Better Pick And Place Machine DIY

[mikeselectricstuff]

WRT to feeders and diy pp machines. I've thought about it too and personally I think the whole feeder caper is insurmountable for the DIY'ers. I think the best idea if someone is intent on doing a machine is simply to limit themselves to cut tapes and stick vib feeders. So build a really large xy table with a really fast screw and get a crap load of 8mm and 12mm cut tape linear guides machined out of aluminium and mount them down at one end of the table. Doing it that way they may get a couple hundred parts per strip of cut tape laying down on the table which may be good for a run of 100 or so boards before you need to replace the tapes. pulling back the tape covers will be up to the operator to do before running the job. ie no feeders, no fishing sinkers, limited runs of about 100 or so boards.

I strongly disagree that a good DIY feeder is not possible, it just needs some clever design. It's not an inherently difficult thing to do, just needs thought applied to design details.
If you go multi-lane, that helps remove the size constraint as well as making it easier to accommodate different tape widths - standard feeder bank width, different numbers of lanes with just a differemt set of guide rails.

Another factor in keeping the widest possible potential market is machine footprint. A low-end machine is going to spend a lot of time not being used, so if it takes up a lot of space, it is going to be less useful to people with limited space.
Large banks of passive tape rails are not conducive to a small footprint.
Another issue here is you need high mechanical accuracy in the placement area but not in the feeder/pick area, so by needing to enlarge the table  size just to accommodatae feeders, you are paying for precision over a wider area than is strictly necessary in terms of mechanical precision. 
This is one reason I like the  SCARA arm approach as it has a small footprint & makes efficient use of space. Although it has an irregular reach area, this can be efficiently used by placing feeders, toolholders and reject bins in the outlying odd-shaped areas of accessability.

[Dago]

I think this is extremely clever.

No "active" feeders, the PP head itself moves the tape.

[Leo Bodnar]

I think this is extremely clever.

No "active" feeders, the PP head itself moves the tape.

Tonnes of code and not even a motion profile with controlled acceleration.
It shakes like it is having a heart attack.
Leo

[mikeselectricstuff]

I think this is extremely clever.

No "active" feeders, the PP head itself moves the tape.

Using the head to move the tapes is a reasonable approach, but not using the tool - should be done with a dedicated pin. However cover tape peeling is still a problem - with static tension it can be a fine line between enough friction to prevent the cover-peel moving the tape, and too much for the head to pull smoothly. 
This machine also has no vision - vision is essential for useable accuracy - you may just about get away without for 0805s but nothing smaller.
Vision is so cheap nowadays that there's no excuse for not using it. What would be nice is of someone figured out a way to use optical feedback for closed loop positioning. Unfortunately you can;t do it by looking at the PCB as you only see solder paste, and you need to align to the pads underneath. I sort of wonder if there may be something clever you could do with an upward facing camera looking at a cover plate with an accurate grid on it.

[EEVblog]

Yes but encouraging them to think about how to make something that is actually useful and not just diving in with a half-arsed approach that will never be more than a toy. My fear is that some over-enthuisiastic DIYer will lauch a half-arsed attempt on kickstarter, get funded by fanboyz/girlz who don't understand the issues until it's too late.

I suspect that's inevitable.

Is there any value is a good quality manual placement machine perhaps?
i.e. one that has either say vibration trays (to flip a random lot of components) or manual tape channels, and vision assistance with good micro adjustment.
Like this one at 9:40


Dave.

[mikeselectricstuff]

Yes but encouraging them to think about how to make something that is actually useful and not just diving in with a half-arsed approach that will never be more than a toy. My fear is that some over-enthuisiastic DIYer will lauch a half-arsed attempt on kickstarter, get funded by fanboyz/girlz who don't understand the issues until it's too late.

I suspect that's inevitable.

Is there any value is a good quality manual placement machine perhaps?
i.e. one that has either say vibration trays (to flip a random lot of components) or manual tape channels, and vision assistance with good micro adjustment.
Like this one at 9:40


Dave.

Probably, but for $8000 it's very limited to commercial outfits that do a lot of hand placement of very low-run boards with very small parts.
Just saw this, which is a similar idea, rather cheaper :
http://dangerousprototypes.com/2012/11/24/diy-low-budget-manual-pick-place/

I don't think either of these would be much faster than tweezer/vac-pen hand placement, the main benefit is precision, so only really useful where you need to use small stuff, and a different beast to any sort of mechnised P&P, where the aim is higher speed/less manual work and fewer errors.

[george graves]

Here one idea on how to center the parts and not even bother with vision if you choose to. (this is a mill converted to a PNP - so it's extremely slow - but you get the basic idea)



I still don't see any problems that can't be solved.  There are several example of good hardware (speed and precision) in this thread.  And example of software that would work as a starting point.

As far as be concerned about a kick starter project...well yea, it's kick starter.  One has to look at all their projects with a raised brow(or should I say there is a sucker is born every minute).  So that's kinda moot.  It's not a reason why an open-source approach wouldn't work.

And here yet another design - I like the idea of it being on some sort of cart that could be put away when not in use.  too bad it doesn't have vision, or better feeders.  Someone just needs to combine all the ideas into one machine.  Problem solved.

