate and PnP really don't go together, nor do slow and PnP.

The big problem with PnP's are that the commercial units have tons and tons of tricks to ramp up speed... but those tricks leads to complexity and reliability issues, and they get outdated and massively expensive to maintain.  I think there is probably a solid business to be had in retrofitting existing machines with aftermarket (simplified) controls.  You start with a really nice rigid structure with all the bells and whistles like ballscrews and optical encoders and the right sized motors... get rid of all the crap like air blow this and laser aligned that and retrofit some sort of standard NC control.  That's what we did with an old gantry style Celmacs machine.  It already had high pitch ballscrews, a rigid frame and all the parts we needed.  It had all the motors, position sensors, encoders, etc.  Just had an outdated proprietary control.  We ripped it out and retrofitted Mach3 and would write G-Code programs as PnP programs using custom M-codes for pick and place.  The feeders were controlled by the pick operation which made things easy, and it was very reliable.

Total cost was $1k for the machine, few hundred in parts, and it would do 1kPPH running at 50% speed.  Ran that sucker for years and sold it to a hobbyist when we got more serious.

Other than the above, the feeders are the crux of the issue.  But feeders are one of those things where if you have them and don't need them, they are essentially scrap metal.  If you need them and don't have them, they are invaluable.  I see them selling for $1,000/ea used, and other times I see them being scrapped out for scrap metal prices.  There must be a brand of feeder that has been the same for years and has been used on countless machines... maybe Assembleon,  Juki,  Mydata,  Siemens, Universal, Fuji... someone must have lots of machines out there that used the same feeders that are highly available, cheap, and easy to interface with.  Seems that would be a great starting place for a home brew PnP?

[fcb]

Lets take the QSP2 from Quad.  Weighs something in the region of 2 tons. Moves a part not weighing more than 1g perhaps 12 inches. Does it with a precision of perhaps 0.03mm.

Take a CNC machining centre perhaps a Hurco VM10, weighs 2.7 tons. Moves a 250kg bed around at 6M/min whilst making chips at 15HP, to an accuracy in-the-region of 0.03mm - substantially better if you slow it down.

Why do many PandP machines weigh millions of times more than the part they move, common answer is rigidity. And that's true, but at huge expense.. Big head means bigger carriage means even bigger gantry means huge motors, etc....  Story doesn't stop there though - you need lots of feeders, these are heavy and take up alot of space, so do nothing to decrease the mass of the machine.

[mikeselectricstuff]

I think the key to real progress in cheap P&P would be to replace mechanical grunt with clever sensing - if you have an accurate way of measuring head position directly at the head. then you can use closed-loop control  and greatly reduce the need for high rigidity and weight, as you'd only need to move a very lightweight lift/rotate head, and home in on the exact position when you're almost there.
There would also be scope for self-learning of the dynamics to get closer to the desired position each time by looking at previous feedback data, reducing time spent on the final fine-tune. 
I think something could be done using laser time of flight and/or interferometer type sensing.
It  actually only needs to be repeatable, as absolute accuracy/linearity could be calibrated using the fid camera and a calibration sheet.
Maybe a ring of retroreflective tape on the nozzle, just above where the part sits, and two steerable mirrors pointing at it (or steerable complete laser assemblies) - these would not need to be accurate or move fast, as they will know roughly where to look for the nozzle, and if the measurement reference point is at the centre of rotation, the rotation angle won't affect accuracy.

Feeders get a lot easier if you use the head to advance - I think this is a reasonable compromise. The TM220 almost gets this right, but fails by not having flexibility in lane widths - that can't be too hard to do better with movable partitions/guides. Feeding the waste tape under the work area is a neat way to simplify the mechanism.
And picking loose parts from trays or short strips shouldn't be too hard with decent vision.


 

[Eclipze]

Closed loop vision feedback would help quite a lot.  Rigidity is still pretty important though if going for reasonable speed.  The TM240A head is pretty small and looks light, but I still cannot run it feed speed on what I considered a very study and heavy table. 

The TM220A and TM240A have fixed feeders, however it's not a far stretch to replace that block with an alternative bank.  That mounting block takes 6 bolts.  If Neoden had an alternative, it would take much to load a different block before setting up for a run.  If anything, I believe a practical and accurate approach would be to divide up this tape block into 4 or 6 sections.  So you can unbolt and change over some of them to suit the loading job, without having to unload some of those common parts.  This could be a cheap and position accurate reliable approach.

