DIY PnP machine with SSCNET servos

[Dmitry Yurtaev]

been lurking for a while, but finally had my machine doing something to show off. :)

spent too much time with the hardware that i couldn't resist to spend even more playing with the software to control it. :) so now it's under linuxcnc control with a bunch of g-code routines and a small python script doing the CV.

as i'm using digital bus connected mitsubishi servo drives some features come for free - like torque limiting. using it for rigid stops homing, component placement force setting and z-probing for PCB warpage. played with the later a bit - works great, easily feels silkscreen paint thickness on the pcb without any touch trigger sensor needed.

wasn't sure about tiny solenoid valves i used (SMC114A, 0.22mm2 orifice), but no problem so far, seem to be quick enough.

nozzle steppers need some more work: angular precision is awful. trying to avoid re-checking position after initial vision correction. but not very successful so far. gonna try different stepper drivers instead of DRV8825.

next are automatic feeders, have some ideas to test, but no finalized design yet...

a couple of videos. accelerations are set conservatively at 1g for X and 0.5g for Y. need to better fix the shelf the machine is sitting on to the wall... notice the  trajectory blending during moves. and it seems to help a lot to reduce jerking...

https://youtu.be/v-mSEl6G7rg

https://youtu.be/AVbiUh-vBU8

[bert shivaan]

Beautiful!!!

[Mike Menci]

Nice, which nozzles are you using and are A1 & A2 as well servos or steppers?
Thanks for shearing
Mike

[Sandra Carroll]

That is one slick machine.  Great work.

Can’t wait to see it once your have your issues worked out.

I would be curious to know how good is it’s placement without vision?

 

Sandra

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/13ff28ba-3c91-40e3-afa0-c10e00867436%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Dmitry Yurtaev]

Mike Menci wrote:

> Nice, which nozzles are you using and are A1 & A2 as well servos or steppers? 

nozzles are home made anodized aluminium. tips on those pictured are dimensioned after juki 503, 505, 508. also plan to make a 502, but the tip will probably be too fiddly to drill and too flimsy - thinking about gluing a piece of suitable hypodermic needle and grinding the end flat.

A1 A1 are common 20mm thru-hole chineese steppers most folks are using. i'd love to have servos instead if the machine was a little larger... 10W Mitsubishi HG-AK0136 would be nice.

Sandra Carroll wrote:

> I would be curious to know how good is it’s placement without vision?

i'm not sure how it would be with automatic feeders, maybe parts will somehow orient themselves. but looking at the tape strips - parts are pretty chaotic in pockets. so it can't be any better than that... the nozzle positioning accuracy should not be a limiting factor.

when setting up strip feeders with top vision i try to detect centers of tape pockets, whenever possible:

https://youtu.be/MmCc2ojoHds

[Florian Chende]

Hi Dmitry,

Beautiful work, congratulations.

The pick and place speed is quite good, as expected for servos and ball screw. But the vision looks to me that is's a bit slow. Why is that?

I am curious about some build details if you don't mind sharing them:

-what ballscrew are you using (size, lead, brand)

-some details for AC servos (power, torque, encoder resolution)

-are your controllers set for max torque allowed by motors, is this the full speed of the system?

Thank you.

[Mike Menci]

Solution might be making body and using only Juki tips: https://www.google.com/search?q=juki+nozzles+tips+aliexpress&safe=active&rlz=1CDGOYI_enID614ID615&hl=en-US&prmd=ivsn&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjG7OSz-ILjAhWLzqQKHc3aAW4Q_AUoAXoECA4QAQ&biw=375&bih=621&dpr=3#imgrc=XzX3L_T_oPH9FM&imgdii=aE0BGIZnKxLFIM

Mike

[Dmitry Yurtaev]

thanks!

the machine was mostly built around components i had on hand. namely the THK d15/20mm pitch ballscrew, a couple of matching length 15mm wide linear rails (unmarked), a cast aluminium base and a few others.

yep, the vision speed leaves a lot to be desired. it was quickly hacked together and debugged with the goal to make it working most of the time. there seem to be at least 200ms delay in the USB/V4L2/OpenCV pipeline and my CV routine samples 8 frames @30fps to average the measurements. that's another 240ms. will surely revisit it.

the servos are mitsubishi MR-J3-B series, with HF-KP13, HF-KP23 and HF-KP053 motors on X, Y, Z respectively. they are 100W, 200W and 50W 3000rpm, 

with 18 bit absolute encoders. the X axis also has a 1um resolution Renishaw RGH24X linear encoder plugged into X axis MR-J3-10BS servo amplifier. the drives are in position control mode with 0.44ms servo period, via my home brewed SSCNET III interface PCI card.

the speed is nowhere near the maximum available from the servos. here's the video of a dry run of the same job with accelerations cranked up somewhat and all dwells and slowdowns disabled. still not at the max. but i'm afraid that the overall machine rigidity and belt drives won't allow to push it to the limit, even when i fix the machine base...

https://youtu.be/hb4kSznglo0

/dmitry

[Florian Chende]

I love the last video, the way it jumps off the table :)

I also have a machine in design, but it's all on paper now. 400W panasonic servos and TBI 25(fi)/50(mm) lead. It will be interesting.

