"centering" moves after pickup from tapes?

[fcawth]

I have been using liteplacer for a couple of years now, and it seems like a lot of the error sometimes come from the parts being at different places in the tape pockets. Has there ever been any thought to lifting the part partially out of the tape hole, then moving it diagonal back and forth and then back to the center of the tape hole to center it? You might need to have the centering move sizes set into each tape since the tape pockets vary in size, but you could probably have a rectangle or two in the vision setup to just match to the tape size and part size or something. How hard would it be to add x and y centering move sizes to the tape setups (defaulting to zero). It could then pick up the part, move just a small amount up in Z, then do the centering moves in X and Y both ways, then back to the "center" and then lift the rest of the way out. This seems like it would work at least for chip components.

BTW I have a set of 14 panasonic panasert ratcheting feeders integrated into my liteplacer (I'm soon going to expand to about 20). I made an arm off the Z extrusion with a bearing on it to push against the feeder lever, and it works just fine without modification to the liteplacer software. I feel like it flexes slightly in Y when pushing the lever, but the Z motor is definitely strong enough to advance the feeder and the Z travel is just about perfect to properly advance the feeder.

I feel like I need just a little more accuracy to not have to tweak the part placement before reflow. I think I might have some mechanical inaccuracies as well that might sometimes be adding to the part shift in the pocket, but some kind of alignment moves might make the difference and might be faster than having to move over to the up camera (once that gets implemented). It would be great to not have to manually move parts around before reflow, and I"m talking about things like SOT23 and 0603 that others seem to be able to place more accurately.

[AnalysIR]

Interesting idea .... The offset inside the pocket sure looks big when viewed thru the camera with zoom on! (In reality less so)

I also notice that the position of components on the nozzle does vary a bit after pick-up, which I presume is mainly due to this issue.

Other software, probably gets around this by using the up-camera to make the required adjustments...which is of course the ultimate solution.

Unfortunately, I don't have the spare time to test this, it would be useful if it could be validated, before writing code, by just running some manual gcode commands to see if the position can be reliably nudged. Hopefully, nudging would not cause adjacent components to jump out of the pocket?? Also, rather than trying to re-position to the centre of the pocket, it should be easier to nudge to one of the corners of the pocket & and apply correction to adjust for nominal pocket center

####
I just checked a random 0805 resistor data sheet.
- the pocket length is 2.3mm vs component length of 2.0mm
- the pocket width is 1.55mm vs component width of 1.25mm
- So the potential/max offset in both directions is +/- 0.15mm (W+L)

Thus the component/pick-up window should always be 2.15x1.4mm from nominal pocket centre. (assumes everything else is 100% accurate)

Then for the 0805 footprint in the same data sheet...
- Length is 3.0mm => no potential error (assumes everything else is 100% accurate)
- Width is 1.3mm => potential error of 0.1mm (assumes everything else is 100% accurate)
- I don't use 0603 often, but I presume the result would not be worse than this.
- ... if these numbers above are correct, it seems there may not be a case for improving this, based on the assumption that the error is due to position within the pocket.

I use KiCad and tend to use a larger 0805 footprints/pads where possible & if there is enough board space. (also helps if any manual re-work is required later at the cost of using more paste)
- Length is 3.2 mm => no potential error (assumes everything else is 100% accurate)
- Width is 1.4mm => no potential error (assumes everything else is 100% accurate)
- Takeaway may be to design footprints to allow for pocket movement & let re-flow re-align as required.
- Still I do see occasional placement offsets well beyond the 'potential' pocket error, which I presume is due to other issues related to LP calibration - even though the measured positions for pick-up/placement seem to be spot on, within what looks like a very small tolerance based on a manual visual check using the camera.

Obviously, many ICs etc. would have less opportunity to adjust footprint sizes. (some may support increase in 1 dimension e.g. SOIC)

[wormball]

Cool idea! You may not need to know exact cell/part sizes if you do it like degaussing works, i. e. move the head several times back and forth with decreasing amplitude.

The major obstacle i think is that you need to know Z coordinates of the head and the tape with decent precision. I am not sure that the mechanical end switches can give such precision. Maybe optical end switches will help.