Very low cost Pick and Place kit

[ottoragam]

Howdy people,
I recently built a pick and place machine to help me with some low volume assembly jobs. I wanted something simple and low cost, but reliable and expandable. The machine worked very well, I was able to do in a few hours what would've taken me 2 days to do by hand. So I thought more people might benefit from such machine, and I'm planning to offer a kit starting at about $500 USD for the base configuration.

The head uses a hollow shaft stepper to allow a DC suction pump to suck the components. The DC pump could be an optional accessory, to allow the final user to use something else to provide vacuum in the case less noise or different performance is needed.

This assembly also has a down looking HD camera with an interchangeable lens that can be manually focused, with a similar optional up looking unit.

The nozzle travel is 35 mm and it uses Samsung CP45 units, with an automatic nozzle changer option.

The kit will include the custom CNC milled and turned parts, Al extrusions, and linear motion components.
Work area is between 350 and 400 mm in X and Y (could be bigger), and the linear motion is achieved using linear rails, timing belts and stepper motors. Everything is controlled by a pair of GRBL boards (I need to design a custom head mounted control board, and a base motherboard to plug everything in nicely), and OpenPnP provides the brains for the operation. At this moment, the machine manages around 750 components per hour using computer vision and it handles 0402s just fine (I've not tested 0201s yet).

Gino

[spanner888]

OK, interesting, but please clarify some details.

1. Are you selling the modules assembled as per photos?
2. Lost cost = great and from your comments, you have not yet finalised all the parts, but give us a ballpark idea at least!
3. Gotta link us to a video of pnp in action and some close up photos of finished PCBs

[KE5FX]

Sounds interesting!  Definitely would be good to see a video.

[ebclr]

What about the software, to drive those things?

[ottoragam]

OK, interesting, but please clarify some details.

1. Are you selling the modules assembled as per photos?
2. Lost cost = great and from your comments, you have not yet finalised all the parts, but give us a ballpark idea at least!
3. Gotta link us to a video of pnp in action and some close up photos of finished PCBs

Yeah, the idea is to sell a kit containing the prefabricated mechanical en electrical modules, so it's easy to ship and assemble by the final user.
I believe the cost will be around $500 USD. I plan to offer tape trays, an automatic nozzle changer, and some other items as optional extras too, but the base model will be capable of running complete assembly jobs.

Here's a little video I made of the first test. Keep in mind that I was connecting and disconnecting the pump by hand, so it's a bit sloppy. I'm designing a PCB that will serve as a calibration job. I'll post more info as the project progresses

[ottoragam]

What about the software, to drive those things?

See the comment above, it is a video of my first test. I'll post a complete assembly job in a few days.

[ottoragam]

What about the software, to drive those things?

The machine uses OpenPnP to prepare and run the jobs. It works really well, and I'll be providing a configuration file, manual and tutorials for using the software.

[ottoragam]

Here's the assembled machine:

[electrode]

This is pretty cool. I notice the nozzle is (softly?) sprung – does this make (auto)levelling quite easy? Can you just tell it the board thickness and the tape thickness and let it rip?

[ottoragam]

This is pretty cool. I notice the nozzle is (softly?) sprung – does this make (auto)levelling quite easy? Can you just tell it the board thickness and the tape thickness and let it rip?

Yes, the nozzle tips can be compressed to allow for some Z error. And indeed, you need to tell the software the Z position of the surface of the PCB and where the component tapes are mounted, and the thickness of each unique component. OpenPnp handles the rest.

When I shot the video I was using off the shelf Juki 500 series nozzles, and I recently got some Samsung CP45 nozzles because I want to take advantage of its tapered body to design a very simple automatic nozzle changing station.

[langwadt]

Here's the assembled machine:
(Attachment Link)

I think it could be reduced even more, the mgn rail is probably straighter and stronger than the 20x20mm alu extrusion  so I don't  much need for that on the X-axis

if using two different size of rails the hole pattern on the smaller rail fits the hole pattern on the bigger carriage

this pic mgn15 and mgn12, but the same goes for mgn12 and mgn9

and if you stand the x-axis stepper like the y-axis the x/y-axis can be flat plate

[Kjelt]

I think it could be reduced even more, the mgn rail is probably straighter and stronger than the 20x20mm alu extrusion  so I don't  much need for that on the X-axis

But Inthink you need the height for the z axis and components ?

[langwadt]

I think it could be reduced even more, the mgn rail is probably straighter and stronger than the 20x20mm alu extrusion  so I don't  much need for that on the X-axis

But Inthink you need the height for the z axis and components ?

it doesn't change the height of the x-axis

[ottoragam]

I think it could be reduced even more, the mgn rail is probably straighter and stronger than the 20x20mm alu extrusion  so I don't  much need for that on the X-axis

if using two different size of rails the hole pattern on the smaller rail fits the hole pattern on the bigger carriage

this pic mgn15 and mgn12, but the same goes for mgn12 and mgn9

and if you stand the x-axis stepper like the y-axis the x/y-axis can be flat plate

I'm using the 2020 extrusion to allow attaching stuff easily to the machine frame, like feeders, fixtures, a table... It's more of a feature for the final user, makes expanding the machine easier. Also, the 2020 modules are easier to ship than a base plate.

