BGA vision/placement demo

[Mark Harris]

Inductance to Digital converters

How cool is that?

I think that might also be an answer to the Z-limit-switch with Juki nozzle quest... I need to check how I could connect to that spring at the tip. Or just integrate a coil into the nozzle holder; it's inductance would also change when the inner nozzle tube moves upwards.

Thanks for this hint!

Those inductance to digital converters are pretty awesome to use. You can also use them for measuring fluids as the "core" of a helical inductor changes. So many uses.. so little time! The thing that gets me though is the resolution, you can get 28 bit versions of them... 268,435,456 count resolution for rotation should be good enough... using them for linear distance you can measure vibration of people walking around the office just sitting them on a desk in a flexible holder - it actually works better than an accelerometer lol.

[WayOutWest]

I would be confident that in our application, microstepping will get you to the right quadrant and maybe in the right 1/8 region. I'm not sure about the latter and I wouldn't bet that 1:16 would be any more accurate, even if it would be available.

Yep, that's my assessment too.

The one exception would be a feedback loop involving the camera -- there are situations where I can imagine getting extreme resolution by using the camera for positioning rather than counting steps.

I think that might also be an answer to the Z-limit-switch with Juki nozzle quest... I need to check how I could connect to that spring at the tip. Or just integrate a coil into the nozzle holder; it's inductance would also change when the inner nozzle tube moves upwards.

Honestly in my opinion the headache with the evil !@*#%^!(*%^ Juki z-switch mechanism has more to do with how tiny the components involved are. It's like a freaking Swiss wristwatch in there. Whatever facility manufactures the nozzle holder itself clearly has the ability to make precise stuff on that scale, so I really think that the electrical contact for the nozzle tip ought to be part of the nozzle holder in the first place, or at least a component that is produced together with the main holder and sold as a package.

Regardless, if you do ever get the z-switch situation solved I don't mind buying another nozzle holder. And I've got a few other random quibbles (from the past two-ish months of using one) if you're still serious about producing an improved version of it.

[Picky]

Also AMS. Magnet strips <10$/300mm, ICs <6$/1k, sub-um accuracy. http://ams.com/eng/Products/Position-Se ... on-Sensors

I can confirm that stuff from AMS works amazingly well. It does not solve all X/Y positioning problems, but takes belt backlash out of the equation. The trick is - you need a closed loop motion controller which can take advantage of servo feedback.

[Picky]

...At rest a microstep is the result of making the motors' two phases fight against each other, with some fixed ratio (1:2, 1:4, 1:8, etc) between the drive strength applied to the two H-bridges. But just setting a particular ratio in drive strength doesn't guarantee you get that ratio of position, and it radically increases the number of factors involved in accuracy...

Certain higher-end stepper drives do employ DSP's with proprietary algorithms and claim they can make things happen. Most of them cost hundreds of dollars but some are available for well under $100. I just described one example in more details here: http://liteplacer.com/phpBB/viewtopic.p ... t=30#p1966

[Picky]

...Those inductance to digital converters are pretty awesome to use. You can also use them for measuring fluids as the "core" of a helical inductor changes. So many uses.. so little time!...

Where were they five years ago?! I spent nearly a year developing contactless ultra-sensitive saltwater conductivity detector as part of a complex hobby project. I wonder how they deal with the interference when the liquid core of the inductor acts as an antenna? That was the showstopper for me back then. I really needed 20-22 Bit resolution for the things I was measuring and I couldn't deal with the drift of the signal influenced by human bodies walking in the same room. The only solution was to place the whole system into a Faraday cage, which was impractical. A fish tank was attached to it and salt water was a very good conductor

[WayOutWest]

I wonder how they deal with the interference when the liquid core of the inductor acts as an antenna? That was the showstopper for me back then. I really needed 20-22 Bit resolution for the things I was measuring and I couldn't deal with the drift of the signal influenced by human bodies walking in the same room. The only solution was to place the whole system into a Faraday cage, which was impractical.

This would be my concern with anything claiming that sort of sensitivity based on inductance. It's basically a radio antenna; how is it supposed to distinguish between EM noise and the intended signal? Especially in a noisy environment like a machine full of steppers.

