Robin's smt Pick and Place build

[Robin Hewitt]

I've looked at lots of pick'n'place machines and very few people really go for it.

Put down a strip of components and peel it? No, I want full reels of 8mm tape and tubes of SOIC chips.

Change nozzles by hand? You're not really trying, pick them from a rack.

Look at the component from below with a camera then try and estimate where to put it? Sounds like you made it and now you are trying to fix it with clever software. Physical chip alignment on the nozzle for me.

Is it moving? How many components does it place in a week? Some degree of alacrity sounds like a good idea!

The board is stuffed, now how do you get it out without disturbing anything? Shouldn't you have thought about that before you start building it?

What's that ghastly noise? A modified aquarium pump generating the vacuum could drive you insane.

Is that a rubber pick up nozzle? You must have camera alignment because you ain't going to move it! I also have a notion that a pick and place nozzle needs to leak air. Without the leak your pick up force is limited by the nozzle diameter and vacuum pressure. If you let it leak Bernoulli tells us that air forced through a gap will try to close that gap. You can probably do more lifting with a leaky nozzle than with a hermetic seal :eek:

Whatever. I am going to have a go at building a 24 station pick and place machine and this is my build log. It's only purpose is to spur me on, a sort of motivator, I think I can rely on you lot to keep me going :smile:

First the XY gantry which is peculiarly floppy. There simply isn't enough space to make it rigid because I need access to the table on all 4 sides. The front is for loading pcb's, the back is for the tool rack and the chip alignment station, the left if for tape reels, the right is for chips and more reels. I am depending on 4 screw adjustable feet to keep it square.

First pic: The Y axis runs on two 16mm round rails. Each has it's own 3/4" pitch screw fitted as close to the rail as physically possible.

[Jonathan]

Do you have a CAD drawing of the whole machine you could show us?
Finish looks nice on the aluminium plate :smile:

[Robin Hewitt]

Do you have a CAD drawing of the whole machine you could show us?

My CAD drawings are there to make sure everything fits together and can be assembled. The perspective is reminiscent of ancient Egypt, I keep most of it in my head if I'm doing the machining :naughty:

My CNC package works exclusively with AutoCAD DXF files so I extract shapes as and when I want to cut them. There is no 3D pic of the whole thing.

Suppose I should show that there really is some kind of plan though... :smile:

[Robin Hewitt]

Taking shape and precariously balanced together fo a pic.

Machined the two 1" x 1 1/2" bars that locate everything to everything else. The 15 slots on the left are for reel feeders. The 10 slots on the right are for either reel feeders of tubes of chips, (needs more space between tubes than reels).

I think the Z axis and component rotation thingy is the next job.

[HankMcSpank]

Go Robin, go!!

Looking good - I can hear the suction pumps already!

[george graves]

I joined this forum just so I could watch this built. Thanks! Cheers from the US.

[sigicnc]

Hi, Robin!

This looks like something, I allways want to build (if I had skills like you)

Few weeks ago, I converted engraver to have some fun with placing SMD parts:


http://www.youtube.com/watch?v=-CqpbsTfVtM

Will watch your build with great interest!

[Robin Hewitt]

Few weeks ago, I converted engraver to have some fun with placing SMD parts:

Blimey Sigi, I never thought that QFP would go down in place. Did you stencil the paste mask?

I had a bit of a software blip that slowed me down but I'm back cutting metal (and seaching for 2mm screws to hold a microswitch :whistling:).

This is the start of the Z axis and rotation. It lifts amd lowers by cam action with another NEMA17 for the rotation. Had to use a linear slide for the up down because it descends under gravity and I can't risk it sticking.

The lifting cam has to run on steel because aluminium would dent.

I sank the linear slide in to the 8mm plate so that it can't come apart now the tabs are off :eek:

AutoCAD side view to show how it goes together. Plates in cyan, linear slide in green.

Next the 3/4" pitch nut mountings and the component rotation motor plate.

[sigicnc]

Blimey Sigi, I never thought that QFP would go down in place. Did you stencil the paste mask?

