[Eclipze]

I've played around with quite a few designs for a small hobby desktop pick and place machine in the last 2 years. But I've finally come down to a solution I'm going to build :-) It's actually the 20th design model

There is of course many different approaches you can take. I've held a core theme for my design... small capacity desktop machine, light weight, support a flexible design for modifications and support fast movements.

It's not a design for full blown production, rather a design suitable for prototyping or a short pre-production run. I specifically want it to support placement for components with footprint around the size of 0805, SOT23 and fine pitch IC like TSSOPs. The option for two placement heads has also been of interest. This is either for two different pickup nozzles, or to have one setup for paste dispension. I tend to machine my own stencils for small runs anyway, so the paste head mechanism hasn't quite been finished yet.

Ultimately... this will never been anything like having a proper production machine. It's far more of a hobby... some people build ships in a bottle, others are happy to spend many late night procrastinating over a feather weight pick and place machine hahaa.....

I am concentrating on the basics. Not the end goal. Vision based alignment, auto head changing, automatic feeders are all future upgrades. If I spent too much time looking into all the extras, I'd never actually get the first model up and running. So please bare that in mind with any comments.


Basic design specs...
X Axis Travel is ~160mm
Y Axis Travel is ~440mm
Z Axis Travel is ~25mm
X & Y step resolution is 0.0125mm
C Axis step resolution is 0.14 degrees
PCB area is flexible, currently shown as 220x180mm

I really wanted the design to have a lightweight head, to enable fast acceleration and also to keep the cost down. I've looked at many approaches using double rails and ballscrews, which always add a significant amount of weight and cost. However the more I played with the idea of light weight, the more I realised I could achieve a rigid design with just 3 linear bearings, belt driven. I have the igus Drylin T Series rails for X, Y an Z, and 2.5mm pitch, 6mm wide belts.

The X and Y are using NEMA 17 motors, and the C is using a tiny 20x20x32mm stepper. All are bipolar 1.8 degree step. The Z axis are all controlled by RC servo motors. I sure a lot of people are going to choke on that one hahahaa.... I have picked servo motors for good reasons. They are light, very fast, inexpensive (consumable) and easy to drive with the electronics (no stepper drivers). They have a radial movement, which means their translated linear movement has variable precision. The way I have the mechanism set, I achieve the highest precision as the placement tip descends close to the board. In addition, if the tip over travels (tries to crush the part underneath), it can't. The only downward force the tip has is the weight of the carriage/head. Once touching the component, only this weight will be on the component, after which the servo bearing disengages from the mating surface. I'm hoping that although I may not have fantastic precision, I can safetly overdrive the placement without risk of any damage.

The head has a separate tape indexing pin to feed the tape with. Each strip of tape also has a square box where the component is picked up, so component alignment by bumping can be done at the point of pickup. Rather than traversing to a specific cavity to perform the function.

The C-Axis uses a rotary tube fitting, which is held in place by a large bearing. A gear is attached to the rotary tube fitting, whereby the stepper can rotate the bottom half of the tube fitting (which is connected to the pickup needle). The top isn't fixed, however the air tube will restrict it's movement anyway.

I'm using ribbon cable for passing all data/power. All the actuators are so small, that the power requirements are relatively low. I still use 16 wires for the Y Axis stepper (4 wires for each stepper wire) just to be sure. I really like the idea of using ribbon cable, as it tracks so nicely without any sort of guiding mechanism. It's so very light too. It should impeded movement like a track conduit system would, and it much lighter than having an overhead cable bundle.

I have an adjustable camera on the head, so placement can be watched via PC. Expanding to vision based feedback is something for future work.

I have tried to create a flexible mechanical design solution. The use of angle aluminium is the basis to achieve this. If something isn't performing as I'd like, or an upgrade is desired, it's likely that only a bracket needs to be changed or modified. I have used 7.6mm polycarb where where thicker material is needed. The baseboard and the component feeder bases will be made out of corrian. It's a nice hard/stiff material that machines well, and should be nice and rigid for this application. I'm just using freebie off cuts. The frame is inch square aluminium with plastic joints. This leads to future possibilities for what is mounted underneath.

