Very low cost Pick and Place kit

[mrpackethead]

Besides me personally am completely fed up with the capatalist pressure to keep on expanding your business till you have tens of machines and tens of personell, and you are only doing the payslip signing and talking to the customers and your staff is doing the fun things.

Having good machinery, that does the job efficently and quickly, has meant i've not had to have tens of stuff,  you just need a couple of people putting stuff in, and taking it out.  When it breaks you can fix high end machines as well yourself.        I guess we are all on a continium of experience.  I started out with a home built PNP... and well, hard expereince, i have old but good commerical machines now, and it was absolutely the right decision for me.   

There is nothing wrong with a small scale one or two person run business, have fun , this is your life, live it now and enjoy instead of waiting till retirement.

If you want to spend your time feeding and watering PNP machines then cool.  I want to spend my time to doing fun things.. Building boards is not a fun thing. Its thing that means i can do fun thngs

[ottoragam]

I'm pretty sure you understand the advantage that having several tiers of the same product provides. These kind of machines are not meant to be used to handle big and complex assembly jobs, but there are many business opportunities where electronics assembly is required where a machine like this is all that's needed or can be justified. And many times outsourcing PCBA is not a good choice either.

I also don't find building boards too thrilling, that's why this PnP came to life.

[mrpackethead]

I'm pretty sure you understand the advantage that having several tiers of the same product provides. These kind of machines are not meant to be used to handle big and complex assembly jobs, but there are many business opportunities where electronics assembly is required where a machine like this is all that's needed or can be justified. And many times outsourcing PCBA is not a good choice either.

I also don't find building boards too thrilling, that's why this PnP came to life.

Yes, i understnad differnet teirs. I dont' have a line that can place 200kpph, likei Samsung does in Malaysia. I have a line that can do about 12kpph ( real numbers )       But I've done also done micro pnp , in several guises much like your machine.  I would argue that there may well be a negetive efficency with 'entry level pnp'.   It will cost more to use than well organised manual placement.      For some jobs / situations that may be acceptable.   For someone making 50 boards per month,  maybe not.    But there is a LOT of good in building and learning to use one of these. You'll learn a lot.  and any learning is awesome.

[ottoragam]

I would argue that there may well be a negetive efficency with 'entry level pnp'.   It will cost more to use than well organised manual placement.

I understand that. I was worried about that too when I started considering building the PnP. So far I don't regret it at all. I spent about $800 USD (less if you count only what ended up in the final iteration) for the stuff I used to get the machine through various design changes. Manual stations cost several hundred dollars too, so the cost difference isn't very significant. However, I don't need to do the actual component installation myself, which is a huge advantage due to the tedium of the repetitive task, and my back doesn't get tired due to being hunched over the board.

Maybe I need to babysit the machine a little, but that's also true for other CNC equipment. Configuring the assembly jobs and loading the cut tape feeders is way better that the other part of this ordeal.

[Kjelt]

If you want to spend your time feeding and watering PNP machines then cool.  I want to spend my time to doing fun things.. Building boards is not a fun thing. Its thing that means i can do fun thngs

And that is the difference you look at it as a tool, the starters look at it as a cool project to learn.
And upto a certain amount of boards this is really feasable IMO.
Above a certain amount and you don't like it, let the pro's handle it esp with extreme large quantities.

But using old 2nd hand gear can get you sometime when spare parts become unobtanium or rediculously overpriced since someone thinks he has gold.
Yesterday I tore down an Assembleon AX2.. machine, I got the TH pick and place head partly. Looked online they ask $8k for a NOS 
I also got the bottom camera unit with a Marlin firewire camera. NOS $900.-

Looking at the industrial PC with two processorcards and six! Other cards booting probably Win NT3.x , the firewire cameras and DC brush servos, control card running on an 8044 with LM628 motioncontroller, makes me realize that if something breaks down you are in deep sh*t. Good luck getting the service manuals etc.

But hey if it works for you great, put it in a topic where it matters, but not in the VERY LOW COST P&P topic, you are barking up the wrong tree 

[mrpackethead]

And that is the difference you look at it as a tool, the starters look at it as a cool project to learn.
And upto a certain amount of boards this is really feasable IMO.
Above a certain amount and you don't like it, let the pro's handle it esp with extreme large quantities.

No, i'm looking at this has a 'learning' project as well.     I wrote this several posts ago.

These kind of machines are awsome learning and experimental things.   And they are great for that.    And as long as you understand that, and dont' try to build a business around using these machines to deliver products/services all is great.   

There are some use-cases that this class of machine will cater for.  The problem that seems to occur is that some peoples expectations, dont' match the reality of the limitations....