[SMdude]

Dude! What a beast!
I'd like to see how well it hangs onto components working at full speed like that!

Well done! And you definitely need autofeeders...

Mick

[Florian Chende]

Dmitry,

That plate onto which you built everything, is it a full plate or is it hollow inside? I see 3-4 cm of thickness there.

Thanks.

[Dmitry Yurtaev]

hollow. it is a cast tray, with 7-8mm thick top/sides. there was a the
photo of it from underside in my initial message...

/dmitry

> --
> You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/O1j_0x9tP6U/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.

> To post to this group, send email to ope...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/dba89e74-2520-4e19-85db-33ab18f78790%40googlegroups.com.

[Florian Chende]

Oh, you're right, sorry. I missed it. It looked a bit odd to jump like that if it was a full plate.

[John deGlavina]

I just had to check to make sure the video wasn't set to x16 speed. That is incredible work for a home made pnp! Also, one of those motors costs more than my entire machine. You just had them on hand? haha  You should name that thing машина

[Dmitry Yurtaev]

hehe, thanks :)

i've been playing with those mitsubishi servos for years, so yeah, have collected a number of them. true, not cheap when purchased new, but that's what ebay is for... :) e.g. i got a dozen of MR-J3-10B drives locally from a dismantled DVD production line for about $20 each. HF-KP053 motors, $50 for a couple from ebay. i think i spent about $700 on a stuff purchased specifically for the machine, including a PC, some spares and a number things which end not being used...

/dmitry

[Mike Menci]

I see that this servos & motors you use are quite resonably priced on eBay, I purchsed Yaskawa drives and motors and I plan to change my stepper with servos for X and Y only. Software will still remain by Open PnP and Smoothieboard i.e use of STEP and Direction pons before the stepper driver... I never used Servos before all I did up to now is testing 100w servo with Yaskawa drive - jog from controler. Setting up config file on Smoothie might be chalanging with Yaskawa?
any recomendations ?

[Dmitry Yurtaev]

important: before buying any servo drive and motor find and read the manual. make sure the drive supports a control method you need (position control with direction/step pulse train), compatible with the motor, power supply voltage/number of phases is what you have in your area. all letters and numbers in model designation on the drive/motor nameplates do matter. there're many customized drives on ebay which don't have any public documentation and may not work as standard production units.

yaskawa has excellent documentation, i see not problem hooking it up to whatever controller you want. beware, most industrial drives have optoisolated pulse train inputs intended to be driven by either a differential transmitter or open collector output with 24VDC. but usually can be connected to a 5V TTL outputs via suitable resistor. all other GP inputs most likely will need 24VDC.

check the availability of all required connectors/terminal blocks and cables - those can be quite expensive.

wiring and configuration is usually well described in the manual as a step-by-step procedure - just follow it. 

one step required to get the performance out of the drive is PID loop tuning. manufacturer's configuration software may help here with some kind of autotuning function. the machine mechanics is a huge deal in performance: the more rigid is the transmission the better performance you can achieve. preloaded ballscrews with a bellow coupling directly to the motor shaft are good. long sloppy rubber belts are bad :)

/dmitry

[Mike Menci]

Thanks- yes all points are important and well noted. 

There is a web of hooking up Smoothie with Omron - http://blog.bouni.de/2015/driving-a-yaskawa-sigma-ii-ac-servo-with-a-smoothieboard.html which I quite near to Yaskawa.... 

Thanks

Mike

[Dmitry Yurtaev]

yep, if STEP/DIR outputs are 3.3V they can be hooked directly and should give about 10ma thru the optocoupler LED. but i'd check the datasheet to make sure if the micro on the smoothieboard is rated to supply 10mA*2*3axes=60mA thru those pins...

omron drives are re-badged yaskawas

/dmitry

[Mert Capkin]

Nice looking machine! Congrats on the build! I was also thinking about building one with ball screws instead of belts. I saw that most machines are with belts, do you know why?

Op maandag 24 juni 2019 20:16:19 UTC+2 schreef Dmitry Yurtaev:

[Mike M.]

If you are looking for precision ballscrew is the solution - specialy with servos (belts can be but as short as possible!)

Mike