Design Variations
The base board defines the basic layout, which I currently show with placement area in the middle and tape strip feeders on each side. Those tape strip feeders are 8mm tapes, 16 per side. But these can be variations of other tape widths and component trays. I hope to have standard trays with most used component values, then others to cater for more specific needs. If I have a short run to place, I should have enough feeders that I can double up on components that are used more, or swap trays in/out as needed. I've also got the idea to mount reels underneath... if it's worth going to that effort of course. Including upward facing vision for component alignment is quite possible... it's just a variation of the base board to provide the real estate to mount it.


I've spent a lot of time considering many different options. Input is welcome, however at this stage of the game I'm reluctant to make any significant modifications in order to get to first base - something that does the basics. I'm mostly into construction/build mode, modifying the design to suit build/machining requirement. Please keep in mind how small this machine is and it's intent. Aiming for design elegance, not design overkill :-)

I'II be machining the parts over the next couple of weeks, and will post some pictures as I go. Hope everyone finds it of interest!

[Eclipze]

Here's the first part done... the stepper to pulley coupler.

I milled a flat spot on the 10-tooth pulley, which mates with the coupler. It's a tight press fit, and also has a set screw for good measure. The stepper shaft also has a perpendicular pin going through it, so I have a tight slot in bottom of the coupler to mate with this pin.

There was no way I could mount this 10 tooth coupler on the 5mm stepper shaft... there wouldn't be enough metal left after machining the hole hahaa... I initially was going to have the pulley on it's own axle and bearings, then a coupler to the stepper - as you normally see. However I decided to make best use of the steppers bearings and to design a coupler to mate on the outside circumference of the pulley for a rigid solution. This was a far smaller footprint and less mechanically complex solution. Time will tell if the solution has a low enough deflection, or even if a minor angle offset with the stepper mount will suffice.

[scsi]

Hey Eclipze-

This looks really nice. The use of RC servos on the Z-axis is rather clever. Just keep in mind that they tend to drift over time and with temperature swings. Also the lifespan is rather limited on the cheap ones because of random gear failures and encoder wear.

Can your design accommodate for Nema 23 motor on the Y-axis in case if Nema 17 does not cut it? Just looking at the clearance under X-axis and I think that the belt may rub there.

Also, beware of the static electricity that may develop on the plastic parts rubbing against plastic. Like those clear covers on the feeder guides. If that is made of polycarbonate or acrylic, try to get the ESD dissipative variety if you can.

Another observation. I think it may work just fine but I still feel uneasy about those unshielded flat cables carrying stepper voltages. The RC servos may not like the interference and the solution may be to put the stepper drivers right next to the motors. Since you are looking to use small steppers, you can find or build some really tiny driver circuits like this one here: http://www.pololu.com/catalog/product/1201. It can handle 2A and 24VDC just fine and will be able to drive Nema23 no problem.

Regards,
Kyryl

[Eclipze]

The X axis for me is the long axis. I assume you're looking at the belt clearance on the Y axis. It only has 2.2mm as it goes over the brown camera bracket support, which is less than I'd like. I look at machining a slot on that bracket to extend the clearance.

The potentiometers in the servos do tend to wear out, which I'm aware of. You pay for what you get with these things. I've got some cheap GWS servos to start with, which after they die I'II replace with a good quality set. I plan to have an adjustment control via the software to control the servo offset, something that can be adjusted on the fly.

I already have a Gecko G540 for this project, which is a 4 axis 10 step stepper controller. I need 10 bit to get the resolution I was aiming for, while relying on their mid band resonance control to get fast/smooth acceleration rapids. I'II run them at a high enough voltage to get good torque at higher speeds. I could fit a NEMA 23 for the X axis by pocketing it lower in the base board. I've actually already bought it too, a 185oz/in suited to fast RPM apps. It seemed such a big motor for the application that I was worried about it being strong it might cause damage if I hit the limits (even though I have limit switches). Picture below shows the size comparison. The NEMA 23 is on standby.

