Better nozzle with auto change?

[JuKu]

This started on the feeder thread, but deserves a thread of its own, imo. Mrandt wrote a great start: http://liteplacer.com/phpBB/viewtopic.p ... t=20#p1297

The OpenPnP community is discussing about using Juki nozzles, and one member is going to (or at least thinking about) offering a nozzle adapter for those:
http://portfolioabout.me/store/. Click the image, and you'll see a high resolution version. I don't see why the shaft needs to be that big, though. The big black spring is for the change mechanism, I suppose. In production environment, the Z values are set, and the small spring in the nozzle takes care of the down force and minor mechanical variations. In prototyping environment (LitePlacer) this is tedious, so the heights are measured. Using these nozzles, I would think it is possible to make (3d print?) two small half circle disks that click together to form a solid disk. This would go in the nozzle at the bottom end of the small spring (referring to the image, between the small green part and the spring in the middle, between the small black part and the spring in the bottom). Another fancy (3d printed?) part holds a limit switch so that it is triggered by that disk.

A nozzle set is reasonable priced: http://www.robotdigg.com/product/467/NO ... MT+Machine. Not so cheap than some other type of nozzles but tolerable, imo. Using a hollow shaft stepper motor is feasible, and greatly simplifies the rotation mechanism: http://www.robotdigg.com/product/43/Nem ... pper+Motor or http://www.robotdigg.com/product/85. The first one is tiny, but maybe still powerful and fast enough; after all, rotation needs about no force.

The only thing missing is a adapter for the vacuum tube. I would like to see a sealed thing that goes to the other end of the hollow shaft from one end and takes the tube to the other end, so that the bottom end rotates when the upper end doesn't. Anybody know about pneumatics enough to show us something like this? Alternatively, the current way the vacuum tube goes inside the pickup tube and seals itself (not quite, but good enough) is kind of a hack, but works. Maybe it could work again with this:
http://www.robotdigg.com/product/256/Li ... 5+to+8+L25.

This start to sound like an upgrade kit / enhanced design!

[mrandt]

Hi Juha,

thanks for pulling this topic into a separate thread

I am adding my comments right away.

I don't see why the shaft needs to be that big, though.

I think the Juki nozzle holders could easily be shortened for our use case. William mentioned in OpenPNP forum that he cut the shafts shorter on a lathe.

Using a hollow shaft stepper motor is feasible, and greatly simplifies the rotation mechanism

I have ordered one of those Nema 8 hollow shaft motors a while back. I think the torque will be more than enough for fast rotation of parts. The shaft is 5mm outer and 2.75mm inner diameter.

The only thing missing is a adapter for the vacuum tube. I would like to see a sealed thing that goes to the other end of the hollow shaft from one end and takes the tube to the other end, so that the bottom end rotates when the upper end doesn't.

For the airtight coupling, a great solution are rotary joints. I use them on my machine already. Misumi sells them with one touch coupling for standard tubing on one and threads on the other end. An R1/8 thread would probably fit the Nema 8 shaft diameter - only problem is to cut the thread as it needs to be perfectly centered. Also, I don't have any imperial unit (inches) thread cutters... I wonder whey MISUMI don't have M3 as an option - as M5 is also available.

Article number is RTCN - see here:
http://us.misumi-ec.com/vona2/detail/11 ... 70/?Inch=0

Another (hack) option I considered is using a short piece of tube as an adapter - cut an M5 thread to one end, drill up the bore to 5mm at the other end - simple to make on a lathe. I would then glue that adapter to the stepper shaft using loctite or similar metal bonding glue and screw the rotary joint into the top. I guess you had something similar in mind with the blue aluminium connector you posted a link to.

For mounting the nozzle holder to the motor shaft I dislike the set screw solutuion. It will probably kill the centricity. Either we need three set screws ar 120° angle to make it adjustable or also use a threaded version which screws onto a shaft. Press fit could be another option, but I am not sure how good the tolerances are on the motor and holder from China.

