FYI: Citrus CNC / CrowdSupply OpenPnP based SimplePnP

[John Plocher]

Just saw Gino's well under-$1000 SimplePnP on CrowdSupply:  

https://www.crowdsupply.com/citrus-cnc/simplepnp 

Looks interesting as a low end starting point, with single or dual heads, strip feeders only, top and bottom camera versions....  Design files on github (https://github.com/ottoragam/SimplePnP)

Features & Specifications (from campaign materials)

• Performance
  • X Travel: 300 mm
  • Y Travel: 300 mm
  • Z Travel: 23 mm
  • Components per Hour: 750 with vision assist
  • Component Size: As small as 0402 (1005 metric)
  • Vacuum Heads: Single vacuum head (base model), dual head optional
  • Vision: Top-vision and optional bottom-vision
  • Camera Specs: 1280x720 @ 30 fps
• Requirements
  • Power Consumption: 60 W With a 12 V power supply
  • PC Interface: Three USB ports (four with bottom-vision)
  • Footprint: Bench mountable (500 mm × 450 mm × 200 mm)
  • Weight: 4 kg
• Components
  • Stepper Drivers: TMC2209
  • Motion Control CPU: Microchip Atmega328p running GRBL firmware
  • Nozzle System: Juki 5xx with tool changer capable holder
  • PC Control Software: OpenPnP

  -John

[Airhead Bit]

Looks like they will reach their goal, $200 left to reach it in 40 days. I'm wondering what they need 4 USB ports for?

[John Plocher]

On Thu, Dec 19, 2019 at 6:24 PM Airhead Bit <airhe...@gmail.com> wrote:

Looks like they will reach their goal, $200 left to reach it in 40 days. I'm wondering what they need 4 USB ports for?

2x cameras (top & bottom) and 2x GRBL controllers (1 for head, 1 for table)

  -John

 

[Airhead Bit]

Ouch! I didn't realize they simply added the TMC driver to a two axis GRBL, I thought since they used the TMC they had designed their own controller... Like OpenPlacer, Mega 2560 stripped with three different drivers for the motors.

[Gino Magarotto]

Howdy, creator here.

I separated the head and table controllers to simplify wiring (cables for the Z home switch and 3 steppers are replaced with tx, rx and power). The controller will be mounted in the head, but I still haven't shown the configuration because I'm currently working on the board layout.

The table controller will also be a custom design. Also, both atmega328p microcontrollers will be user replaceable, to allow upgrading the board for very little money to a more powerful 32-bit CPU in the future (I plan to use a PSoC, need work the GRBL port). Every design file will be made available on the repo.

Gino

[Wayne Black]

If you do the Psoc port of GRBL be very careful of the step IO pin placement. My homebrew CNC controller w custom psoc pcb worked just fine when PWM configured, but falls flat on its face using same IO for GPIO 'Pulsed' as in GRBL psoc port. I dont recall all the specifics, so bust out your scope and test pulse timing before you layout the PCB to psoc. FWIW I love the psoc :)

--
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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/15fd7996-edfb-488a-89ba-a864a530ad86%40googlegroups.com.

--

Wayne Black

Owner

Black Box Embedded, LLC

black...@blackboxembedded.com

1.831.682.4964

[Gino Magarotto]

Thanks for the heads up. I plan to add support for quadrature linear scales via the programmable logic to the GRBL port. Already did a similar project with a PSoC4, so it shouldn't be too problematic to merge the two. And I love them too, the hybrid microcontroller/PLD stuff is a lot of fun!

[Wayne Black]

I use the psoc4 in a consumer device and know it well, kinda limited in io vs the 5lp. Be warned that the 059 Kits are dirt cheap, but when you go to purchase just the CPU for end design theyre twice the cost of the kit :(. I just started using psoc6 as modules for internal testing jigs and the dual core is pretty sweet.

On Fri, Dec 20, 2019, 4:34 PM Gino Magarotto <otto...@gmail.com> wrote:

Thanks for the heads up. I plan to add support for quadrature linear scales via the programmable logic to the GRBL port. Already did a similar project with a PSoC4, so it shouldn't be too problematic to merge the two. And I love them too, the hybrid microcontroller/PLD stuff is a lot of fun!

--

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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/29d2307b-19df-436d-aa0b-c8f1d1749bd7%40googlegroups.com.

[Gino Magarotto]

Yup. Been using the C8CY5288, very good price/perf for the capabilities vs something like an ATmega2560. I still haven't got to play with the PSoC6. For some reason Cypress (or should I say Infineon) is too focused on their IoT applications, and they kinda nerfed them with the lowish amount of logic blocks. The dual core ones are very interesting, for sure.

[Alexander Goldstone]

I have been waiting for a design / kit that looks accessible to someone that has no mechanical experience...

I don't know enough to identify whether this is something I could use for small scale production (and maybe evolve over time) or of it is likely to be more trouble than it is worth...