The ribbon cable may not be good enough without shielding. It's definitely a low cost and very low movement resistance, which is why I'm keen to see if it could work. But if necessary, I have the room to put tracks in to take normal cables. Otherwise, I may need to do some filtering for the servos or try to use some thin flexible metal to adhere to the cable.

[scsi]

The X axis for me is the long axis.

I thought it's your Y that is longer from reading your first message. Never mind, I was referring to the clearance of the belt driving the long axis but didn't think about the possibility of sinking the motor down. I'd definitely go with 23'th frame on the long one.

The ribbon cable may not be good enough without shielding.

Yes, look for adhesive copper foil on ebay or elsewhere. Make sure it's copper and not aluminum or nickel - you'll have to solder it to the ground. But before you do foil, see if a simple isolation of the servo conductors with ground conductors will work. Another sensitive circuit would be your video feed from the camera. It will pick up EMI like crazy unless well shielded.

Let me know if you have troubles locating the vacuum ejector assembly. I have several collecting dust here.

[Eclipze]

I could always use the NEMA 23 and set a lower current limit... so in future I can easily up the power for more speed. Would also mean I'd have margin to tighten the carriage bearing for better lower tolerance (at the cost of friction). I'II have a look at the design tonight

The vacuum ejector assemblies I've seen online look rather BIG. Appreciate the offer though! I'm probably more keen to see how well the mechanism I've designed will work hahaa...

Copper adhesive foil definitely does sound an attractive solution. I hadn't decided yet if the air tube and USB camera cable would run alongside the ribbon cable, or if the video feed would also use the ribbon cable. I'm going to wait until the table is moving so I can get a better feel for how the ribbon cable and/or other cables will move before committing to an option.


For now... the Y-Axis pulley brackets are done. Bearings are a nice press fit. I could do a better job of them a second time around, however the pulley is aligned perpendicular to the rail and should do the job nicely

[Eclipze]

The X-Axis platform and stepper motor mount is done

[rgee]

hi guys
just coming to the end of a job had to load 30 boards with 98 smd parts
so i converted my old engraver to do the job, i have made a gcode post processor that takes a protel file if any body is interested i will post when i have cleaned it up a bit. i agree that it would be nice to have interaction between machine and placement data but that is just one way. i am using mach to run my machine as it is what i have.
machine basic info
old A0 pen plotter
the table is 1mm sheet galuv and is feed back and forth with the paper feed
the z axis was an OLD hard drive
the A axis was made from stepper motor out of old scanner
the nozzel was made from K&S brass tube square, one slides inside the other with a spring for pick/place override
the vacuum is made by a compressed air ejector and turned on and off
with the spindle command
the tape is advanced by placing the nozzle into the pocket and dragging it forward then going back and picking part
some parts are picked from a tray no moving of the tape
a small piece of silicone tube is placed over end of nozzle
to pick the bigger parts
alignment of board and other tasks is performed with a web cam
Peter R

[phomann]

hi guys
just coming to the end of a job had to load 30 boards with 98 smd parts
so i converted my old engraver to do the job, i have made a gcode post processor that takes a protel file if any body is interested i will post when i have cleaned it up a bit. i agree that it would be nice to have interaction between machine and placement data but that is just one way. i am using mach to run my machine as it is what i have.

Peter R

Hi Peter,

I'm very interested in your pot processor. I have a soldering robot that I want to drive via Mach3 from Protel files, as well as set up a Midi Router for paste dispensing.

Cheers,

Peter

[epineh]

This looks good...I am in to see how it goes

Russell.

[rgee]

hi all
regarding photos when i ave finished the job i will take machine apart and take more, i still have the throu hole to do yet and six boards to load to day.
machine was taking 20min to load a board, i have found that i can override the feed by 20% and i had a .3 second vac delay reduced that to .1 that saved just over a minuet so i guess it now is about 15 min per board. this job was made available to me 6 weeks ago so this has been very rapid pro typing and that is why it is a bit rough. my day job is running a Japanese mad PNP for a contract assembler so i have plenty opportunity to dream up ideas on how to not spend $500 000, same job at work would load in about 1min if that. now i know what the other $499 000 are for!
regarding a post for a soldering machine i think your best bet would be to place a pad on an unused layer for every place you want to apply solder,
then output a gerber file of that layer. then all you have to add is the dwell time, approach angle, solder feed, ECT , sounds easy really. gerber file is just a text file full of coordinates. My post has been created in Visual basic 2008 express this is a free download. I am not a programmer, of all the things i do this would be the hardest for me, could not do it with out Google.
here is a little tease, this is the start of the gcode file and the first placement.

