Why CNC?

[saxonhawthorn]

Plans? What are they? I've never had any plans for anything! In 25 years of writing commercial software I never even once got a specification from a customer. They always wanted me to tell them what they wanted.

I hadn't thought of brass as a screen. Not a bad idea. Not sure if I can afford a laser powerful enough to punch it though. I'd have to etch it.

I envision picking up components from the reel with a vacuum head on a robot arm, then holding them over a camera with some vision recognition software to square them up (that's where the tiny worm & wheel come in) before plonking down on the pasted-up PCB.

I bought the large worm & wheel (£18) because - like Mount Everest - it was there. I didn't have any particular use for it, but I had a hunch I'd find one. When it arrived I was really delighted with the quality, and I'm sure it will be adequate for the job.

As regards Tribbles point about volume: yes, and I can indeed rattle out boards manually at a rate of knots if I have to. But now that I'm liberated from the shackles of somebody else's factory I'm planning new products which I intend will go well, and anyway somebody's got to give the Chinese some competition! Once you've got a production line set up of course, it's easy to switch from one product to another. You just load the appropriate product code, and away you go.

Ian

[Robin Hewitt]

I don't think you have to laser your own paste stencil any more than your assembler would, there's a ready made industry out there. First hit on Google...

http://www.smtstencil.co.uk/?gclid=C...FU0A4wodF1p6rA

Don't think you need a worm for the rotation, a dinky stepper will easily microstep anywhere you want, with no load mirostepping really works!

Is an elbow arm really the best solution? As soon as you locate over a camera you can't "teach" the arm and you are back to positioning by co-ordinate :naughty:

[saxonhawthorn]

Robin,

Actually, I thought about this in bed last night, and grinned when I realised the answer. CNC! All I need to make holes in the solder mask is a CNC drill. I'm fairly sure the CAD outputs a Gerber file that will either do the job as it stands, or will do so if I just fiddle a bit with the hole sizes. But thanks for the link to that mask manufacturer. I've made a note of them, just in case. The price is certainly right. As a general rule I like to keep as much as possible in-house, for both cost and quality reasons. But sometimes it can make sense to contract out.

And yes, it's true that I don't really have to use a robot arm. I could just as well do it by 2D XY positioning. But that's just a tad boring, and I have to have a challenge in it somewhere to keep up my interest. Besides, a robot arm will impress the hell out of the customers; and although it's irrational, in real life that's an important part of getting contracts. :) Then I can make vids like the one Tom posted!

Ian

[saxonhawthorn]

Robin,

About micro-steppers.... when I mentioned using an arm I was working on the assumption that the minimum angular resolution of a bog standard stepper would be 1.8 degrees, and to get anything better I'd need a worm. Was that assumption wrong?

Ian

[Robin Hewitt]

The perfect current for stepper motor coils would be two sine waves 90 degrees out of phase. Simple drivers approximate to it, pukka drivers try to create it. Intermediate points are a bit springy but rotating a component is a no load situation so tiny changes in angle are possible if your driver is willing and able

[irving2008]

The perfect current for stepper motor coils would be two sine waves 90 degrees out of phase. Simple drivers approximate to it, pukka drivers try to create it. Intermediate points are a bit springy but rotating a component is a no load situation so tiny changes in angle are possible if your driver is willing and able

For this requirement, where full rotation isnt required, i guess, but accurate partial rotation is, I wonder there would be mileage in a purpose built sinewave phased driver rather than a stepper driver as such? The load is small but not zero.. presumably you'd be rotating the suction head which would have a vacuum tube attached which would give some tangential load.

[saxonhawthorn]

Irving,

"presumably you'd be rotating the suction head which would have a vacuum tube attached"

Correct. And you are also correct that I am thinking in terms of designing my own driver board - mostly from sheer habit, but also because that way I end up with exactly what I want (or alternatively, have only myself to blame if I don't). I'm planning to spend time thinking through this project over Christmas (although I have the wife in bed with 'flu at the moment, and I'm therefore having to do all the running around town that she normally does) and because of family commitments little will actually happen in terms of either hardware or software until Q2 of 2010; but by then I expect to have thought through all the details and be ready to move quickly once I start in the Spring. I'm looking forward to it greatly now, because almost everything I do has a processor and a PCB in it somewhere, and to be able to knock those out quickly and easily will be really useful. I'll post pictures and circuits when I get there.

Ian

[irving2008]

Look forward to seeing the outcome of your deliberations....:idea:

I am mentally picturing a 'Wallace and Gromit' scenario...

Wallace sitting at his PC designing the board, and ceremoniously pressing a big green button, activatiing a robot arm (complete with shirt sleeve and oversized cartoon hand) which picks up a blank piece of PCB, places it in a CNC PCB router which cuts the board, then the arm picking it up and placing into the SMT placer, and then again into the solder oven, and finally (with singed fingers and a 'boing' sound effect) putting it on Wallace's desk!:heehee:

I'll get back to work now.....

[Robin Hewitt]

Correct. And you are also correct that I am thinking in terms of designing my own driver board - mostly from sheer habit, but also because that way I end up with exactly what I want (or alternatively, have only myself to blame if I don't).

You probably won't believe this but I think I have to mention that you may be making beginner's mistake. I won't be saying, "I told you so", but you can assume that :naughty::rofl:

Yes, you could gain vast extra control by making a seperate computer to interface Windows to your machine, but you would do well to stick with commercial stepper drivers and PSU.

Given the price of a humble MSD542 it really isn't worth the nightmare of trying to recreate it

[Tom]

Sounds great! Here are some plans.... (tee hee)

[ame="http://www.youtube.com/watch?v=YZ_xyh4ehNo&feature=related"]YouTube- Gigabyte Mainboard Factory[/ame]

I've been in a board manufacturer before, and this is the method that they used:
Screenprint solder paste (stainless screens), 3D vision system to check for paste registration (and volume), pick and place machine(including pick, scan to check orientation, rotate to correct, and place)(as in the video above), then another vision system to confirm, and finally the reflow oven. You could obviously skip the pre and post paste vision systems (what I saw was pretty high volume).
If the boards are small, consider processing them joined together then do a final op (punch, or route) to separate into individual boards.

It sounds fun! I'm looking forward hearing all about it!