An interesting video: Jon Oxer and his p&p machine

[mawa]

Hello everybody,

here's an interesting new video I just found. Jon has some very interesting solution ideas and he references liteplacer and some other current P&P projects.

https://www.youtube.com/watch?v=dvwZeb2uwkg

Especially his Z axis solution is very keen because it works on gravity! I already smashed 2 components due to a disabled z limit with Juhas powerful z-axis drive.

Also the idea of placing an arduino on the gantry to control the solenoids, Z-Servos and, what Jon does not seem to do, the A-steppers and vacuum measurement.
That significantly reduces the cabling to the gantry.

Instead of using RC servos for $70 I would prefer the twin stepper solution from smallsmt. It also works on gravity and only needs one stepper. Take a look here:

http://www.smallsmt.biz/pick-and-place-machine/

By this you get two pickup heads which will speed up the placing. Making it 4 heads is an additional option.
This could diminish the necessity of nozzle changing!
And you don't need to probe pickup and placement heights and keep track of them in the tape information.

As local stepper drivers I could recommend this part for only 25 Euro (without the stepper motor)

http://www.elv.de/intelligentes-schritt ... usatz.html

which I am using in another maker project.

This controller works excellent and extremely smooth with up 16 micro steps and up to to 32 of them can be controlled via i2c.
The only disadvantage I found so far it that the step range (position counter) is limited to a 16 bit which excluded the controller for X & Y axis. But for Z and A they are superb.
The controller does the ramping, position count, speed control and jam-, overheat-, undervoltage- and event step loss- detection.

[mrandt]

Hi Manfred,

thanks for sharing.

As you seem keen to follow other DIY PnP as well, check out Anthony's build for OpenPNP - looks like a great machine:
https://hackaday.io/project/9319-diy-pick-and-place

Head moving down by gravity and being pulled up by belt and pulley is a concept that Volker Besmens used on his machine.

The other approach I've commonly seen is heads being pulled up by springs and pushed down by stepper - Jason von Nieda used that on his initial OpenPnP HW.

RobotDigg sells a similar head construction:
http://www.robotdigg.com/product/674/PN ... 11+Headset

There is also a very active discussion going on in OpenPNP Google Group. Peter Betz has designed a dual stepper head which is looks very promising:
https://groups.google.com/forum/#!topic ... XGpRTdaTrk

All these approach have one downside in common though: They don't have integrated height probing (or not yet).

I must say I like the Z-level control Juha's lead screw had offers. Also, the height probing makes setting things up very easy.

If you look at Adam's and Louis recent posts, I think that mounting the nozzles directly on the stepper shaft also has some cons, such as low A-angle resolution and potentially dangerous sideloads during nozzle change.

Still, I thing it is amazing how much and at what speed the DIY PnP community as a whole is progressing. When I first build my FirePick and attempted PnP, not much of all this was available...

We totally live in the future

[mrandt]

Peter Betz posted details and CAD of his head design here:
http://www.betztechnik.ca/blog/ice-pick ... place-head

[mawa]

Hi Manfred,

All these approach have one downside in common though: They don't have integrated height probing (or not yet).

Malte, could you please explain why you think height probing is necessary?

I can understand that you need it in a spindle driven Z solution like Juhas. There you need to know the height of the PCB and the component height to get the correct placing height and not crush the part after reaching the Z springs compression distance and the same applies for pick up.

But using gravity you don't have to care. Looking at professional P&P machines you will recognize that all important Z levels (pick up feeders, trays, PCB, drop offs) are on the same - let's call it zero Z level.

You know how far the Z nozzle has to travel and you can adjust your Z travel distance accordingly so that it just moves a bit below zero.

Take a look at the pockets in the plastic tape. They have as far as my samples show a depth equal to the components thickness so that the pickup Z is always around our Z zero. Paper tape has the parts also quite level to the paper top.

On placing you can have thicker components like ICs, FETs etc. and you can have high components like large caps, sensors etc. But you know their height from the data sheet and you put that height into the tape data. For those parts the Z travel can be decreased (if at all necessary).

Well Juhas table solution does not have that common Z zero that open frame P&P machines have. But you can get there very easily. Using your really great tape tray blocks their pickup Z height is 9mm above table zero. They dictate the Z zero level. Now mount the PCB approx. 7.5 mm above table zero will bring the PCBs surface to our Z zero. Finally adjusting up camera focus to that height should be no problem.

So why must I probe heights? What do I miss?

[mrandt]

Malte, could you please explain why you think height probing is necessary?

Sure. I don't think it is a must have but I still think it is very convenient.

As you point out, it is possible to build your feeders, trays and cut tape strip feeders in a way that they all have common Z-level.

I think that is a good idea anyways, and it is also smart to bring your PCB to same level (using holders) as that will eliminate focus issues and perspective errros (pixel-mm ratio is dependent of viewing distance with non-telecentric lenses).

But, as you also point out, on placement you need to factor in the height of the component you are placing and this value depends on the physical attributes of the part. When placing, your Z-drive needs to "stop" at just the right level.

If you use direct Z-drive with a spindle like Juha or toothed rods like Peter Betz, this is a requirement. When using gravity (like in a belt+pulley design) you will not "force the part" further. But still, the mass of the moving head assembly might cause damage to the part being placed or the nozzle - especially if accelerated by gravity at high placement speeds.

Of course, you can calculate the correct Z-coordinate if you manually input the height of each part. Afaik, this is how OpenPNP works - and most commercial PnPs.

If you use a nozzle system with a spring (like Juki), this will also increase the tolerances - so you don't have to be too precise.

But when building prototypes, I don't want to further increase my setup times by having to enter the height of each and every part into a form... Also, if I make a mistake and don't have Z-probing / limit switch, I will likely break something with directly driven Z-axis (not so much on gravity / spring driven).

I think Juha's integrated Z-probing made setup easier and operation more fool-proof.

Just my two cents

[Pixopax]

I had my Z-Switch triggered a few times, and was happy to have one.
It would otherwise have ruined my board or the needle.

The gravity-thing is nice, I think of building one, as it has much less moving parts. With a small linear guide its easy to make one.
But I still would make a limit switch on it, just in case the machine drives crazy.

[WayOutWest]

Hi Manfred,
All these approach have one downside in common though: They don't have integrated height probing (or not yet).

Malte, could you please explain why you think height probing is necessary?

It is not strictly necessary, but it is very very very very helpful and useful.

I used to have it, and now that I have had to give it up (in order to use the Juki changer, which technically allows for a Z-sensor but it is incredibly difficult to make it work), I miss it a lot.

I also mount LGA parts (Land Grid Array, like BGA but with solder paste where the balls would be) and for these parts you MUST NOT push down too hard on them or it will smear the paste and produce shorts. Granted most people here are not mounting LGAs, but...