[george graves]

Top and bottom vision on a DIY pick and place:

[pickle9000]

A novice comment again. Does Kicad or any other program export info (component location) that a pick and place system can use? I assume the answer is no.

As a follow up I guess you would need a pickup location and perhaps a mask to indicate orientation, if not a mask a package name so you could create one. Just thinking.

[Thor-Arne]

A novice comment again. Does Kicad or any other program export info (component location) that a pick and place system can use? I assume the answer is no.

As a follow up I guess you would need a pickup location and perhaps a mask to indicate orientation, if not a mask a package name so you could create one. Just thinking.

Yes, they can create .pos files.

Generated from the video card sample that comes with KiCad:

Code: [Select]

### Module positions - created on 25.11.2012 03:30:58 ###
### Printed by Pcbnew version pcbnew (2012-01-19 BZR 3256)-stable
## Unit = inches, Angle = deg.
## Side : Cuivre
# Ref    Val                  PosX       PosY        Rot     Side
C2       100nF               13.6050     2.6550      90.0    Cuivre
C4       100nF               14.0150     3.0750       0.0    Cuivre
C5       100nF               13.6300     3.8300     270.0    Cuivre
C6       100nF               13.6050     2.8850      90.0    Cuivre
C7       100nF               13.3150     3.4550     180.0    Cuivre

[gregariz]

WRT to feeders and diy pp machines. I've thought about it too and personally I think the whole feeder caper is insurmountable for the DIY'ers. I think the best idea if someone is intent on doing a machine is simply to limit themselves to cut tapes and stick vib feeders. So build a really large xy table with a really fast screw and get a crap load of 8mm and 12mm cut tape linear guides machined out of aluminium and mount them down at one end of the table. Doing it that way they may get a couple hundred parts per strip of cut tape laying down on the table which may be good for a run of 100 or so boards before you need to replace the tapes. pulling back the tape covers will be up to the operator to do before running the job. ie no feeders, no fishing sinkers, limited runs of about 100 or so boards.

I strongly disagree that a good DIY feeder is not possible, it just needs some clever design.

Its possible.. some commercial feeders are very simple as I mentioned...you could just copy an existing feeder design... its just that I'm not sure its worth it. Feeders for some of the older common machines are available for around $100 per feeder. I get the impression people want something in their shed for a grand or two... I'm not sure you'll get near that point with a bank of 20 or 40 feeders included ( you will with cut tape strips though). (not to mention weight. I have a small cnc. My feeder bank weighs more than the cnc). Then you're in the realm of picking up a commercial machine for 5 grand at auction with feeders included....which way to jump?

You could look at the Madell approach which seem to make machines that look somewhat like what a DIYer could be expected to come up with. Instead of making their own at first they decided just to steal a commercial feeder and use that (can't remember which... fuji feeders?). Then when they did come up with their own you needed to run down to the tackle shop and buy a bunch of sinkers. Its probably just not an economical prospect.

[benemorius]

...
Then you're in the realm of picking up a commercial machine for 5 grand at auction with feeders included....which way to jump?
...

I didn't realize you could pick one up in that range. Can you elaborate a bit?

[gregariz]

...
Then you're in the realm of picking up a commercial machine for 5 grand at auction with feeders included....which way to jump?
...

I didn't realize you could pick one up in that range. Can you elaborate a bit?

just keep your eyes on the auction sites.

I think whats driving it... at least from what I see is that this last 2 years has been a real killer for the industry. This year alone I've seen a whole bunch of shops shut down, not just OEM's but some of the smaller contract shops have been going to the wall. As a result alot of used gear has been going up, not just pp's but all sorts of stuff. Downsides are most of it literally weighs a ton as it is commercial gear. You hardly ever see the small proto type systems. The other end of it is that most of the gear will only do down to 603's in a pinch.

[JonnyBoats]

Here is a very low cost DIY manual pick and place that looks quite nice: http://vpapanik.blogspot.gr/2012/11/low-budget-manual-pick-place.html

[EEVblog]

The other end of it is that most of the gear will only do down to 603's in a pinch.

That's no drama for most projects. If you know your machine only does 0603, design accordingly.
If you could get a commercial machine for $5K here in Oz, I'd be temped to pick one up myself, for kicks.

Dave.

[george graves]

Being that the low end of a *new* usable machine is ~30k USD - the lowest I've seen for a used machine is 10k - and always sold "AS IS" - as auction houses and ebay sellers tend to do.  No gaurentee that it works at all  So you are out 10k and you don't know if the machine works or not - and if it's 20 years old, good luck on getting parts.  I don't know about you - but I could never wright a check for 10k or even 5k with out knowing something works.....

I have head of a hacker space that picked up a non-working unit for something like 3k(?) and then replaced all the electronics.  I don't think they have vision or anything working on it - it's all just trusting the steppers don't skip and then hand re-aligning the parts that don't get placed right.

[mikeselectricstuff]

A novice comment again. Does Kicad or any other program export info (component location) that a pick and place system can use? I assume the answer is no.