If we went to order new motors anyways, an even better option could be article number 11HY5402-14S8 from RobotDigg - scroll down on the following page to find the specs:
http://www.robotdigg.com/product/85/Nem ... pper+Motor

It is Nema 11, so slightly bigger but not too big. The shaft has M8 outer thread at the bottom end to mount a nozzle holder and an M5 inner thread at top - which would be the perfect match for a Misumi RTCN6-M5 rotary joint.

They show that motor with a Juki nozzle holder in one of the pictures, but it seems that holder is not sold by RobotDigg. At least I could not find it in their shop - might be worthwhile to enquire about it.

Otherwise I think the guy selling the nozzle holders on eBay could make a custom version with a shorter shaft and M8 inner thread:
http://www.ebay.com/itm/Customized-SMT- ... SwAYtWOGI0

I would think it is possible to make (3d print?) two small half circle disks that click together to form a solid disk. This would go in the nozzle at the bottom end of the small spring [...]. Another fancy (3d printed?) part holds a limit switch so that it is triggered by that disk.

This is the part I don't like. Additional disk and switch would probably interfere with vision and also looks clunky

Karl mentioned that the nozzle has an inner tube that pushes upwards when the nozzle spring is compressed. Maybe that is the better solution to replace the lower Z limit switch - although cable routing and slip rings pose a challenge given the limited space...

I believe we have to put some more (collective) thinking into that problem.

[JuKu]

> For mounting the nozzle holder to the motor shaft I dislike the set screw solutuion. It will probably kill the centricity.

Maybe, but maybe not. The image shows 5.03mm inner diameter, so pretty accurate. The Robodigg small motor is spec -0.015 max, so the centricy error should be less that 0.05mm. Besides, with this scheme the centric calibration is likely a setup phase operation, not run time as it is now.

> Additional disk and switch would probably interfere with vision

It wouldn't. This scheme might work without calibration and even if not, is at most a software tweak away. The vision can be programmed so, that it looks only for stuff close to center, masking off or ignoring the disk and switch.

> and also looks clunky

Agreed. On the other hand, it isn't very pretty now either.

[mrandt]

so the centricy error should be less that 0.05mm. Besides, with this scheme the centric calibration is likely a setup phase operation, not run time as it is now.

The bearings in the motor will probably be not 100% centric either - so why add additional error? This error (though small) could be eliminated by using multiple set screws or alternative mounting.

> Additional disk and switch would probably interfere with vision

It wouldn't. This scheme might work without calibration and even if not, is at most a software tweak away. The vision can be programmed so, that it looks only for stuff close to center, masking off or ignoring the disk and switch.

Hopefully nozzle wobble calibration would become unnecessary with these nozzles (or a one time setup thing for the holder and motor).

Still, I think we will see increased need for up camera part detection and offset correction - thinking of TQFP and QFN No-Leads packages and BGAs. Nozzles already have nice green shades so they can be disregarded by camera. If I could avoid additional distractions, I would do that.

To toss in two wild ideas in addition to Karl's already working approach with contact pin:

1) Optical detection of nozzle compression
- Use some transparent coupler (acrylic) at top of motor
- insert a thin plastic pin through motor shaft and coupler - need to find a way to mount it without obstructing flow of air
- if nozzle pushes inner tube upwards, the pin is lifted and becomes visible in acrylic coupler
- if nozzle moves out again, said pin falls down again and disappears from coupler window due to gravitiy (or additional spring)
- we could easily use IR reflex sensor or light barrier to detect the pin appearing in the transparent coupler
- alternatively, we could fix small magnet on top of pin and use a hall effect sensor

2) Create secondary, spring loaded linear rail for stepper mount and use micro switch
- Mount stepper to short linear rail
- Use spring or gravity to push it downwards in normal operation
- If nozzle hits obstacle it is compressed, once fully compressed it will move stepper assembly upwards
- upwards movement of stepper could be detected using micro switch, optical or other sensor
- during calibration determine necessary backoff to only compress nozzle spring, not move stepper during normal operation
- stepper assembly will only be moved on secondary linear rail during height measurement - once all Z-levels are calibrated the backoff will make sure the pushing force on parts is minimal (software already does that)

What do you think?

