[stewi]

I like to share some insights on a multiple nozzle revolver head.
This is just a drawing and I have no intention to build this. However, I will assist whoever wants to build one.

Starting with a 5 mm thick plate, which is covered with another 1 mm plate I attached a disc stepper to it.
At the front side I have a circular track with two openings. At the lower opening the head segment will exit here for Z-motion, to the right the head segment will engage with a stepper motor for nozzle rotation.

The plate is milled out with two paths, one for vacuum (air/kiss) in the lower pick position, one larger one for the rest of the nozzles.
I've designed an eight nozzle head, but any other number is feasible.

The lower hole for vacuum has a 3 mm opening, all other 2mm.
At the end of the path, I will attach vacuum generators. Likely one for pick and place and 3 for the other.

The wheel hub is two parted. The longer one is attached to the motor shaft and the "star wheel", the shorter containing 8 3mm tubing will slightly press on the plate via spring tension.

I've tapered all holes in the Nylon or Teflon wheel, in order to assure the vacuum will not drop, when the wheel turns.

[stewi]

Front and rear of the revolver head.

There are two discs spaced out by 20 mm. At the front side, I have the tubes coming out from the segments, which will be connected to the tubes from the wheel hub via silicone tubing.

From the rear side we can see some pins sticking out with small bearings, which will engage with the circular track of the main plate.
Motor driven levers will engage with these bearing at Z position and at the rotating station a segment will be pushed out by a few mm in order to rotate the disc with the nozzle plunger, by yet another stepper motor.

I'll stop here for a few days, attending some other business. However, I'll be reading any feedback.

[Nails]

Facinating.....I can see how the vaccum is acheived. But I am still confused about how the rotation of each nozzle is done.

I guess from what you have written, it is locked into position by friction and a spring, and when it is pushed out by a few mm it is unlocked. I am also confused how the rotation station engages to perform the rotation.

[stewi]

Yes, you are on the right track.
In a hurry I couldn't find a longer spring, but I guess you get the idea.
There are two bushings with a pretty tight fit, that we don't get much of an air leakage. The hollow shaft is forced against the lower bushing (with the right spring size).
Upon nozzle rotation at the turning station a stepper motor shaft with a friction wheel (O-ring), which will be located perpendicular to the disc from the segment, will push back on the disc, than rotate the disc with nozzle and then the segment will be moved back to the track.
I'm designing all this as I write. Please be patient.

[stewi]

In my ACAD design, I implemented the location of the stepper motor shaft for nozzle rotation. I'll likely select one NEMA 14 or even smaller.
I also found that the track opening is in the wrong location. I'll change that as well.

Than I'll add the Z-motor and the small motor, which pushes the segment out at the turn station. I guess a smaller 25 mm tin can stepper with gear head will do for both of these jobs.
At the end of the gear shaft, I'll mount a lever, which will require only a few degrees of motion for Z and towards the theta motor shaft.

Again, this drawing is evolving and it will change a few times. Particular the main plate will change with motors and component camera added.
And, of course I also do mistakes here and there.

Hope, my drawings explain better than I do.

[stewi]

I placed 3 NEMA 11 motors in proximity in order to drive the Z axis, theta of the nozzle rotation and one driving the segment against the rotating station.
Since I'm more inclined using servo motors, could you guys telling me what you think is a good speed I could be driving the steppers with?
If I expect 5 rev per second, that makes it one in 200 ms, so if I rotate the nozzle by 180 degree, I could get this in ~ 120 ms (100 + ramp up and down).
Most often the nozzle rotation does not impact the machine performance, if we rotate only by a few or 90 degree. For pushing the segment against the rotating station, I should expect 20-30 ms.

Else, I designed for an 18 mm Z-stroke. In practice Z-travel should not require more than 12 mm. If I want to get 100 ms out of Z travel, I should rather count 40 ms up, 40 ms down and 20 ms for the air kiss.
If you guys could confirm these numbers, than I can continue in my design.

Also, I hope you are ready writing software code for this head. Keep in mind, while one nozzle picks a component, another component is likely in 12 o'clock position for camera inspection and yet another component at 3 o'clock for final rotation from the two previous camera inspections.

[dandumit]

This is just a drawing and I have no intention to build this.

What a pity that you don't intend to build it.
I see that you're so opinionated about this.

I would be interested to give it a try but I am pretty sure that it's not that can be done in a home workshop.

Kind regards,
DAniel

[stewi]

What a pity that you don't intend to build it.

I will not totally erase that thought.
If somebody pays me to build this head, I would do.

Regards,
Stefan

[dandumit]

Like you have presented it until now seems to be doable by a hobbyist ...
I see it like a lasercut base board and few stepper on it. I suppose that troubles are coming on the moment when you start to build the nozzles mechanism.

[stewi]

Getting started could be a big hurdle too.
I am only proposing a design plan, which can be modified.

Also, I am a great fan of this open source concept. We could discuss any obstacles in this project and help each other to resolve issues.
I have quite some knowledge about failure analysis http://www.steering-inc.com/about.html but admit finding bottlenecks in my own design is most difficult.

However, I don't mind critique if it is constructive.

Here is another thought of building the "star wheel" with 4 mm linear shafts.

[Nails]

I don't know whether to thank you or curse you for this information. I've been going round and round with ideas about how to make things a little better and easier for building. But I keep coming back to the conclusion that many elements of this design do indeed work out the best.

Things I have changed are using 7mm wide linear rails/blocks to guide the nozzles, air cylinders for the up/down motion and rotational station engagement (with linear rails), timing pulleys and belt for the rotational motion, and an encoder on the shaft so I can use the index pulse to know the orientation at start-up. I still have details to work out, so I can't post a picture.

I have a few questions.

How small were you able to get the revolving wheel? I've gotten it down to 4 inches. With my lightning nozzle holders and nozzles hanging off of it, it ends up being just shy of 6 inches from tip to opposing tip on the other side. I can't see it getting any smaller.

About fine pitch components; I found some Siemens literature that says the rotational accuracy for the 12 station head is +/- 0.7 degrees, and the 6 station head is +/- 0.3 degrees. It appears the rotational accuracy of the rotary head design is a function of the disk diameter located on the nozzle.

http://www.automation.siemens.co.uk/...0606_final.pdf

So there is some limitations for very fine pitch, large BGA's such as FPGAs. Hence the option of the "TwinHead" mechanism for these parts, which has +/- 0.07 degrees accuracy.

The spring on the nozzle...it doesn't make much sense to me for how it is drawn. Can you include a post showing a transparent body housing with the bearings and spring?

About only picking up 7 components on the first round. You could check the fudicials first and then load the parts to avoid this. But I suppose this is just an optimization so you can already pick up parts while the next board is loading, right?

[stewi]

Looking at the wire frame ACAD drawing, I get a 142 mm diameter. 4" is very nice.
I anticipated a spring which forces the pins in the hollow shaft against the lower bushing. Upon placement, pick up and rotation pin and shaft will deflect the spring.

Wish you'll a happy and successful new year!!

[stewi]

Adding one more thought to your discomfort.
Although I like the idea with linear tracks and re-circulating ball carriers, it makes the head quite expensive and heavy. I would rather consider less expensive slides and easy head exchange feature. Ideally you plug the entire head into a single connector, using alignment pins and a single latch holding the head in place.
Into the same place you can plug in another spare head within seconds or even a single nozzle head, while maintaining the other head.
If your placement head becomes too expensive, you'll likely not effort an extra head.

[gattuso]

Hi,

actually i am confused in the vacuum distribution design.
how it work actually?
Possible to share me the detail?