(Posted by OsPnP by P.J.R v1.0)
(4/24/2011 11:41:25 AM)
G0 G49 G40.1 G17 G80 G50 G98 G21
G0 Z14
A0
M06
(align fiducial)
G1 X90 Y106 F3600
M06
G0 X80.13 Y-18.23
M06 ( zero machine )
(Placement Number 1)
(Comp ID C4 100N X-75.79 Y158.32)
G1 X-75.79 Y158.32 F3600 (move to pick pos)
G1 Z-0.97 F1000
G1 Y154 F600 (move tape)
G1 Z7 F1000
G1 X-75.79 Y158.32 F3600 (move to pick pos)
G1 Z-0.5 F1000
M03
G4 P.3
G1 Z7 F1000
G1 A90 F30000
G1 X40.704 Y34.754 F3600 (move to place pos)
G1 Z-3.93 F1000
M05
G4 P.3
G1 Z7 F1000
G1 A0 F30000
(------------------------------------)
(Placement Number 2)


photos aligning the fiducial
post processor screen shot
closeup of nozzle
machine head

by for now
peter r

[phomann]

regarding a post for a soldering machine i think your best bet would be to place a pad on an unused layer for every place you want to apply solder,
then output a gerber file of that layer. then all you have to add is the dwell time, approach angle, solder feed, ECT , sounds easy really. gerber file is just a text file full of coordinates. My post has been created in Visual basic 2008 express this is a free download. I am not a programmer, of all the things i do this would be the hardest for me, could not do it with out Google.
here is a little tease, this is the start of the gcode file and the first placement.


by for now
peter r

Hi Peter,

Thanks for the reply. My plan is to use the Protel P&P file, using the x,y,theta data along with the footpad id. In Mach3 I'll have a macro for each of the Footpads that will be called.

For instance there will be a macro for a 0805, SIOC8, SIOC14 etc. The GCode will call the macro with the X,Y,theta position. The macro will then do the paste dispensing for the pads within the component.

Cheers,

Peter.

[Xerxes]

Nice work Eclipse! I think the design has many great ideas.

However, I'm bit suspicious about accuracy of Drylin combined with stepper and belt drive. All of them are quite inaccurate devices in some circumstances. You may end up having only 0.5 mm accuracy. I woud use at least some preloaded ball rails and servos to get within 0.1mm. Also speed can be much higher with servo as you can push it to limits without worrying losing steps.

I'm also starting to design a pick'n'place machine at my work. Instead of SMD it will be handing unmounted laser chips and doing automated testing on them. It has great accuracy requirements but speed is less critical. It will use machine vision to pick-up parts precisely.

[Eclipze]

I wanted to make sure I had a step resolution close to 0.01mm, but the actual repeatable accuracy is certainly more difficult to put a number on. I've be very happy with 0.1mm. If it was 0.5mm, I think I'd be pushing it to place 0805 reliably. TSSOP packages with 0.5mm pitch would need operator assisted adjustment/release... come to think of it, that would be needed for SOIC's too. I'II be tweaking/tightening the carriages up before I accepted 0.5mm hahaa...

The rails/carriages are overkill with respect to the load they carry. Keeping the weight of the head assembly low will hopefully contribute to the overall accuracy and speed. The Y-Axis rail is quite stiff with respect to the load, and with the weight of the head assembly being low I'm not concerned about it bobbing around under acceleration forces. If there is bobbing, I would be more suspicious about the flex in the baseboard first. However the design is for prototyping and pre-production, where speed isn't as important. I'd be more than happy to get around 500..800cph. But have no expectation of hitting 4000-5000cph compared to a production loader. Servos are expensive, and for such a small design these steppers (particularly that massive NEMA23 on the X-axis now!) is more than ample to move the load quickly.