[NorthGuy]

Meah if you are a startup and sell up to 50 boards a month and your self built openpnp machine places the components correctly , why not? You can earn the money for a better quicker machine if needed.
If something breaks down you can repair it your self since you built the machine. 

You certainly can, but automated PnP head is not exactly the whole machine. You will also need feeders, otherwise your PnP head is mostly useless. Feeders are far more complex. This is not something you can build easily or cheaply.

[SWR]

You will also need feeders, otherwise your PnP head is mostly useless. Feeders are far more complex. This is not something you can build easily or cheaply.

Another more low cost option is to go with fixed feeders. I know it's not acceptable to an EMS with many different customers and PCB designs due to the longer setup time, but if you're only doing a few differen't boards of your own design, then it might be a viable option.

I've got a small desktop P&P with 79 reels, 22 cut tape and 7 sticks plus trays. I can cover my different designs by choosing as many possible from my common components on the reels and then placing special parts on cut tape, sticks or trays.

If I need higher quantities of a non-common component I have to mount a reel which takes longer than swapping a feeder, but it also takes time to mount the reels in a feeder although it's less than the fixed reels.

The big downside of feeders is the initial cost. Buying 100 or 200 feeders would cost more than I paid for the machine including the 79 fixed reel feeders. It's one of those tradeoffs you might be willing to take if you prioritize low purchase cost over time saving.

I think it's time well spent to do a bit of analysis on your current and near future needs regarding component diversity and needed flexibility before choosing a machine. I was in the lucky situation that I had a list of projects I want to do during the next 3-4 years that was fairly easy to assess on the hardware side. Most of my complexity will be in the software development.

[ottoragam]

Another more low cost option is to go with fixed feeders. I know it's not acceptable to an EMS with many different customers and PCB designs due to the longer setup time, but if you're only doing a few differen't boards of your own design, then it might be a viable option.

I've got a small desktop P&P with 79 reels, 22 cut tape and 7 sticks plus trays. I can cover my different designs by choosing as many possible from my common components on the reels and then placing special parts on cut tape, sticks or trays.

If I need higher quantities of a non-common component I have to mount a reel which takes longer than swapping a feeder, but it also takes time to mount the reels in a feeder although it's less than the fixed reels.

The big downside of feeders is the initial cost. Buying 100 or 200 feeders would cost more than I paid for the machine including the 79 fixed reel feeders. It's one of those tradeoffs you might be willing to take if you prioritize low purchase cost over time saving.

I think it's time well spent to do a bit of analysis on your current and near future needs regarding component diversity and needed flexibility before choosing a machine. I was in the lucky situation that I had a list of projects I want to do during the next 3-4 years that was fairly easy to assess on the hardware side. Most of my complexity will be in the software development.

Agreed. There are also low cost 3D printable feeder designs available on the web (like this one https://docs.mgrl.de/maschine:pickandplace:feeder:0816feeder). So if you have a bit of time, you can print yourself some feeders to accommodate for a larger component library.

[Harjit]

Love the simplicity!

I've started down this road and had some questions for you - where did you end up getting the "plates" that you use to connect the y axis rails to the X axis rail, the X axis motor bracket, etc.?

How are you attaching the belt attach point to the plates?

Are you getting your Samsung nozzles and holder off of AliExpress or elsewhere? How is the quality? I need to order some, which is why I'm asking.

So, far, I've got the Al extrusions, the rails, steppers, pulleys, power supply.

[ottoragam]

Love the simplicity!

I've started down this road and had some questions for you - where did you end up getting the "plates" that you use to connect the y axis rails to the X axis rail, the X axis motor bracket, etc.?

How are you attaching the belt attach point to the plates?

Are you getting your Samsung nozzles and holder off of AliExpress or elsewhere? How is the quality? I need to order some, which is why I'm asking.

So, far, I've got the Al extrusions, the rails, steppers, pulleys, power supply.

I milled the extrusion, motor, pulley and camera brackets. The head plate and nozzle holder are also custom machined. The most critical part is the nozzle holder, as low run out will allow for better part placement accuracy.

The belts are attached using a 3D printed clamp where I slide the teeth. I haven't published my design, but there are several available on the web if you wish to make your own, or you can buy some metal ones.

The nozzles are from Robotdigg. I tested some Juki 5xx ones too, but I find the Samsung ones to be better made. Their spring loaded stroke is shorter, for what is worth.

Let me know if you have more questions!

[Harjit]

Thank you.

I wonder / think the head design is ripe for innovation. The current designs use stepper motors with shafts that have holes. These steppers are way more powerful than they need to be. They are also heavy. This then makes the motor that pulls them up and down larger. Which makes the X-axis motor larger. Which makes the Y-axis motor larger.