I still think the software side is more complex than a lot of people first consider.  The movements are just a small part.  There is a database requirement for all the parts, banks, pickup positions, tape index lengths etc... Managing failed pickups, starting or re-starting at different parts, fidicuals and board rotation correction, panelisation with placement, nozzle changing, teach modes, file imports... and all the vision libraries for various parts.  It's a lot of software.  I used Mach3, but it was a rather fixed format for basic operation... and was very easy at the basic function level.

For tape feeding, indexing using the head is such a practical approach for an entry level machine.  The mechanical overhead and cost is so small.  Which also translates to lower complexity and less that can go wrong.  The tradeoff is lower placement speed, but at the practical speeds for an entry level machine, it's not a significant factor.  The friction drive on the tape spooling used on the TM220A/TM240A has worked really quite well.

[Corporate666]

I think one of the issues at play is that it's very easy to run into a level of complexity where it's just much easier and cheaper to buy something off the shelf.  That's where I ended up.  I have three Quad 4C's at the office.  Each of them is still supported by a local company who bought all the parts/tooling right from Tyco and still offers factory warranty support... but the machines are simple.  I paid around $7k for each machine and they do around 1,800 PPH and came with maybe 20 feeders each.  They are dead simple to set up and run and have up and down looking cameras and vision, handle vibratory feeders, waffle trays, etc.

Before that I had a Fuji that I got for cheap money because it just went out of support with Fuji... an IP2 or something?  The quality of the Fuji was phenomenal - but it was so complex with sensors on everything, needlessly complex robotic loader arms to shuttle waffle trays from one area to another, etc, etc.  A real pain to keep it running reliably.  Before that I had a pair of Dynapert Chipshooters... 16,000CPH.  Like gatling guns... Dynapert was a local company and I got them cheap and managed to get one running reliably... got the original invoices too - over $400k each.  But soooo complex and always needing tweaked, cleaned, lubricated, adjusted, fixed. 

The retrofitted Celmacs was great because we built it from start to finish so we knew it's innards completely and when something went wrong, we knew what to do.  But other than that, nothing really beats the Quads... super simple, and it's nice being able to call the manufacturer if I need do (haven't had to yet).  I really think it's a very hard business case to make doing low volume PnP for the small guy... just really hard to make the numbers work.  It seems like it should be so easy, but it's just really hard to really do it.  Accuracy and feeders are hard nuts to crack.

[fcb]

Yes, but can the 4C be carried upstairs by one person and shoved in the boot (trunk) of a car.

[Eclipze]

I know in the past I've had issues with having floor space for such equipment.  For many hobbists, being able to run a compressor to run a commercial pick and place machine would be a deal breaker alone without getting complaints from the neighbors hahaa...  But there can also be other limitations, such as needing 3-Phase or high current wiring.  Even for my reflow oven, I had to have a 32amp circuit all the way from the meter box and 25A socket and plug installed/fitted to run it.  Another possible trap for people would be to get a pick and place, then realise just how much fumes they'll get from reflowing the boards.  If you can't reflow the volume of boards, there isn't much need for a pick and place.  I used to have a ducted fan setup to exhaust the fumes outside, but still had the roller door up to keep the workshop clear enough to occupy.  Have a filtration setup now, so at least the neighbors don't get suspicious and wonder what I'm cooking back there hahaa...

[Corporate666]

Yes, but can the 4C be carried upstairs by one person and shoved in the boot (trunk) of a car.

Yes, his name is Mr. Hulk


No, but... how often do you need to do that?  I realize a hobbyist might want to put a PnP in his attic but if someone is doing such low volume that they want to put their PnP in the boot of their car and carry it up stairs, why not just put the boards together by hand and/or outsource the boards?  I think that the rigidity required for sufficient accuracy for PnP is going to be mutually exclusive with the lightness required for hand carry which will also be exclusive with the sort of reliability that PnP requires for accurate placement.  Maybe it's possible with a lot of electronic sensing and processing like Mike describes above, and maybe that is future of the hobbyist PnP machine... but I have never seen anything close to a successful homebrew hobby PnP machine.  I've seen tons of attempts with aluminum extrusions, or frames like 3D printers or the like, but they just never work - and that's exclusive of the feeder problem, which is a whole other can of worms.