As a follow up I guess you would need a pickup location and perhaps a mask to indicate orientation, if not a mask a package name so you could create one. Just thinking.

Any serious PCB software will generate a report listing component locations, rotations, types & values.
If you can unify the package names used by your PCB and P&P system (and make sure orientations are consistent relative to feeder orientation!) then data setup can be very quick & easy. A minor fly in the solderpaste is SO packages that can come in tubes or on tape, which usuallly need different orientations.
You also need to export fiducial positions to get the absolute position references - I do this by having a special 'fid' component that my conversion software recognises and uses to create reference points in the P&P config file.

[mikeselectricstuff]

Being that the low end of a *new* usable machine is ~30k USD - the lowest I've seen for a used machine is 10k - and always sold "AS IS" - as auction houses and ebay sellers tend to do.  No gaurentee that it works at all  So you are out 10k and you don't know if the machine works or not - and if it's 20 years old, good luck on getting parts.  I don't know about you - but I could never wright a check for 10k or even 5k with out knowing something works.....

I have head of a hacker space that picked up a non-working unit for something like 3k(?) and then replaced all the electronics.  I don't think they have vision or anything working on it - it's all just trusting the steppers don't skip and then hand re-aligning the parts that don't get placed right.

This is where you really need to know what you're doing- some machines will be easily fixable, others will need unobtanium bits, like custom made leadscrews.

 If you're seriously  looking it's worth talking to specialised second-hand dealers - the good ones will know which machine are viable, even as 'projects'.

You generally also need a fair amount of space, and access (wide doors, no steps etc.) as most old machines are big and heavy and don't break down into smaller pieces easily.
When I was looking for a P&P  I found several in the £5-10K range with feeders that would have been fine if only I'd had the space for them.

[mikeselectricstuff]

Here one idea on how to center the parts and not even bother with vision if you choose to. (this is a mill converted to a PNP - so it's extremely slow - but you get the basic idea)

Mechanical centering is a dead end IMO. A few low-end commercial machines of old used it - the only reason I know is whenever I hear them mentioned it is to bitch about how bad they are.
It may work for chip components but not readily scalable for the whole range of  pars without a lot of fiddling.
The machine already has to have the ability to rotate and move parts, so all you need is a camera, and cameras (and the processing needed) are now trivially cheap so no reason to slow everything down by stopping to centre each part.

[mikeselectricstuff]

The other end of it is that most of the gear will only do down to 603's in a pinch.

That's no drama for most projects. If you know your machine only does 0603, design accordingly.

0603 is a pretty reasonable limit for a low-end machine - the difference between 0805 to 0603 is a lot more significant that 0603 downwards. something that could only do 0805 would be quite limiting - e.g. by the time you 'd got a couple of 0805  decoupling caps around a QFP micro, you've lost a lot of routing space.
0603 is now pretty much the default industry size for non space-constrained designs.

If you could get a commercial machine for $5K here in Oz, I'd be temped to pick one up myself, for kicks.

Dave.

If there are any SMD/factory equipment dealers left in Oz it's worth  emailing them - they often aren't too interested in low-end stuff but if they know someone is looking they may come across something & let you know instead of letting the factory they're clearing scrap it.
The problem is that with any old machine that is still useable, the feeders tend to be in demand from other users, and can often have higher value than the  machine.
Industrial SMD stuff is an area where there is very little info on the net - the only forum I'm aware of (other than mine ) is http://www.smtnet.com/Forums/

[JuKu]

..and if someone wants a suggestion on something to design to make a real contribution to low-end SMD, the market is crying out for a cheap (<£200)  stencil printer that will handle random sized unframed metal stencils, with proper tensioning.

Could you elaborate? In principle, what is wrong with manual method, like this? I've done only two boards like that, using a steel stencil, and the print turned out perfect. Was I just lucky? Why a machine is needed, why tensioning is needed?

[Psi]

That video has no sound (To save others from wasting time checking their PC cables )

[JuKu]

Industrial SMD stuff is an area where there is very little info on the net - the only forum I'm aware of (other than mine ) is http://www.smtnet.com/Forums/

Also: forums.adafruit.com/viewforum.php?f=42

[peter.mitchell]

CAUTION; STRONG USE OF CHINESE LANGUAGE!

http://item.taobao.com/item.htm?id=19322348120

Looks amazing though. looks to be about $3500 AUD @hoeken linked it on twitter a while ago.

http://www.tudou.com/home/SMThz
Quicker loading videos.

I'm also zipping the vids and putting them in my dropbox; it's quite impressive, the later videos show it can pick up 2 components at the same time and place them, halving the head movement mostly.

[mikeselectricstuff]

CAUTION; STRONG USE OF CHINESE LANGUAGE!

http://item.taobao.com/item.htm?id=19322348120

Looks amazing though. looks to be about $3500 AUD @hoeken linked it on twitter a while ago.

http://www.tudou.com/home/SMThz
Quicker loading videos.

Very interesting - could be a great opportunity for an English-speaking company to translate & market it.

Not clear what if any vision it has though, or how it handles different tape widths.
Putting two heads on a low-end machine is awesome!