[dave]

I think I have the perfect solution now for the top bearing and air coupling + also a DIY slip ring to sense Juki Nozzle from inside the shaft like karl. I will try do another drawing later and share with you as you as I am sure you can give me more input.

I ordered some slip rings already last night from china with a 5mm hole for £10 each, but I think they are just to bulky and tricky to attach. so I decided to order bits to make my own.

Ok. ontop of my motor I am going to put Vacuum hose. 5mm ID. 10mm OD. around 40mm length. it will fit between my motor and rotory bearing and fix using some hose clips (I use m5 SMC brand bearing). Remember this coupling dont need to be so rigid, just airtight so plastic hose should work as coupling.

I am going to drill a tiny hole in the side of the vacuum tube coupling and drop down some very thin recored player tone arm wire. down to the same pin configuration as Karl.

I will strip the wire outside the vacuum hose. and then put copper pipe with a 10mm ID over the vacuum hose. this should contact the stipped wire without solder.

I then have a nice round insulated conducting surface around the coupling, I will make a copper arm that is sprung against this rather than the mess of wirewool and glue karl has.

[mrandt]

DIY slip ring to sense Juki Nozzle from inside the shaft like karl

I get it. One suggestion: Instead of making your own slip ring, use a ball bearing as I just suggested to Karl here:
http://liteplacer.com/phpBB/viewtopic.p ... 1312#p1312

Should be pretty easy to get one which fits the air hose you mentioned.

Instead of a hose, a custom made plastic part could connect motor shaft to rotary joint and at the same time support a bearing or slip ring contact. Should be fairly easy to either 3D print or cut from acrylic or other plastic.

Maybe afterall the electrical contact inside the nozzle holder could be the simplest and a very reliable solution? I think I have to see the holder and nozzles in more details to make a final judgement for myself

[dave]

Oh yes. Nice one, great tip on the bearing. I probably already have something suitable.

I already ordered the all the Nozzle parts Karl has. I am confident in this solution.

[jarekk]

I can show you my prototype - a bit different idea.
Uses professional Siemens Siplace nozzles ( got bargain for them, but had to buy full box - 100s of them different types)

The plate is so large because it is used for activating induction sensor when nozzle presses component ( there is spring system - the bearing slides over motor shaft 2..5mm up)

Zdjęcie-0137.jpg (98.3 KiB) Viewed 44452 times

Zdjęcie-0138.jpg (105.78 KiB) Viewed 44452 times

Zdjęcie-0140.jpg (93.77 KiB) Viewed 44452 times

Zdjęcie-0142.jpg (91.27 KiB) Viewed 44452 times

[Spikee]

Update:
Samsung nozzles came in:

As you see there is no need to change the current "trigger" system as it works fine with this kind of holder.


3D print a adapter and attach it to the current system and you are done
Clearance height is more or less the same as with a stock needle.
The nozzles themselves are from hard steel and are really flat.

You can auto nozzle change via the same method as the juki's

[mrandt]

Hi Spikee,

thanks for sharing the pictures - your nozzles look good.

You can auto nozzle change via the same method as the juki's

I believe the nozzle coupler for Juki is more complex. It has a ball lock similar to those you find in compressed air connectors. You have to push the collar of the coupler upwards to release the nozzle - Karl seems to use a angled surface to achieve this while sliding the nozzle into the depot.

Samsung seems to only use a pair of O-rings and rely on friction to keep nozzle in place - so changing means moving nozzle into some kind of tool depot holder and retract coupler upwards - and vice versa.

3D print a adapter and attach it to the current system and you are done

If one keeps the spring mechanism of Z-axis as it is, people could also use a fix coupler (no springy mechanism needed) instead of the one you have bought.

It looks easy enough to make yourself from brass, if you have a lathe or know someone who does. That way it could be made to perfectly fit the steel tube (just add inner / outer thread) - could be a good retrofit option.

I think I will still try using a hollow shaft stepper and Juki nozzles - will let you know how that turns out.

Regards
Malte