Upgraded mechanics

[Spikee]

To improve the speed, accuracy and repeat-ability it is inevitable that the mechanics of the lite-placer have to be upgraded.

In the recent years accurate linear rails and little to zero backslash ball-screws have been pretty cheap when sourced from china.

I think that a mechanical upgrade of 500-1000 euro will benefit the projects future greatly. Servo's or steppers with feedback are not needed unless super high speed is a priority.


I have ball-screws and heavy duty linear rails on my cnc machine and it is awesome. 6 meters / minute is doable with a moving table and a 10kg cnc spindle on it.



In the cnc world the majority of users buy from: linearmotionbearings2008 on ebay. He does custom machining etc for a low price. The quality is great.

I'm searching for the cad file of the liteplacer now , maybe I can make a drawing in cad of my idea.

[mrandt]

To improve the speed, accuracy and repeat-ability it is inevitable that the mechanics of the lite-placer have to be upgraded.

Do you have any proof for that claim? Results from tests you have run on your LitePlacer?

Of course one could always build a faster and more accurate machine. To achieve that, components like the one you mentioned would probably be needed.

But that is not the point. The use case for LitePlacer is prototyping and maybe very small production runs - thus it should be cheap and easy to build. Adding linear bearings and spindles would likely almost double the kit price - especially for larger work areas.

Different from CNC, the Gantry only carries little mass and there are next to no forces to overcome in X or Y direction. The belts have little backlash when installed properly and I have not had a single situation in which the steppers missed a step. Also, the TinyG firmware takes care of smooth acceleartion and decelaration, so the stress due to mass inertia is further reduced.

For my purposes (and I guess those of many existing users) the current mechanics with timing belts, steppers and makerslides for the linear movements work good enough. I can place 0402 and chips with 0.4mm pitch and do not intend to go smaller anytime soon.

What is your use case? Maybe LitePlacer is just not the right tool? On the other hand, the nicest thing about open hardware is that you can base your own design on what's already there and build something great yourself. If you do, please share your experience!

If I were to further improve my machine, I would probably focus on the head, nozzle and feeders first - just my two cents

[Spikee]

I have a liteplacer, it is ok but needs some more development to be really useful.
It is and will be used for assembling prototypes. The extra 500-1000 euro would not be an issue.

I have tested my cnc machine at high speeds and the result is very high repeat-ability of <0.01mm.
The idea is that the mechanical are doing 90% of the accuracy work , the vision doing the last 10% .

Medium to high end pnp machines also use these kinds of rails and ball-screws.

The build might even end up being cheaper than the current lite-placer mechanics (as it needs lots of parts and lots of assembly work).

This is not really aimed at the hobby maker community but more for the semi-pro users. And yes I'll just do my own thing and post the results here.

For the semi pro world there is not much choise:
go pro and get the M10V and spend 20000 euro , go the china route and get for ~10000 euro.
The china route can work out ... or not and when it is broken in 3 years you are probably fucked.

Visionbot is aiming at a ~3000-4000 euro price-point which is ok. But is as far as i can see not open source and the mechanics are kinda fiddly.
This seems a nice middle way which can be done for about 1000-1500 euro.

[mrandt]

Since we're talking other machines:

I think we all agree that if we wanted to buy and could afford 20k+€ professional machines (and that is most basic without feeders and accessoires), we would probably not be here.

I keep watching the low cost / maker / DIY space and there are a few other interesting machines:

- VisionBot IMHO is a bit overrated. I mean, congrats to the young guy who made it work. But afterall, it is not much different from LitePlacer or some DIY projects documented on the web - except for the steeper price tag.
Mechanics use different linear bearings but are still belt driven. Also uses USB cameras and a single nozzle. I guess accuracy would be about the same.
I had a glimpse at the software and did not like it much - looks very playful and cluttered to me (might be based on personal taste...).
Also, it is all closed source, so extensions hard or impossible to do - this dismisses it for me. But see for yourself:
http://visionbot.net/

- There is a bunch of German guys who demoed a nice looking P&P at maker faires in Germany last year. Software screenshots look compelling, hardware seems well engineered.
Not open source either and not available yet - also no pricelist has been published. They said the machine kit would sell around 2k€ but who knows...
Check it out here:
http://www.varioplace.com/