Anyone more knowledgeable able to look at the photos / video and advise what they think based on experience of other DIY or low cost machines?

[Simon Merrett]

Hi Alexander, I am interested in the design too and Gino has kindly accepted my offer to help with the design / documents. I would say that the design looks like one of the simplest motion platforms you can get, from my experience of 3d printers and cnc routers. Because Gino is producing an open source design, there is a github repository which includes some documentation. You will be able to see it developing over the coming weeks and months. https://github.com/ottoragam/SimplePnP

[John Plocher]

Opinion:

if one were to compare with the 3D printer world, this looks more like a DIY reprap printer starter kit than a fully functional plug and go system, or, in the laser world, more a K40 than an epilog 🤓

Expect to spend as much time making this work as you would if you built it from scratch, just without the uncertainty of getting that first bit of “where do I start?” out of the way...

The fun begins when you start replacing and upgrading and expanding this starter system...

...as long as you don’t expect more from it than it really is.

  John

[Simon Merrett]

John, I think that's a fair opinion and you rightly identify the hackability / incremental upgrade potential of this design. Would you care to give your opinion on the Z travel/accuracy and angular resolution required? I asked about this in a separate thread and I won't cross post further but this project would really benefit from those who have other pnp systems letting us know what specs the head would need to achieve to be useful.

Simon

[John Plocher]

> Would you care to give your opinion on the Z travel/accuracy and angular resolution required? 

Z travel (I.e. distance between pcb and retracted head) is directly related to the max height part that can be placed:  

Height[part] = ~1/2 max Z height of head 

Thought experiment: if you are placing electrolytic capacitors that are 5mm tall, and your Z height is only 4mm, it will scrape off as soon as the head flys over the pcb...  if Z is 7mm, the flying part will hit other tall parts...

Getting lots of Z height is hard- both mechanically and in travel time.  Long strokes take longer, and every part pick and part placement needs to spend that time...

The usual trade off is to realize that most parts are short, and exceptions can easily be hand placed.

5mm to 10mm seems a common compromise.

As for rotating parts, solder surface tension fixes many alignment issues, but it doesn’t work well on badly aligned fine pitch ICs if the leads aren’t on the right pads.

You will want at least 270degrees of rotation, with 360 being much better.  (Think diodes that have polarized placement combined with random feeder pickup alignment)

As I haven’t seen specs in this area, I’m not sure what angular resolution is needed; many someone else can step up and give details?

  John

[Simon Merrett]

John, great points - thanks. My Juki nozzles are in the post so I don't know whether there's any give/elasticity in the Z direction on them. What would you say is a useful level of accuracy on the Z height - 1mm is probably not right but is 0.5mm? Or are we talking 0.2mm, 0.1mm or better?

[John Plocher]

The juki's have spring tips that compress on contact, so there is some leeway.  However, you don't want to "mash" the parts - they easily pop components out of feeder tapes, squish out the solder paste, under the component,  bounce the PCB board, etc - all badness.  If you simply plop Z down to PCB height to place a part, you will often end up "placing" an empty tip into a blob of solder paste and clog the tip...

You will want to be able to adjust the Z pickup and Z placement heights pretty accurately based on component height; again, the resolution and adjustment pitch will translate directly into increased cost/complexity as well as poor speed-vs-cost tradeoffs (i.e., you can get microprecision and repeatability, but either very slow & cheap or very fast & expensive...)  -  it may be that a resolution of "at least 1/2 the height of the smallest part you wish to place" works for your budget and speed needs...

Remember, SimplePnP is aimed at Good Enough rather than Perfect - you want to design away from naively stupid & broken without getting too far into the complex and expensive.  This philosophy should lead to a slow machine that always works as expected, on the basis that it is easier to throw money at faster than it is to fix stupid.  After all, that's why one can spend north of $1M on a machine that places 008004 parts (0.25 mm × 0.125 mm) reliably at hundreds of thousands per hour...

  -John

On Tue, Dec 24, 2019 at 9:24 AM Simon Merrett <smerr...@gmail.com> wrote:

John, great points - thanks. My Juki nozzles are in the post so I don't know whether there's any give/elasticity in the Z direction on them. What would you say is a useful level of accuracy on the Z height - 1mm is probably not right but is 0.5mm? Or are we talking 0.2mm, 0.1mm or better?

--
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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/39f704aa-f8bd-4f8d-86ba-d657138dc160%40googlegroups.com.

[Simon Merrett]

Yet again, John, great points - thanks. From a quick search, it seems that 0402 resistors are 0.35mm high generally. So half that will round to 0.2mm, which is a good starting point for Z accuracy.

The SimplePnP strategy should be to enable people to successfully and repeatedly populate boards, at the cost of speed, which is one of the reasons I initially asked for requirements for the Z travel and nozzle rotation in a different thread, as this isn't the core of SimplePnP.