This thread has some discussion about an alternate: https://groups.google.com/forum/#!topic/openpnp/UexQ1whelYE

I want to take the ideas at the above link a littler further and put a tiny vacuum pump either on the head or on the head carriage. This will eliminate the long tubes and reduce mass and drag.

[Fire Doger]

Thank you.

I wonder / think the head design is ripe for innovation. The current designs use stepper motors with shafts that have holes. These steppers are way more powerful than they need to be. They are also heavy. This then makes the motor that pulls them up and down larger. Which makes the X-axis motor larger. Which makes the Y-axis motor larger.

This thread has some discussion about an alternate: https://groups.google.com/forum/#!topic/openpnp/UexQ1whelYE

I want to take the ideas at the above link a littler further and put a tiny vacuum pump either on the head or on the head carriage. This will eliminate the long tubes and reduce mass and drag.

Too much vibration or very low vacuum.

[ottoragam]

Fire Doger is right. I prefer to have an air line connected to the head so I can use a silent vacuum source, and the small pumps cannot deal with a tiny leak, they don't provide enough flow and pressure.

[Harjit]

Valid concerns.

FWIW, the vacuum pump I was thinking of using is the same / similar one to the one used in the Hakko 394. I have one and took it apart and got the part number from there.

[SWR]

I wonder / think the head design is ripe for innovation. The current designs use stepper motors with shafts that have holes. These steppers are way more powerful than they need to be. They are also heavy.

A few ideas:
How about a dual head built with 3 micro servos?
They are cheap, light weight and low power.

One servo on each head is used for component rotation.
The best resolution I've found is 1024 steps on 180° equal to 0.2° resolution.
Example: Tower Pro SG90 Digital 9g with 1us dead band (SG92R Digital has similar resolution).
That is 0.065mm or 2.5mil steps on the corner of a 30x30mm component.
This can be turned into 360° positioning by rotating the servo before pick-up and again before adjustment/placement.
The nozzle adapter with vacuum tube attachment and servo spline connector can be 3D printed.
The exact cone shape could be improved by turning a soldering iron tip with the exact conus shape,
and heat it to the proper temperature before pressing into the 3D printed conus shape to smooth it out.

The last servo is driving a small tooth belt with the heads attached to each side.
When one head goes up the other goes down and both are level when the servo is centered.
If you need 10mm vertical clearance on each head you need 20mm movement.
That is 0.02mm or 0.8mil vertical resolution.

With a small ESP32-CAM communicating via WiFi and driving the 3 servos and the camera, you only need 2 wire dc power.
This would require good capacitive decoupling of the dc power due to current spikes drawn by the servos.

[mrpackethead]

Using some of the components that you can get as spares from some of the commerical machines is not a bad idea either.    Theres a guy in Australia thats taken some assemblys of a Yamaha machine.     Its also well worth considering using a venturi vacumn generator on the head ( very light ) and just supply compressed air.  Ideall you want both positive and negetive pressure, as being able to 'blow' the compoennet of the nozzle is a really nice thing to have.   They often just wont' drop' off.   You also then have a really simple way of using off the shelf nozzles.

[Kjelt]

The main challenge is still the valves. On commercial machines you see very tiny very expensive swiss made valves and sometimes a custom aluminium milled block of valves.
They have multiple pressures which can be achieved with a pressure regulator, one for the pneumatic head to operate usually in the order of 5 bars pressure and one for blowing of the component which is very low pressure order of 0,5 bar above the roompressure. Crucial is the timing by placing: drop head, release vacuum, apply blow of pressure, release blow off pressure, raise head.

[SMTech]

The valves in the head of our Essemtec are made by SMC and if you look up the part number are made specifically for surface mount machines. Not about to pull the cover off right now but from memory they are DIN rail mounted, no fancy milling and have pressure adjustment grommets on them. Most machines (including this one) use a venturi, it may even be a built into that SMC valve assembly, not about to take the hidden bits apart to find out. Notably Europlacer & Mycronic have built in vac generators, Mycronic needs no air supply at all IIRC and Europlacer uses a little, mostly for the feeder trolley clamps. In software you can define how long before the device is reached the vac turns on, how long it remains at rest before moving up, blow off durations are very short.

The old Versatronics machine we used to run, has a pump in arm, it was more effective than you might think and the lack of blow off never really caused an issue, but then it wasn't even capable of picking or placing anything exotic. However blow-off is a fairly inaccurate name, for the most part its just about clearing residual vacuum, tiny little pumps probably leak enough for that not to matter.

[SWR]

I absolutely agree that you need the blow off to get somewhat decent reliability.
I saw these cheap mineature pumps on Aliexpress.
The 6V version sould fit nicely with the servos for a single DC power supply.