- "Redfrog PnP" features impressive woodwork and some nice circuitry, but I have not found out which software to use with it - also rather on the expensive side
https://www.buildyourcnc.com/PickandPla ... dFrog.aspx

- FirePick Delta takes a different approach - a very interesting delta robot that is cheaper to make and uses less space than a traditional cartesian machine. It is intended to be a multi-purpose machine with exchangable tool heads which could do anything from 3D-printing, to P&P, solder paste dispensing or decorating cakes...
I have a Beta build but to be honest, pick & place lacks semi automatic feeders and a nicely integrated software at this point. Might be interesting to watch, especially if you only place a few components and like to get 3D printer functionality for little extra money. Open source + hardware.
http://delta.firepick.org/

I would also like to point out OpenPNP, which is intended to be a machine independent open-source PnP software. As it is written in Java, it will run cross-platform, even on your RasPi 2 (or other embedded). Is anyone running LitePlacer with that software?
http://openpnp.org/

In the DIY scene there are some impressive builds - in no particular order:
http://briandorey.com/post/DIY-Pick-and ... plete.aspx
https://www.vbesmens.de/de/bestueckungsautomat.html
https://amesberger.wordpress.com/2013/0 ... e-machine/
https://github.com/openpnp/openpnp-openbuilds

I hope I have not missed something cool?

Long story short, I do honestly believe that Juha delivers the best package at the moment - and I don't get commission

His machine kits are readily available, resonably priced and ship instantly.

Hardware is easy to build, nice and stable. You still have to put some thoughts and work into feeders / trays and there are no automatic feeders at this point - but for prototypes and small panels that is not a killer.

The software may not be perfect and you are forced to use Windows - but it works without much tinkering.

Most important, everything from hardware to software is open-source - so you can enhance and change whatever you like.

If I could go back in time and revise my decision, I would get a LitePlacer again - see this as a testimony from a happy LitePlacer user

[JuKu]

Thank you, nice summary.

>Is anyone running LitePlacer with <openpnp> software?

Not that I know of. Openpnp does support TinyG, so it should be possible to adapt it to LitePlacer hardware relatively easily.

[mawa]

Visionbot is aiming at a ~3000-4000 euro price-point which is ok. But is as far as i can see not open source and the mechanics are kinda fiddly.
This seems a nice middle way which can be done for about 1000-1500 euro.

That guy is a young science project competition winner from Romania. And he expresses that fact more than the qualities of his machine.
My impression his software is the same as Maltes and no open source.

Did you see what "professional" parts he is using? A camping gas cartridge for his vacuum and.. I could continue but see for yourself.

Juhas Solution is in a currently unbeaten price performance range with excellent parts for a fairly large work area.

It offers an unprecedented and as far as I experienced nice working solution of picking up the parts from cut tape strips without any special hardware or feeders. It is intended for prototypes and small volume series.

As Malte mentioned there are some minor enhancements on the wish list and I am optimistic that the feedback from us users will inspire Juhu to look into realizing some of them.
My priority: nozzle, nozzle limit switch (IMO 2-3N pressure on the part on first Pick/place is at the documented limit), nozzle changer, individually mountable tape guide trays ( a tray with a set of tape guides where you slide in the cut tape strips and the guide has a white or black background behind the tape holes)

[Spikee]

individually mountable tape guide trays ( a tray with a set of tape guides where you slide in the cut tape strips and the guide has a white or black background behind the tape holes)

There are a few on thingyvers you can get 3d printed in white or black.

[protovoltaics]

.

[Spikee]

I'm also a eevblogger so I have seen your machine before. I am also surprised to see the VarioPlace it looks great. But not open source would be a deal killer.
For the pricepoint the liteplacer is great. But it needs some more development to be more useful for the semi-pro market.

[WayOutWest]

I would also like to point out OpenPNP, which is intended to be a machine independent open-source PnP software. As it is written in Java, it will run cross-platform, even on your RasPi 2 (or other embedded). Is anyone running LitePlacer with that software?

I have been grabbing large chunks of OpenPNP's codebase. I might end up using it wholesale eventually.

The fact that it is written in a non-proprietary programming language was the overwhelming deciding factor for me.

OpenPNP has a nifty mini-language for writing OpenCV pipelines called firesight (originally created for the firepick, but not specific to it). Very neat. Once your pipeline gets to be more complex than two or three stages even the fanciest GUI is no longer appropriate.