Certain higher-end stepper drives do employ DSP's with proprietary algorithms and claim they can make things happen. Most of them cost hundreds of dollars but some are available for well under $100. I just described one example in more details here: http://liteplacer.com/phpBB/viewtopic.p ... t=30#p1966

I'd be curious to know what driver IC they use. Or is it done with discrete components (i.e. four actual separate NFETs on a PCB)? If you still have one of these boxes and don't mind taking a picture of the innards that would be appreciated...

Also thanks for posting your videos, I'd actually come across those before! Very cool to know who made those. Can you comment on your magnetic nozzle-changer? I am so fed up with fighting the ball-lock style Juki changer that I've finally decided it's easier to just make my own nozzle changer based on magnets (I'll still use the Juki tips and epoxy a ring of ferrous metal to them, plus a high-power magnet and rubber gasket on the pickup head). Were you happy with the magnetic approach? It seems so much more straightforward than fighting with the stupid ball-lock mechanism. I assume the only drawback is that it can't be used near ferrous components. The only ones I've come across are clock cans -- some of them have casings that will stick to a magnet.

[Picky]

Certain higher-end stepper drives do employ DSP's with proprietary algorithms and claim they can make things happen. Most of them cost hundreds of dollars but some are available for well under $100. I just described one example in more details here: http://liteplacer.com/phpBB/viewtopic.p ... t=30#p1966

I'd be curious to know what driver IC they use. Or is it done with discrete components (i.e. four actual separate NFETs on a PCB)? If you still have one of these boxes and don't mind taking a picture of the innards that would be appreciated...

In the smaller one (DM422) they use a DSP chip with a bunch of discrete FETs on a PCB. Haven't taken the big ones apart, but likely it's the same story.

...Can you comment on your magnetic nozzle-changer?..

I wasn't very happy with the magnetic changer from Madell which used fuse holders. It required very accurate positioning due to tolerances and wasn't very forgiving to any mishaps. No problems with the magnetism though.

Then I developed the plastic changer prototype published in this video: https://www.youtube.com/watch?v=2yi8mou9qAc - it worked pretty well, but then the whole pick and place project was canceled for reasons unrelated to technology. I feel sorry for those beta testers who jumped on board and didn't get a 100% finished product.

The pancake shaped stepper motor in the video is quite amazing. It's a round version of the NEMA14 with hollow shaft and 0.9 degree steps. I believe I have 50+ of them laying around in the garage somewhere. You're close to Seattle?

[WayOutWest]

You're close to Seattle?

Not really, but I am in town a few times a month. Sent you a PM.

[WayOutWest]

So I finally converted back to a Juha/Karl style off-axis 0.9-degree stepper with 160:360 gearing to the axis. WOW the results are amazing.

I use the balls on the bottom of the BGA (fit a line to each column, average the slopes of the lines) to determine the rotation angle of the part, so the angle measurement is very accurate. With the original 0.9degree-NEMA14 and gearing and 1:8 microstepping I can easily align to below 0.100 degrees, and the step increments are 0.050 degrees (as expected) plus or minus 0.002 degrees. Wow.

I'll post pictures/videos soon.

[WayOutWest]

...Can you comment on your magnetic nozzle-changer?..

No problems with the magnetism though.

Then I developed the plastic changer prototype published in this video: https://www.youtube.com/watch?v=2yi8mou9qAc - it worked pretty well,

Yeah, that's the one I wanted to know about. Very glad to hear it worked.

It turns out that the metallic part of the Juki nozzles is ferrous, and it's pretty easy to find powerful ring magnets with a hole larger than the upper narrow part but smaller than the lower part. So this shouldn't be too difficult.

I think the ball-lock mechanism is just a disaster for DIY machines. Hooking up the z-sensor inside there reliably inside there requires precision-machined parts... if you had to have somebody else make the nozzle adapter for you then you'll have to have somebody else make the z-sensor contact for you. You need a tube that's large enough to let the vacuum through but small enough (or insulated enough) not to short to the metal body. Then you have to position it very precisely in the z-direction, to within about 5mm. Finally you have to fix it in place VERY well, because every time the sensor triggers it will exert pressure on the contact... eventually the contact gets pushed up so high it no longer makes contact. If the machine ever overtravels in the z-max direction it will exert a LOT of force -- in theory this should never happen, but in practice it happens every once in a while, and then you have to take apart the entire pickup head to fix it. The whole thing is just completely fidgety and unreliable.