The X-Axis belt plates are done! The belt is going to be cut to length and both end secured at the carriage. The belt ends will loop up through the hole (as shown in post #7) and retained by the mating track in the plate on the top. So they cannot slip out with use.

This top plate with the fine pitch tracks was a pre-test to make sure it was a good fit before trying to the same with the Y-Axis carriage. It's a good fit too, very happy. The depth of that track still leaves the belt slightly higher, so I have some compression when it's bolted in place. I also made the track wider than needed, so I had some sideways adjustment and also because the square edges of the tracks won't butt up nicely to the pockets machined with a round end mill tool.

The plate that sits underneath has two purposes. The first is to hold the belt nice and straight, pointing the belt at the pulleys. The second is the vertical part is going to be used for the opto-switches that will be mounted on the baseboard. This plate will pass through those at the limit of the travel (limit switches).

Now I know the track profile is good... next is the Y-Axis platform bracket.

[scsi]

I think we are misusing the term accuracy here. This particular machine does not have to be accurate or precise (within reason) but repeatability is paramount for the basic PnP functions. The thing will be traveling from point A to point B and it does not really matter how straight and true that motion is. What is important is that the business end hits the destination coordinates every time within 0.1mm. Does not have to be linear motion either - it's just cutting air so to speak.

I'm saying all this assuming that the machine that Eclipze is building will have some teaching function when you can define those points A and B manually for every part. Typically they use down looking camera with crosshair.

On the other hand, the industrial machines must be more or less accurate as they rely on motion linearity a lot. While many machines have manual teaching, the destination point of every placement motion is getting calculated by the computer based on the fiducial correction and part position correction factors.

I don't think Eclipze is going after fiducials with this machine or going to calculate the coordinates on the fly for every part. With this in mind, I'd suggest you target repeatability and not precision of the steppers. In other words, there is no reason to use 10x microstepping trying to get 0.01mm positioning precision if you cannot repeatably position the machine due to sloppy bearings, belts or weather fluctuations. Keep in mind that with high microstepping rates your torque reduces dramatically and the step loss probability increases. Try it with 1x or 2x stepping and see if repeatability is there. If it's not, solve that problem first and then go after higher precision.

For reference, on my machine I was playing with stepping parameters and ended up using 0.02mm/step and doing it throug "macrostepping" if there is such a term. Essentially, the servo drive has software gearing function when it can subdivide incoming step/dir pulses (microstepping) or it can multiply them. In my case I'm using 200x *multiplication* because the native travel distance of the linear servos is 100nm/step and that's a little bit too precise. That's nanometers...

-scsi

[Eclipze]

The linear stepper shaft is rotating on the suck back stroke and some of the downward stroke. I was hoping it wasn't going to. However the motor doesn't have enough grunt to push the paste without skipping steps anyway. So I plunge, pause and retract, skipping steps during the plunge anyway, however it has consistency of travel. Running the linear stepper at 12V too, not 5V as it's rated.

I had to go to a plastic tapered nozzle to reduce the resistance, as the metal pin type was just too much for the stepper to dispense through. I attached four 1/4W resistors around the plastic tip and run some current through, to warm the paste in the hope it was more compliant. Didn't actually make much of a difference.

I could get an amount to eject consistently (without drooling too) if the amount was large enough, but it was a bit much for an 0805 pad.

I think the paste is too dry for small spots and not sticky enough to adhere to the board. Perhaps I need to mix with some thinner. I am using paste from the jar, which I normally use for stenciling. The jar is recent and refrigerated, but the syringe I made up has been sitting on the bench for a couple of weeks now. I was considering buying another pre-filled syringe to see if that paste is thinner. Comparing to the youtube videos of paste dispensing, like the Taig mill one, there paste is very wet and I'd expect much better success with this.

Having trouble with the file upload function again, so here is a link to the picture instead...
http://www.eclipze.com.au/pnp/paste_dispenser_02.jpg
BTW... the meltage on the tip was from heatshrinking the resistors in place <oops>, not from the heat the resistors were pumping out.