I don't know if:
 - 3,2 L/min per nozzle is enough to pick up the component reliably?
 - Do they have adequate lifetime?
 - Can you switch from suction to blow by reversing the motor polarity?
 - Can you monitor the vacuum (detect pick-up failures) by monitoring the motor current?

When using dirt cheap components (these are $3 each) you have to do some testing to find the performance and lifetime limitations, so it's a some-assembly-required approach and it might end up being a dead end.

[Kjelt]

The valves in the head of our Essemtec are made by SMC and if you look up the part number are made specifically for surface mount machines. Not about to pull the cover off right now but from memory they are DIN rail mounted, no fancy milling and have pressure adjustment grommets on them.

I would appreciate it if next time the cover needs to come off you can take a picture of the type number 
Sounds like an obtainable  item 

[NorthGuy]

I absolutely agree that you need the blow off to get somewhat decent reliability.

I placed quite a bit of components for prototyping manually, and I don't remember them sticking to the nozzle (well if you dip the nozzle into flux or paste, that's another story). How come this is such a problem with machines?

[Harjit]

I absolutely agree that you need the blow off to get somewhat decent reliability.

I placed quite a bit of components for prototyping manually, and I don't remember them sticking to the nozzle (well if you dip the nozzle into flux or paste, that's another story). How come this is such a problem with machines?

From what I've read, it isn't a problem. Apparently, when the component is placed, the surface tension of the paste pulls it off the nozzle.

Pumps I found interesting:
https://www.aliexpress.com/item/32863883713.html?spm=a2g0o.productlist.0.0.4d7d73cdfqFFcb&algo_pvid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02&algo_expid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02-4&btsid=ce5a9812-1180-4322-8446-af36515867df&ws_ab_test=searchweb0_0,searchweb201602_7,searchweb201603_52

https://www.aliexpress.com/item/32857665192.html?spm=a2g0o.productlist.0.0.4d7d73cdfqFFcb&algo_pvid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02&algo_expid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02-11&btsid=ce5a9812-1180-4322-8446-af36515867df&ws_ab_test=searchweb0_0,searchweb201602_7,searchweb201603_52

Don't know if they are adequate or not.

[ottoragam]

They look underpowered to me. I tried the head mounted pump approach, but didn't like the results. I also think they're a bit expensive for what you're getting.

[Fire Doger]

I absolutely agree that you need the blow off to get somewhat decent reliability.

I placed quite a bit of components for prototyping manually, and I don't remember them sticking to the nozzle (well if you dip the nozzle into flux or paste, that's another story). How come this is such a problem with machines?

From what I've read, it isn't a problem. Apparently, when the component is placed, the surface tension of the paste pulls it off the nozzle.

Pumps I found interesting:
https://www.aliexpress.com/item/32863883713.html?spm=a2g0o.productlist.0.0.4d7d73cdfqFFcb&algo_pvid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02&algo_expid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02-4&btsid=ce5a9812-1180-4322-8446-af36515867df&ws_ab_test=searchweb0_0,searchweb201602_7,searchweb201603_52

https://www.aliexpress.com/item/32857665192.html?spm=a2g0o.productlist.0.0.4d7d73cdfqFFcb&algo_pvid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02&algo_expid=c5abc8d4-f9e8-4e49-a8fa-f7eb2891df02-11&btsid=ce5a9812-1180-4322-8446-af36515867df&ws_ab_test=searchweb0_0,searchweb201602_7,searchweb201603_52

Don't know if they are adequate or not.

If your loop has no leaks then surface tension is not enough to pull it. Especially if you use the black suction cup at the tip end.
Its a constant balance between air leaking and vacuum.
If your nozzle leaks (simple needle) or loop in general when you stop the pump pressure is equalized fast and surface tension does the job. -> Need more vacuum
If it doesn't leak when you stop the pump it will need more time for pressure to equalize. -> Less vacuum but you need more component in loop (valve)
I believe having big vacuum is the best because
A) You can PWM it or use valves to limit the airflow
B) You can lift heavy parts by using nozzles with suction cup (high vacuum + low leak)
C) You can have higher jerk without worrying about part shifting or falling
D) You can have a more controlled environment by having a vacuum chamber. Direct pump doesn't push - pull air constantly, its more like sawtooth and chamber acts like a cap.

Making lighter head is very important if you use belts for driving the head because momentum couples directly on motor shaft.
If you use trapezoid lead screw most of the momentum doesn't couple to shaft as a rotational vector (depends on screw pitch, smaller pitch = easier to rotate) so you don't need to care about weight and jerk.
The negative is that you loose speed because pitch increased. The cost increases if you use better motors but I would prefer to buy an accurate slow machine than a fast machine that needs constant observation.