[fcb]

I started out with a Versatronics RV, Mike has similar machine - the actual machine is certainly 'carryable' upstairs or through door ways.

Any affordable p&p needs to run off single phase and be easily transported/moved (no pallet-truck/tail-lift, etc..) and fit through an standard doorway.

As for compressor etc... I don't believe this needs to be built in, nice if it is, but not critical - I have a little silent compressor under my p&p bench for my paste dispenser and other services, it was <£300 new.

[mikeselectricstuff]

I started out with a Versatronics RV, Mike has similar machine - the actual machine is certainly 'carryable' upstairs or through door ways.

It seems odd that after all these  years (?20?) the RV is still probably the best value, smallest viable used P&P available (in UK at least).
The only thing I've seen that looks comparable is the Mechatronika M10V - not sure what this costs though. 

[Corporate666]

I started out with a Versatronics RV, Mike has similar machine - the actual machine is certainly 'carryable' upstairs or through door ways.

Any affordable p&p needs to run off single phase and be easily transported/moved (no pallet-truck/tail-lift, etc..) and fit through an standard doorway.

As for compressor etc... I don't believe this needs to be built in, nice if it is, but not critical - I have a little silent compressor under my p&p bench for my paste dispenser and other services, it was <£300 new.

I saw Mike's video of that machine - that's not what I would call carryable... it's just about the same size as a Quad 4C.  You mentioned carrying up a set of stairs or going in a car boot... that Versatronics is like the Quad - you would put it in place and leave it there long term.  I'm not even sure you would want that on the second floor of a wood-floored house, would you?  What does it weigh?  Maybe it's all aluminum?  It looks like a heavy beast.  The Quad 4C is single phase too, but it does require shop air (or an external vacuum pump). 

[Eclipze]

All things considered, perhaps what people want isn't necessarily yet achievable in terms of size and cost.  There is probably a good reason so many of these projects are started, but never finished.  Yet the cheapest machine coming out of China is still much more expensive than many expect.  Even given it doesn't have vision and a few other items on the wish list.  I can tell you now, even unpacking the TM240A out of the wooden crate, it would definitely be a two man job to lift or move it anywhere. 

It is disconcerting when people judge the cost of such equipment to be much lower than the actual price, with no reference for comparison.  It may look like it costs $2k, but you have to consider the entire business case.  The cost of development has to be covered, including wages and all the prototypes.  Tooling costs.  All the software development.  Then there is production, everything from stock management and procurement through to logistics, advertising and even having the warehouse floor space to assemble/build and test them.  Everyone in this process is doing so to earn a living as well, and this profit on top is the mechanism that ultimately puts such a product on the market.  Raw cost of parts is merely one component to the total cost.  It is getting easier to mis-judge cost in a market where toaster ovens are manufactured in the millions and you can pick one up in a competitive market for $30.  Yet the cheapest IR reflow oven direct from China on ebay will set you back $308.  ~10 times more expensive.  Getting to something reasonable, like the T200C+ direct from China and your up at $2.8k not including delivery and the need for a 25A circuit installed.  I'm sure a lot of people wouldn't understand why it costs more than $1000.  Having recently investigated, a descent small batch oven (not a China model) is likely in the range of $4.2k to $7k+, not including delivery.

I probably fit more so in the category for an early adopter of technology, and pay the premium for the privilege.  At the same time, if I found a Quad in good condition when I searched and going for a good price I likely would have jumped on it.  Looking at old second hand machines, finding them in good condition and capable of 0402 placement was too ambitious.  At least in Australia.  I'II stick to prototype and small batch runs, and save the real board loading bulk for the professionals with the professional expensive bling machines :-)

[mikeselectricstuff]

I started out with a Versatronics RV, Mike has similar machine - the actual machine is certainly 'carryable' upstairs or through door ways.

Any affordable p&p needs to run off single phase and be easily transported/moved (no pallet-truck/tail-lift, etc..) and fit through an standard doorway.