.

QFP was easy, because it was positioned precisely, also PCB was positioned well by two pins. I assembled 14 boards (320 SMDs each) and QFPs went on without a glitch...
And yes, I had proffesional stainles steel stencil for paste.

WOW, nice parts you show in last post!

[HankMcSpank]

I fancy having a dabble too (I'm finding that some of the SMD parts I want/need @0.65mm pitch are just too small for this ageing hand & eyes!) - sigi, what the the pickup tool you used (& vacuum pump)...was it 'improvised' ....or did you buy? (if so have you any linkage that I can partake in?)

I'm curious how the pickup nozzle gets a good seal on the part? Also is the pickup nozzle 'sprung' to avoid stress of the component 'pickup surface' should the parts (or board) not all be completely level bed.

[sigicnc]

I fancy having a dabble too (I'm finding that some of the SMD parts I want/need @0.65mm pitch are just too small for this ageing hand & eyes!) - sigi, what the the pickup tool you used (& vacuum pump)...was it 'improvised' ....or did you buy? (if so have you any linkage that I can partake in?)

I'm curious how the pickup nozzle gets a good seal on the part? Also is the pickup nozzle 'sprung' to avoid stress of the component 'pickup surface' should the parts (or board) not all be completely level bed.

First idea was to use adapted aquarium pump - didn't work, vaccum was ok but not enough flow. Also vacuum was kinda "pulsed". For small parts OK, but big parts - big problems
So I used old refrigerator compressor and fiddled up two relays to operate as two valves: first to "let vaccum enter nozzle" and other to "devacuum nozzle" (so component can be released fast). Big mess, all together.

If I can suggest you, buy cheapest electromagnetic valve and smallest ejector, and connect this to standard compressor or source of pressurized air. Check FESTO, SMC or other manufacturers of pneumatic components for prices and options.

Pickup nozzle presses on component by it's weight only (10g maybe?); in video you can see, it moves freely up when I change nozzle.

[Robin Hewitt]

Got a few hours on it today. hope to have the spadework done soon so I can start on the interesting bits :smile:

[sigicnc]

Z axis and rotation looks so compact... as one piece of metal.:eek:

Beautifullll!

[pdl]

Hi Robin

I have an old ECM85 Mk2 that could do with updating but i would like to make a pick & place machine myself.
I would like to make a machine that could pick up programable chip and place them on a programmer IE Pick Place and Program.

What software are you going to us or using?

I am looking at a used Quad 4C machine this week as i need one for the work i do.

Keep up the good work if i can help in anyway please contact me.

Pete

[Robin Hewitt]

Hi Pete

I know very little about P'n'P machines apart from having seen someone elses in action stuffing my boards and a bit of web searching. I know even less about the software.

Progress is stalled until the workshop warms up a bit, I am a fair weather machinist. OTOH it has sprouted a couple of motors and a toothed belt since my last pic :smile:

If you are used to P'n'P software I do have a question about how other people do it before I write my own...

Board alignment: The holes do not always align 100% with the smt pads. Has that ever been a problem? How does P'n'P software help you square things up?

Robin

[pdl]

Hi Robin

The pick and place machine that i have a the minute is Dos based but i have a copy of Versatronic RV1 software.
Versatronics are not around anymore.

Many thanks

Pete

[pdl]

Hi Robin

I am selling my Dos based pick & place machine with lots of feeders it a MAMIYA ECM85MK2.


Pete

[flippersplace]

Hi, I'm in the middle of a similar project. So far I am good for 0805 parts, 3216 LEDs, SOT23 parts.

Re your query about software:

Mine does not have a vision system, it simply depends on accuracy of the machine driven via Mach 3 and GCode. But I don't do manual GCode.

I have a Excel Workbook with a separate spreadsheet for:
- Heads
- Feeders
- PCB Array offsets
- Each PCB
- GCode output

For each PCB I paste in the Pick Place export from my PCB CAD package. I manually redit rotation, and enter a head and feeder. I paste in a block of formulas alongside.