Few things I can do... try paste from a purchased syringe. Find out from the manufacturer what I can use as a thinner. Also, move to a thinner syringe vessel, which will reduce the resistance and give finer precision.

I don't want to go to a larger linear stepper, due to the weight and physical size. I have been very careful with keeping the weight low on the head assembly. Not concerned about the X and Y steppers... they zip along faster and with more acceleration than practical hahaa... Currently, the weight distribution with the head assembly fully extended is giving a weight loading of ~330g. It won't be until I have the X rail bolted to the base board than I'II be able to see what level of flex/movement this produces under acceleration. Until I see the effect, I'm not going to rush out and buy a support rail. I have the provision for it, as it may need it, but not the evidence to justify adding it to the design yet ;-)

[scsi]

There are two types of solder pastes, one for stenciling with higher viscosity and close to 90% metal content and one for dispensing with 80-83% of metal content and much lower viscosity. Those are the "by weight" numbers, not the volume, so the actual difference is much greater than 10% due to the weight of the metals. Try to get the paste designed for dispensing paying attention to this "metal content" parameter.

What gage size needles and tips are you trying to use?

On the motor. Are you using microstepping or full step mode? What current? That motor can take 0.5A or higher intermittently.

On a side note. Have you considered making some parts out of Delrin or other engineering plastics? It would help with weight reduction and wont compromise the rigidity. You are already using plastic in the most critical spot - the bearings, why not use it elsewhere? That would possibly allow you to put a bigger motor on the dispenser.

-scsi

[scsi]

I'm not sure if I posted my dispensing videos here. I'm using custom auger valve mechanism in both, but the first video with 0.5mm dots is nothing special. Believe it can be done with the motor driven piston or regular pulsed air without much trouble. Dispensing paste type-3 with 83% metal content is used here, not the regular stenciling type.
[ame="http://www.youtube.com/watch?v=tPjFQ15Zdp0"]YouTube- Solder paste dispensing with auger valve - test #1 22GA 400x 0.5mm dots[/ame]

The second video shows mixed results in a first attempt of depositing 10 mil dots (0.25mm). Thinner paste and smaller nozzle would help and I'm still working on it. Don't try this at home!
[ame="http://www.youtube.com/watch?v=4EGH5xG-wSc"]YouTube- Solder paste dispensing with auger valve - 0.25mm dots 0.5mm pitch[/ame]

[FlyingElectron]

I think the paste is too dry for small spots and not sticky enough to adhere to the board. Perhaps I need to mix with some thinner. I am using paste from the jar, which I normally use for stenciling. The jar is recent and refrigerated, but the syringe I made up has been sitting on the bench for a couple of weeks now. I was considering buying another pre-filled syringe to see if that paste is thinner.

I've had really good success with the 63/37 solder paste from zeph at http://www.zeph.com/zephpaste.htm . I use it in a solder paste dispenser very similar to yours with a linear actuator pushing down on the plunger of a syringe. The paste is really sticky and sticks to the board very well. I also tried using regular stencil paste in a syringe and the results were that it would not stick to the board reliably. I have a tube of the Kester R276 dispensing paste that I haven't tried yet, but I'm hoping it works as good as the zeph paste since it is cheaper.

[Eclipze]

I was using a 0.8mm nozzle. Stepper was driven with full step. It is a 2.7W 5V stepper, which I was driving at 12V. After a dispension sequence, the driving board de-activates all driver phases to avoid it getting too hot. The paste I was using is a Tin-Lead, particle size T3, T4, T5 with 88.5% metal loading.

The supplier I use, AIM, also have solder paste for the MyData MY500 Jet Printer. That is a machine that uses an auger and solenoid to shoot paste dots. It comes in 30cc syringes too. There is a good animation of this half way through this video...

[ame="http://www.youtube.com/watch?v=f70nZQFDEhs"]YouTube- Solder paste application using Inkjet technology[/ame]


I've had thoughts of considering an auger valve approach.

I have considered delrin before as a friend uses it often, however most of the weight is in the motors, the rails and bearings. The aluminium used is a only a minor contributor.