As for compressor etc... I don't believe this needs to be built in, nice if it is, but not critical - I have a little silent compressor under my p&p bench for my paste dispenser and other services, it was <£300 new.

I saw Mike's video of that machine - that's not what I would call carryable... it's just about the same size as a Quad 4C.  You mentioned carrying up a set of stairs or going in a car boot... that Versatronics is like the Quad - you would put it in place and leave it there long term.  I'm not even sure you would want that on the second floor of a wood-floored house, would you?  What does it weigh?  Maybe it's all aluminum?  It looks like a heavy beast.  The Quad 4C is single phase too, but it does require shop air (or an external vacuum pump).

The RV base+arm unit is heavy but just about carryable - it sits in a steel frame which is just square tube,with a wooden(!) base-board , so not especially heavy. As it uses a SCARA type geometry, the motion transmitted to base/floor is mostly torque, so it doesn't  need  to be as heavy as for an X/Y gantry.
Even then, the arm construction is quite a lot heavier than it really needs to be, as the hardware was originally designed for light machining. The head mechanism could also be made a lot lighter - the rotation stepper is much bigger than it needs to be.

[Corporate666]

All things considered, perhaps what people want isn't necessarily yet achievable in terms of size and cost.  There is probably a good reason so many of these projects are started, but never finished.  Yet the cheapest machine coming out of China is still much more expensive than many expect.  Even given it doesn't have vision and a few other items on the wish list.  I can tell you now, even unpacking the TM240A out of the wooden crate, it would definitely be a two man job to lift or move it anywhere. 

I was unfamiliar with the TM240A machine, but a quick Google search turned up some Youtube videos... I was pretty amazed... at how crap it was (apologies to any owners).  The guy said he paid $5500 for it delivered.  I paid $6k for the latest Quad 4C I bought which was totally refurbished 3 years ago by the manufacturer and came with the latest software, LCD screen, 20+ feeders, the newest "deep field" up and downward cameras, etc.  That was about market rate for the machine.

I will run 500-1000 boards through it every couple of weeks and in the past 2 years, nothing has gone wrong with it at all, other than having to clean solder paste out of the nozzles twice and having to swap out a couple of feeders that got clogged up with tape detritus, but that's tens of thousands of placements with no issues - and around 1,500CPH realistic, and it rarely misses a placement and will do 0402's all day with no issues. 

Here's a 4C in action



It also has a nozzle changer, supports vibratory feeders, waffle trays, can do part inventories, pattern repeats, component fiducials, board fiducials, allows for multiple feeders with the same component and will automatically jump to the next feeder when one runs out - and has SMEMA input and output on motorized transport rails for easy board load in and load out.  Plus I can call a 1-800 number and get support, and even buy an extended warranty.

I'm not disagreeing with you at all - I am totally agreeing with you 100% that people don't take the costs into account.  This guy who spent $5500 on that Chinese hunk of junk wasted his money unless nothing else was available.  In-house board assembly is all about cost... older machines that either require more maintenance or are slower don't have much value in a contract assembly house.  My Quad 4C's are good for high mix low volume.  They are not the fastest machines around but you can load them up with tons of feeders (on all four sides, I think you can fit 130+ feeders on the machine) and it has tooling holes on 1" centers all over the surface of the whole area for any sort of jigs and fixtures you can dream up - so it's very flexible.  But for a contract assembler - it will never be competitive with something like a Fuji CP6.  So you just have to keep your eyes open for something like the Quads - they are fully depreciated and often well maintained and can be bought cheap.  It's false economy trying to save by getting a $2k PnP then spending hundreds of hours screwing with it.

[Corporate666]

I started out with a Versatronics RV, Mike has similar machine - the actual machine is certainly 'carryable' upstairs or through door ways.

Any affordable p&p needs to run off single phase and be easily transported/moved (no pallet-truck/tail-lift, etc..) and fit through an standard doorway.

As for compressor etc... I don't believe this needs to be built in, nice if it is, but not critical - I have a little silent compressor under my p&p bench for my paste dispenser and other services, it was <£300 new.