I then manually enter a step sequence (gives a small leeway to optimise pick up order for efficiency)

The aforementioned formulas reference the Head & Feeder sheets to calculate pick and place offsets for each head/feeder combination, and place offsets for each head/refdes combination

I also update worksheet PCBArray with the PCB name and no of PCB's in X & Y directions for this assembly run.

I then run a macro over the above which for each rotation generates GCODE to:

- Initialise all Mach 3 variables used per feeder to keep count of the the next component to come out of that feeder

Main loop:
- Pauses waiting for operator (me) to place the PCB array holder and use the USB camera to zero offset G55 to the PCB holder calibration point
- calculates and sets an offset for each PCB in the PCB array relative to G55 offset for that rotation
- creates calls to the GCode pick place subroutine for all refdes in this rotation
- creates the GCode pick place subroutine called above

= Repeat until all rotations processed

Once a PCB is setup in the workbook (which only takes 5-10 min) can be tweaked quickly, ie if a head is faulty I just edit the heads per refdes and rerun the Macro.

Seems to work so far.

3533353435353536


3537

[Robin Hewitt]

Hi, I'm in the middle of a similar project. So far I am good for 0805 parts, 3216 LEDs, SOT23 parts.

Must have missed the e-mail notification.

What a splendid looking machine, is that a multi-nozzle head? :smile:

[flippersplace]

Yes its morphed into a bit of a monstronsity. Although I've still a few hundred PCB's to finish running through this machine I've already started collecting bits ready for making a new one based on the experience so far. This one has 24" X & Y screws and I will stay with that size for the new one. The primary factor for speed I'm finding is lower mass, so I'm thinking an overhead gantry type machine, with 1-2 heads on each side. Heads to be pneumatic preferably for both weight, speed and built in 'give' via compression.

For rotation, I think it's actually just easier to automatically rotate the PCB's base rather than each head, particularly as my Excel GCode generator groups sets of pick place operations per each rotation already. Automatic component feeding is always in the back of mind but thus far is not necessary.

But, what I'm really interested in is fast / light mechanisms for centering components, mounted on each head...

3547354835493550

[mcrspark]

I joined this forum just so I could watch this built. Thanks! Cheers from the US.

We want to purchase the automatic Pick & place machine 1500 cph to 2000 cph. can you provide us ? please, send the details on [email protected]

[flippersplace]

Robin, it's been a while and the weather has warmed up over there.. any updates?

[Robin Hewitt]

Robin, it's been a while and the weather has warmed up over there.. any updates?

I'm redesigning the 8mm tape feeders, the design was fragile and complex, I wanted robust and simple with few moving parts.

Then I had decided to forget everything I 'knew' about tape feeders and start again. It was a liberation, I have the basic ideas, but designing that perfect lever system takes time :whistling:

[hixo]

hello.. just wondering, how do you compute for the correct nozzle type for a specific product? is there a standard computation for the nozzle inner hole?
hope someone could answer..

[Robin Hewitt]

Tricky... :naughty:

I don't think you can do it simply as area * vacuum differential > mass of the component.

If you look at the tip it is cut to leak air. Bernoulli tells us that when you force air through an orifice it will try to close that orifice. Perhaps leaky is extra sticky.

[hixo]

so how do you generally select the type of picker that must be use?

[Jonathan]

so how do you generally select the type of picker that must be use?

Pick one :whistling:

[Robin Hewitt]

Aim for the largest that you can reliably pick the part with, allow for the worst case resolution on your tape feeder as well as the nozzle placement.

If physically aligning the part on the nozzle you may need to snatch it back when the tapper opens so a good flow rate helps.

OTOH, if you don't have an autochanger you may decide to compromise on a single tool that picks your entire selection of discretes.

Suck it and see :beer:

[matt-b2]

Hi Robin,

I too am working on a DIY SMT pick and place machine and I've found your work very inspiring.

Can I ask, where did you source the small linear slide for the Z-axis, the leadscrews and anti-backlash nuts from? Also, what shape Z-axis cam are you using?

Thanks,
Matt.

