Needle High Calabration

[dave]

Not sure if this is a Request or a question.

I have designed in Juki Nozzles (Plastic) attached direct to a hollow stepper motor.

There is no bloody switch as a after thought to zero the Z axis like on the liteplacer

Can someone explain the needle calibration, I still dont have Tiny G so not played much with software.

For me knowing the tool length and size of picker head, I then motor the X Axis in the up direction until it touches the upper switch and add a offset rather than zero. I think the Juki Nozzle has around 10mm of spring to remove any error so it wont crash if it like a few mm out. I will be using inductive hall effect switches.

Can I do the above, can a Z zero offset value be added to the calibrate routines.

[mrandt]

I have designed in Juki Nozzles (Plastic) attached direct to a hollow stepper motor.

Please share photos / drawings

Can someone explain the needle calibration, I still dont have Tiny G so not played much with software.

I assume you are only interested in Z-level calibrations?

Needle "wobble" and camera-needle distance calibrations are a different story.

There are multiple calibrations for Z-level:

a) Needle length / PCB height calibration to figure out Z-travel to PCB surface -> this is needed to have nozzle tip at correct height when moving above upwards camera
b) Backoff travel (this is needed as the switch does not trigger right away when needle touches something)
c) Pickup Z-level -> when picking up first part of a given type, the Z-level is determined and stored per component type
d) Placement Z-level -> similar, when placing parts Z-level is also determined and stored

Basically, for each calibration the needle (Z axis) is slowly lowered, touches surface or part and then the spring will be contracted. At some point, the switch "clicks" and movement stops. As "backoff" has been calibrated before, the software now has exact measurement to surface and can use those derived values for future placement.

So you definetly need:
- a spring that allows to touch a surface without breaking anything - the lower the force the better
- a sensor or switch that allows you to determine that the nozzle "hit" something and spring was compressed - the earlier the better

I will be using inductive hall effect switches.

Sounds interesting. How will you mount these to the nozzle? Can you share a drawing?

Can I do the above, can a Z zero offset value be added to the calibrate routines.

Yes, latest software release by Juha has options to control Z offsets and backoff.

[JuKu]

The needle height calibration figures out the following:
- PCB surface, used to set needle wobble measurement height (and trade faire demo sequence)
- switch backoff, used together with placement depth user settable value in placement and pickup

The switch is used to measure pickup and placement z value. The measured values are put in to the tapes data grid. The value used is the measured value - switch backoff + placement depth.

All of the values are user settable - but a lot of work, if your machine does not measure these for you.

[dave]

My head is a bit like the one from Brian Dory Project with Deltron E3 Rails I found second hand on ebay. but will also work with Hiwin 9mm mini rails

To attach the juki I will use this. Although if you can find a 5-7mm solid coupler I think this will work. but you cannot order it from ebay etc. if you have a nice lathe you can save some money.
http://www.robotdigg.com/product/517/No ... ft+Stepper
http://liteplacer.com/phpBB/download/file.php?id=229
The motor is housed in 3"x 2" x 1/4" box section. Imperial measurement sold on ebay from a UK vendor. I will mill it out to make it lighter but its shown as normal box in the image.

The Yellow mark is the location of the inductive sensor which is 8mm in size. the red mark is the location of a standard V4 micro switch that is triggered from a screw in the back of the box section.

The bit marked in Purple I copied from Chinese machines. Its a rotary coupling for the airline. I am now goner change to the right angle version and have the air and wires coming out the front of the housing. but I do all this detailed stuff in CAMBAM on my CNC mill.

https://stevenengineering.com/prod_serv ... /KS-KX.pdf

Finally the crome looking bit is the andonstar camera. with everything mounted the unit is 75mm wide. I will make massive weight reductions because its made from 10mm aluminum. This depth should allow enough to sink the cap head bolts etc

picker.JPG (21.78 KiB) Viewed 5937 times

[mrandt]

Thanks for sharing!

To me, it also looks a bit similar to Jason von Nieda's OpenPnP Head - although you seem to use a vertical lead screw for Z-movement.

I don't get how you will detect that the nozzle hit something. If I understand your drawing correctly, the inductive sensor would only detect upper limit. The micro switch could detect lower limit, but in practice that should never be reached as needle would always hit something first... And that is what you need to sense.

IMHO Juha's design is very smart - he basically combined two linear bearings along Z-axis (maker slide + steel tube). One is motor driven, the other spring loaded.

To build and operate your machine in a similar fashion, you could mount the hollow shaft stepper into some sort of cage which should also move linearly along Z-axis. That cage would move up a bit once the nozzle hits a solid surface so that some switch or sensor could detect it. I think, a simple optical switch (like opto endstops often used in RepRap printers) would do.

