Again, another feeder design
Posted: Tue Nov 03, 2015 7:21 am
I've been working on a cheap but reliable feeder design that is entirely mechanical and does not require any extra movements from the machine, as to not slow down the process. I've come up with this:https://vimeo.com/144454867. It's all been done on my CNC machine and probably cost me ~$5 in materials at the most.
The feeder is actuated by the down-going movement of the head during the pickup process, a lever is being pushed down by the head that is part of an escapement gear setup that tries to do *exactly* 4mm of movement. in theory this works great (this is how clocks work!) but in practice my CNC machine (and probably many others) isn't exact enough to do a given radius sprocket without at least *some* error. And *some* cumulative error is a no-no in feeder design. So i added a little needle that goes down and physically halts the tape once it's come to it's perfect pickup position. This seems to fix the problem entirely.
The cover plastic is removed by some kind of tension, i tried using weights first but found it to be cumbersome and now i'm using a butchered $20 laminator (as can be seen in the video) that should be good for removing the covers of all the feeders i'll ever need and is waay cheaper than any "real" tensioners i've seen. The tension is also important for the feeders operation as it sometimes needs that extra "push"..
At the front of the feeder, just before the tape is removed, is a magnet that sits under the tape. The idea is that it slides the components left and back as they move forward over it, hence vision should not be necessary as the components should always be located in the lower left corner.
The current design also has a holder that you slide the feeder into when mounting it, the back attaches to a standard rail like those used in the LitePlacer.
The design is modular and should be able to do any tape width or hole spacing necessary. I've done some very alpha attempts with 12mm/16mm and 2/3 hole spacing (and any combo within) and it seems to work perfectly.
The current 8mm feeder has *no* visible error when working correctly, though with the last statement comes the fact that i have to test it more. I did a test where i placed >100 components, only had one missing.
My thinking is to do this open hardware with a "hi-you-can-use-it-but-don't-make-money-off-of-my-idea" type GNU contract (sorry i don't know which one that is, BSD?).
Any input on this much appreciated!
Karl
The feeder is actuated by the down-going movement of the head during the pickup process, a lever is being pushed down by the head that is part of an escapement gear setup that tries to do *exactly* 4mm of movement. in theory this works great (this is how clocks work!) but in practice my CNC machine (and probably many others) isn't exact enough to do a given radius sprocket without at least *some* error. And *some* cumulative error is a no-no in feeder design. So i added a little needle that goes down and physically halts the tape once it's come to it's perfect pickup position. This seems to fix the problem entirely.
The cover plastic is removed by some kind of tension, i tried using weights first but found it to be cumbersome and now i'm using a butchered $20 laminator (as can be seen in the video) that should be good for removing the covers of all the feeders i'll ever need and is waay cheaper than any "real" tensioners i've seen. The tension is also important for the feeders operation as it sometimes needs that extra "push"..
At the front of the feeder, just before the tape is removed, is a magnet that sits under the tape. The idea is that it slides the components left and back as they move forward over it, hence vision should not be necessary as the components should always be located in the lower left corner.
The current design also has a holder that you slide the feeder into when mounting it, the back attaches to a standard rail like those used in the LitePlacer.
The design is modular and should be able to do any tape width or hole spacing necessary. I've done some very alpha attempts with 12mm/16mm and 2/3 hole spacing (and any combo within) and it seems to work perfectly.
The current 8mm feeder has *no* visible error when working correctly, though with the last statement comes the fact that i have to test it more. I did a test where i placed >100 components, only had one missing.
My thinking is to do this open hardware with a "hi-you-can-use-it-but-don't-make-money-off-of-my-idea" type GNU contract (sorry i don't know which one that is, BSD?).
Any input on this much appreciated!
Karl
.