[Robin Hewitt]

I too am working on a DIY SMT pick and place machine and I've found your work very inspiring.

Blimey, I've been called a lot of things in my time but... Thanks :smile:

I got the parts off ebay. The Z cam is an eccentric, miniature ball bearing running on an iron rail. Has to be iron so it can't dent.

I got stalled by an old trap called, "The enemy of a good idea is a better idea". I got a notion that I could put the 8mm tapes really close together and lots of them. It goes like this... Mounted at one end of the X rail is a pin feeder and rubber pad. As the placement head moves over the component to be picked it drops the pin and winds if forwards, drops the pad on the cover tape and moves it back. As the placement head leaves it lifts pin and pad then resets them for next time. I am just waiting for the mechanism to pop into my head.

Good luck with your machine.

Robin

[matt-b2]

Good luck with your machine.
Thanks. Interesting idea re. the tape feeders. I've not put a great deal of thought into feeders yet as I'll probably add them after the rest of the machine is built. So far my very embryonic plan is to actuate each feeder individually with its own low cost DC motor and worm gear drive winding the cover tape over a roller to separate it from the main tape and onto a spool. I'll use some sort of hairpin switch to index the tape via its sprocket holes. I am, of course, hoping that most tapes will cope with being pulled forward via their cover tapes rather than their sprocket holes even though this is not how things are intended to be done.

Matt.

[Jonathan]

Have you made any progress with this?

[Robin Hewitt]

Have you made any progress with this?

Some. I got the tape feeder down to 4 moving parts per tape which is OTT. I want one moving part per tape with all the complicated levers mounted on the head so I only have to make them once.

[Robin Hewitt]

I've got a bit stuck working out the tape feeders, here's a whacky idea...

Remember that famous drill that cuts a square hole? I think I can use that idea to convert rotary to rectilinear motion. A conversion that can get complicated PDQ when you try to draw the mechanism.

The problem is advancing taped components. I want to drop a pin vertically into an index hole, move it forwards exactly 4mm then lift it vertically clear.

With one pin for all tapes, I can have lots of tapes because they can be closer together. The pin feeder mechanism moving with the gantry.

So, using the square drill idea, the plate holding the pin has 2 square holes in it. The holes fit over two oddly shaped shafts turning 2mm off centre. One rotation advances the tape 4mm.
Picture shows 18 degrees in 30 degree steps :beer:

[Robin Hewitt]

... shafts turning 2mm off centre...

Okay so I didn't quite understand the mysteries of the Reuleaux triangle but I do now.

My rounded out Reuleaux. without the pointy bits. looks okay but the centre doesn't move in any approximation of a circle :whistle:

The centre of a true Reuleaux, constrained in a square, follows a sort of circle defined by the ends of 4 elipses. Mine looks like a finger painted flower :nonchalance:

I can get true vertical/ horizontal motion if I fix the square and float everything else.

I have ordered gear wheels and lumps of brass, so I have until that arrives to simplify it.

[Colin Barron]

Hi,
are you still working on the project?
regards
Colin

[GEOFFREY]

My CAD drawings are there to make sure everything fits together and can be assembled. The perspective is reminiscent of ancient Egypt, I keep most of it in my head if I'm doing the machining :naughty:

My CNC package works exclusively with AutoCAD DXF files so I extract shapes as and when I want to cut them. There is no 3D pic of the whole thing.

Suppose I should show that there really is some kind of plan though... :smile:

That looks pretty impressive for "some kind of plan. G.

[Robin Hewitt]

are you still working on the project?

I have 2 problems.

I have to switch PCB supplier and I cannot decide who.

I bought a 3D printer and I have a notion that it will make the component feeders simple if only I can figure it out.

[Stupoty]

I have 2 problems.

I have to switch PCB supplier and I cannot decide who.

I bought a 3D printer and I have a notion that it will make the component feeders simple if only I can figure it out.

If your still working away on this as a back burner thing this may give some inspiration :)
http://www.eevblog.com/2014/11/19/eevblog-684-ness-smt-maufacturing-assembly-factory-tour/

Stuart