[langwadt]

I think it could be reduced even more, the mgn rail is probably straighter and stronger than the 20x20mm alu extrusion  so I don't  much need for that on the X-axis

if using two different size of rails the hole pattern on the smaller rail fits the hole pattern on the bigger carriage

this pic mgn15 and mgn12, but the same goes for mgn12 and mgn9

and if you stand the x-axis stepper like the y-axis the x/y-axis can be flat plate

I'm using the 2020 extrusion to allow attaching stuff easily to the machine frame, like feeders, fixtures, a table... It's more of a feature for the final user, makes expanding the machine easier. Also, the 2020 modules are easier to ship than a base plate.

sure you need the square frame, I was talking just the x-axis

[ottoragam]

sure you need the square frame, I was talking just the x-axis

Gotcha, that could be an option too. Also saves a little weight.

[SWR]

sure you need the square frame, I was talking just the x-axis

If you want to make stuff dirt cheap you've got to remove as many components as possible and make the remaining components as cheap (often=flimsy) as possible.
A few options:
- Scrap the frame and bolt the MGN9 rails to a cheap & easy to clean laminated kitchen shelf.
- To get away with the flimsiest dimensions possible yo need to reduce the moving weight.
    - Only drive first axis on one side and let it follow freely on the far side.
    - Put 3 stepper motors on the outside of the near side so the moment of inertia balances out with the cross rail.
        - This will be affected by the position of the head on the cross rail, so keep it light and balance the center position.
        - On the far side you only have light weight rollers to return the thin 6mm tooth belts.
        - One belt will drive the head position on the cross rail (belt in any orientation).
        - Another belt will drive the vertical position of the two pic & place heads (belt horizontal).
            - When one head goes up the other goes down.
            - This belt will have to move synchronously with the head position belt to keep the vertical position stationary.
        - The third belt controls the rotation of both heads (belt vertical).
            - This belt will also have to move synchronously with the head position belt to keep the rotation stationary.
    - All electronic and vaccuum control should be placed with the motors so the only connection is power and vaccuum.
        - Maybe a couple of DC supply wires could be placed inside the vaccuum connection tube to reduce the clutter?
    - Find the mechanical resonance frequency of the assembled cross rail (this will vary with head position on the cross rail).
        - Accelerate and decelerate the cross rail on a cosine curve with the frequency matching the resonance frequency at a given head position.
            - This will prevent vibration when the cross rail has reached its end position (just like you do when controlling a gun turret to fire quickly).
            - You essentially replace structural rigidity with more complex acceleration and deceleration equations (but math is cheap these days).

The only mass attached to the head should be a linear bearing, belts, belt fittings, the suction heads, vaccuum tubes and a very light camera.
Disclaimer: These are just penny pinching ideas - I have no practical experience with this specific application.

[ottoragam]

The current machine structure is for sure not the definitive and final product. Thanks for the ideas, SWR, I still need to further optimize the design, so I appreciate the feedback very much.

[ottoragam]

I wanted to make a video to showcase the general operation of the SimplePnP. The machine is sorting a few scrabble letters using my new Samsung CP45 nozzle holder, without computer vision assistance. Here I'm using a dial vacuum gauge to make sure suction is achieved. Normally, a pressure transducer is used instead, to provide feedback on whether the component was picked correctly or not.

[fki82]

Cool idear!
I’m currently building a OpenPNP based machine myself. But mine has a lot more parts…
I’m looking forward to seeing with how little and simple parts you can get away and it still works.
Are you planning on having bottom vision?

[mrpackethead]

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.   

[Kjelt]

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. 

Go try that at your professional factory where a team has to be flown in from hundreds of kms away to repair your machine. And where you so heaviliy depend upon that quality that this hours of non productivity costs you a lot of money.

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.
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.

[Fire Doger]

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. 

Go try that at your professional factory where a team has to be flown in from hundreds of kms away to repair your machine. And where you so heaviliy depend upon that quality that this hours of non productivity costs you a lot of money.

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.
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.

In other words, hand assembly to full equipped 5digit PnP is a big gap and has it's market. PnP owners are a bit fanboys some times...
From my research the most expensive part was the header, cameras and step controllers. If you have a mill you can make a very nice machine, slow, but accurate...

[ottoragam]

Cool idear!
I’m currently building a OpenPNP based machine myself. But mine has a lot more parts…
I’m looking forward to seeing with how little and simple parts you can get away and it still works.
Are you planning on having bottom vision?

Yes, I use the same camera model for bottom vision, just with a shorter focal distance. I still need to do some work on an illuminated enclosure for it tho.

[ottoragam]

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. 

Go try that at your professional factory where a team has to be flown in from hundreds of kms away to repair your machine. And where you so heaviliy depend upon that quality that this hours of non productivity costs you a lot of money.

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.
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.

Spot on.