If you did it that way, a spring loaded nozzle (like Juki) would not make much sense, would it? You could just use a stiff nozzle like the ones Fuji uses:

http://de.aliexpress.com/item/Free-ship ... eb201560_9

Another nice option could be the custom milled nozzles used by FirePickDelta - completely open source, CAD model is available so one can make their own:

mrandt_FPD_nozzle_01.jpg (25.31 KiB) Viewed 5926 times

mrandt_FPD_nozzle_02.jpg (34.74 KiB) Viewed 5926 times

These do not use a magnetic but simply a mechanical link - I think it should still be possible to implement auto changing.

I also use a rotary joint to connect the vacuum tube to the steel tube on my LitePlacer - I cut M5 threads to each end so I could connect both the Luer lock and joint. Works like a charm:

mrandt_vertical_tube.jpg (41.24 KiB) Viewed 5926 times

I later also switched to a 90° elbow joint - I am using part RTCNL by Misumi:
http://us.misumi-ec.com/vona2/detail/11 ... s2products

I see three benefits in your design:

1) lower moving mass in X and Z direction, for Y direction you still have to move the heavy Gantry - not so sure you will speed up significantly

2) probably less "wobble" in A axis; hollow shaft motor should have accurate bearings - might be able to work without needle wobble calibration which would be great, especially when changing nozzles often

3) potentially less force needed to detect Z-axis limit when nozzle hits component - currently it requires about 3N, maybe a bit too much for some electrolytic caps and delicate ICs

Have you considered leaving space for a second head? Would require many enhancements to software but might be a nice way to speed up placement.

All in all I think this is interresting.

I would appreciate if you kept the forum posted about your progress

[dave]

I never realized it was important to detect you have hit something each pick operation. Thinking now there is nothing in the software to input component height. I have experience in MACH 3 CNC milling and assumed it was not important to know as the height of everything was known. You just calibrate the needle then it dont crash and you have a few mm on the nozzle if shit happens.

I was hoping that because the tool is always the same length down calibration was not important. I was just goner set the down micro switch so most the head can travel was to the table and so it dont drive to the end stops of rail also. the software I though would stop it crashing and compensate for component height but maybe thats not the case

Dose the solenoid release vacuum when you hit something. or is solenoid software controlled.

I almost designed this with gravity belt and a spring to set the down force, maybe the open PNP did this I will check now. Check VB project though he dose for sure, It wont be to hard to implement I just change from screw to belt that only drives in up direction. I can make it lighter again like this, The reason I did not use this option at first was because I was worryed the control would be a bit sloppy in the down. but I can always add weights and springs to control that. Now you mention denting Electrolytic caps I really consider the VB project belt drive much greater.

This guy talks abit about his experience. I notice he dose not use a spring in the up direction. So long as the weight of the head is not enough to crush a cap the spring might not be needed. but it will be a good idea to implement the up spring in the design in advance, I notice a number of Chinese machines also have a up spring and gravity head.

https://www.youtube.com/watch?v=jVllrAoW7lU

I have ordered enough stuff to build 2 heads. I could put another on the rear or the camera and the other head next to. but I am keen to put a feeder advance arm / solenoid and push Fuji CP6 feeders from the rear before putting 2 heads.

In part the only reason why I dont have the Feeders and Dual heads is because the software dont support it. Machining it and implementing it will not be a problem for us. If Someone wants to make the software I am crazy enough to build it, Maybe I will have some time to make software mods also but now I have to much on my mind haha.

I will update you with another drawing soon. Thanks for your input, I think seeing your post has made me reconsider the design a little

[JuKu]

> Dose the solenoid release vacuum when you hit something. or is solenoid software controlled.

It is under software control.

[dave]

How about the detect of component down.

Can the software run without knowing the needle has touched down.

IE motor down all the way to top of the PCB. and allow spring and gravy to compensate for component height without to much needle pressure hopefully.

[dave]

I just noticed pickup and place Z variable in the config. I guess this is component height.

Surely the spring in the Juki Nozzle will be enough to stop crashing components.

Or is a gravity and spring system goner be better.

[mrandt]

I think you *could* make the software run without a sensor to detect the needle touched something as you can enter most Z-levels as a configuration value - and if not, might be easy to add in source code.

Question is, why would you?

There are quite a few calibrations that use that switch / sensor:

- Needle / nozzle length
- Z level of table surface
- PCB thickness
- Pickup + placement height for each individual component

I would not want to measure all of these and put it into configuration. It is much easier to let the machine figure out at run time - this is what the switch is used for.

For those P&P machines with fully automated reel feeders I assume Z-level is less of a problem. At least pickup height is always the same. For placement, you would still need to measure each and every component if you drive the Z-axis with a lead screw.

If you used gravity and belt, it would be a different story - as you basically could not drive Z too far. However I am not sure if I really like that approach... Although I don't have reasonable arguments aggainst it, it somehow feels like a hack