I saw Mike's video of that machine - that's not what I would call carryable... it's just about the same size as a Quad 4C.  You mentioned carrying up a set of stairs or going in a car boot... that Versatronics is like the Quad - you would put it in place and leave it there long term.  I'm not even sure you would want that on the second floor of a wood-floored house, would you?  What does it weigh?  Maybe it's all aluminum?  It looks like a heavy beast.  The Quad 4C is single phase too, but it does require shop air (or an external vacuum pump).

The RV base+arm unit is heavy but just about carryable - it sits in a steel frame which is just square tube,with a wooden(!) base-board , so not especially heavy. As it uses a SCARA type geometry, the motion transmitted to base/floor is mostly torque, so it doesn't  need  to be as heavy as for an X/Y gantry.
Even then, the arm construction is quite a lot heavier than it really needs to be, as the hardware was originally designed for light machining. The head mechanism could also be made a lot lighter - the rotation stepper is much bigger than it needs to be.

Ahhh, that's different than the one I have seen then... the one I saw looked virtually the same size/shape as a Quad 4D - about 4 feet square and maybe 5 feet tall, box steel frame with a metal plate about 3.5 feet up and a cylindrical arm in the center of that.  There must be different models then....the one in the pics above looks substantially smaller.

[Eclipze]

I paid $6k for the latest Quad 4C I bought which was totally refurbished 3 years ago by the manufacturer and came with the latest software, LCD screen, 20+ feeders, the newest "deep field" up and downward cameras, etc.

WOW!!!  That's ridiculous!!!  I remember talking to several dealers at the Electronex Expo, as well as those that followed up with phone calls and emails.  I literally had one guy give a decent belly laugh at me when I said i was looking for a 2nd hand machine under $10k.  As an example, there was a 2008 Autotronik BS-384 with a great range of feeders that had my interest.  Until the price range was disclosed $30-$35k.  I'd love to hear from other Aussies about this sort of thing, as often I wonder if the respective climate here is just massively different to the UK/USA.

[mikeselectricstuff]

Ahhh, that's different than the one I have seen then... the one I saw looked virtually the same size/shape as a Quad 4D - about 4 feet square and maybe 5 feet tall, box steel frame with a metal plate about 3.5 feet up and a cylindrical arm in the center of that.  There must be different models then....the one in the pics above looks substantially smaller.

The RV is basically this unit sitting in  a frame with feeders either side. And I do mean just "sitting". It doesn;t rely on the frame for any more than support and to hold the covers, the latter being for protection from dust & mechnaical hazard ( it can give you a nasty whack  ), and to help control light levels - the vision likes to have consistent ambient light levels.
You could use it without the frame - you'd just need the base board which supports the feeders.

[matkar]

The only thing I've seen that looks comparable is the Mechatronika M10V - not sure what this costs though.

According to http://www.elektroda.pl/rtvforum/topic2318954.html more than 19000EUR...

[mikeselectricstuff]

If it includes feeders, 19KEU isn't that bad if it works well

[nctnico]

It looks nice but if I do the math outsourcing is still cheaper.

[Eclipze]

Make sure your math includes the overheads associated with using an external party to do the board loading.  Small board run done yourself.  Much cheaper unframed stencil, no paperwork, quoting, chit chat, building reference samples, validation sample, handling build varients, handling of loose/tape strips, no expensive chip loading component losses etc...  For a descent size job, spending 2-3 days to establish the production run with an external supplier makes total sense.  But everytime you have a small run job, spending 20 mins to setup the machine and loading the build as you need within a day is a big savings.... both in time and outgoing expenses. 

[mazurov]

I also have TM240A, it is a nice machine. It is discussed quite extensively here -> http://dangerousprototypes.com/forum/viewtopic.php?f=68&t=4903&sid=8d99a69d51b41610859d3a2be0bd57de .

[Harvs]

Can I ask where you got it from/through?  Last time I looked at threads on it there was some considerable difficulty acquiring one if you weren't versed in Taobao.

[Eclipze]

I looked at Taobao... wasn't worth the risk and trouble to save a couple hundred.  Deal direct with the manufacturer... http://www.neodentech.com/
Mia quite good and quick to deal with.  Had stock on the shelf and super quick to dispatch.  Just what you want if sending funds overseas.
Mia@neodentech.com

I would recommend to ask about the "Export Necessary Accessories", which is an extra pack of various parts.  May come in useful.