$4K Chinese Pick'n'Place machine

[EEVblog]

The speed is very impressive!
http://item.taobao.com/item.htm?id=19322348120

Dave.

[mikeselectricstuff]

2 heads seem a bit overkill but obviously they were going for high speed, however on a small unit I'd have thought things like feeder flexibility would be more important.

Hard to tell for sure but looks like the number of 12 & 16mm feeders lanes is limited and fixed - if so that could be a big deal.

Also not clear if it has vision or not - if not there are some parts that won't be very  useable - the vid showing a qfp seems to pick it from a tapered bin not the original tape

There are some apparently cheaper options shown but not clear what the differences are  - I'd assume there's a single-head version. Also seems like they are selling heads seperately.

From one of the vids on their site it looks like it's standalone with internal PC (winCE?) 

However it's by far the most plausible  cheap P&P I've seen to date.

Just hope the software doesn't suck as much ass as the avarage Chinese product, and someone translates it. Could be a good business opportunity for someone to handle it for the English market.
 

[48X24X48X]

Watched one of the videos (loading is super slow) on how to use the LCD interface. Although it is in Chinese, it seems to be intuitive (I don't read Chinese!). They didn't mention anything on using tray (probably loaded at the back?). But, this is awesome for small assembly and will definitely make lifer easier!

[StubbornGreek]

I'm having a hard time understanding how they can price it this cheap, yes even for China. For the money, I can't even bring myself to criticize it..its just (potentially) awesomesauce with sprinkles of wonderful.

These will either crash and burn quickly enough or they'll become quite popular, IMHO.

Nice find, Dave.

[EEVblog]

However it's by far the most plausible  cheap P&P I've seen to date.

I agree.
As I've said before, if you have your own machine with X capability, then those limitations simply become part of your design considerations.

Dave.

[EEVblog]

Even if it's not very good at QFP's or other large devices, and your designs had a zillion small parts and passives, then it could still be very useful in doing most of the work.
Say you had a board with a hundred small parts and passives, and one big QFP micro or something. You could still place the micro my hand, no biggie.

Dave.

[BravoV]

Also not clear if it has vision or not - if not there are some parts that won't be very  useable - the vid showing a qfp seems to pick it from a tapered bin not the original tape

At the last video on led board with circular pattern, it looks like the head turned on the led for lighting when started, possibly for the camera ?

If you watched closely at about 00:08, the head moved quite fast on the led's with vertical & horizontal position in the beginning at constant regular rhythm, but once it started on the diagonal position, the head stopped above the board for split seconds sort of processing something before proceeding to place the diagonally position led, camera in action ? 

[Psi]

Someone here says there's no vision system

http://dangerousprototypes.com/forum/viewtopic.php?f=2&t=4903&p=47444

[mikeselectricstuff]

If you watched closely at about 00:08, the head moved quite fast on the led's with vertical & horizontal position in the beginning at constant regular rhythm, but once it started on the diagonal position, the head stopped above the board for split seconds sort of processing something before proceeding to place the diagonally position led, camera in action ?[/quote]

could just be the time to rotate
 

[BravoV]

With that price tag, is it realistic to have a visions ?

[mikeselectricstuff]

With that price tag, is it realistic to have a visions ?

Yes - vision costs peanuts in hardware terms, and there are free software libraries like openCV.
P&P vision is a pretty easy task as you can have a very well-controlled and optimally lit field of view with nothing apart from the part visible, so no significant work on finding the area of interest.

[BravoV]

Mike, talking about the hardware, is that includes the cpu/mcu to run the heavy recognition tasks ? Say like equivalent to powerful tablet's arm cpu ? or vision needs quite some serious processing power like running it on a PC externally ?

[Psi]

It doesn't need much cpu power.
But obviously you want to do it fast.

[mikeselectricstuff]

It doesn't need much cpu power.
But obviously you want to do it fast.

Not hugely fast in processing terms though, as you have the travel time from camera to place location (less rotation time) to do it. A few hundred milliseconds would be fine.

[mikeselectricstuff]

More info here :

[Zad]

My first thought was; Phew that's cheap. My last thought was; Hmm, that's expensive.

Like a lot of Chinese-ified stuff, it looks pretty good and it seems quite fast and flexible, then you actually use it and find that there are insurmountable problems that turn it into another dust collector. The lack of vision seems a bit of a killer and the touch screen looks annoying (aaaargh Comic Sans!) and if you have to mess around with the setup calibration for each board then I can see it just not being used. Add a $50 webcam and macro lens, delete the cheesy colour touch screen and drive it directly from a PC via USB instead.

As it already has a pneumatic pump, it should be possible to add an automated solder dispensing facility.

[StubbornGreek]

This really seemed exciting at first (coming from someone with absolutely no hands-on with an actual pick-and-place) but I can see the shortcomings now - still, a good concept/start...?

[mikeselectricstuff]

If they've not even bothered to put in switching for the pump(s?), what else have they skimped on... I wonder how long the mechanics will last.
And WTF is up with it needing to be told how to use both heads in the data....?

Doubt it will be useful for anything smaller than 0805s