ds23man wrote:Do you have a download link?
My build
Re: My build
Re: My build
That will be a problem focussing. Thanks for the link!
Re: My build
Mounted a new down looking camera.
Perfectly aligned:
With this camera also a strange angle:
I think there is something wrong with the sensor placement on the circuitboard, it is perfect inline with the lens.
Perfectly aligned:
- down1.PNG (698.28 KiB) Viewed 8076 times
- down2.PNG (821.78 KiB) Viewed 8076 times
- _DSC2565[1].JPG (321.06 KiB) Viewed 8076 times
Re: My build
Consider that with these kinds of cameras, the lenses aren’t prefect and also the mounting won’t be
(Both the sensor inside the camera, and the mounting of the module on the board). Getting a camera
pointing at at a location precisely and squarely 40mm away is virtually impossible. Even rotating the camera
Lens will significantly change the alignment.
That said, none of it really matters, once you accept that your vision and machine isn’t going to be perfect,
you can then deal with the errors to minimize them.
Essentially you can throw all your concerns about alignment out of the window provided you use accurate
methods to quantify it. Sure it’s good to be in the ballpark, but don’t beat yourself up trying to get it perfectly accurate.
The reason is because you should make sure everything is calibrated at the same level, the top of the pcb, provided you
do this, cameras not straight, and axis not straight don’t matter. Part of doing this calibration, means coming up
with a suite of techniques to accurately measure.
Remember, the machine doesn’t even have accurate fine adjustments to adjust these things anyway.
** However the issue you were having before where the lens was moving, that is important, as it effects repeatability **
Repeatability is different, and it is critical. If you can’t accurately position to the same spot, then you’ll never get adequate
accuracy.
I do have an alignment system I’ve written up for OpenPnp, and if you want good accuracy with Liteplacer, OpenPnp is the way to go (see my post under the OpenPnp forum).
I’m getting superb alignment on small pitch parts (0.3mm / 0.4mm), 0.8mm BGA’s and 0402’s. But everything is at the same level (bottom of component on up camera, top of the component on pickup and top of pcb on placement). If for example I was off by 0.5mm on the height of the component, then a 0402 might be not centered.
(Both the sensor inside the camera, and the mounting of the module on the board). Getting a camera
pointing at at a location precisely and squarely 40mm away is virtually impossible. Even rotating the camera
Lens will significantly change the alignment.
That said, none of it really matters, once you accept that your vision and machine isn’t going to be perfect,
you can then deal with the errors to minimize them.
Essentially you can throw all your concerns about alignment out of the window provided you use accurate
methods to quantify it. Sure it’s good to be in the ballpark, but don’t beat yourself up trying to get it perfectly accurate.
The reason is because you should make sure everything is calibrated at the same level, the top of the pcb, provided you
do this, cameras not straight, and axis not straight don’t matter. Part of doing this calibration, means coming up
with a suite of techniques to accurately measure.
Remember, the machine doesn’t even have accurate fine adjustments to adjust these things anyway.
** However the issue you were having before where the lens was moving, that is important, as it effects repeatability **
Repeatability is different, and it is critical. If you can’t accurately position to the same spot, then you’ll never get adequate
accuracy.
I do have an alignment system I’ve written up for OpenPnp, and if you want good accuracy with Liteplacer, OpenPnp is the way to go (see my post under the OpenPnp forum).
I’m getting superb alignment on small pitch parts (0.3mm / 0.4mm), 0.8mm BGA’s and 0402’s. But everything is at the same level (bottom of component on up camera, top of the component on pickup and top of pcb on placement). If for example I was off by 0.5mm on the height of the component, then a 0402 might be not centered.
Re: My build
Small update:
Do not wire the camera cabling with the Z axis limit switches, the camera shifts between the up and down positions due the flimsy camera mount plate. Second observation: adjusting camera plane will have effect to x axis aligment. The camera will rotate....
Thanks to Jet's exe file with adjustable backlash, the machine has no offset anymore shifting between several locations. The standard 0.4 backlash compensation is NOT enough. Please make this adjustable in a next update!
Do not wire the camera cabling with the Z axis limit switches, the camera shifts between the up and down positions due the flimsy camera mount plate. Second observation: adjusting camera plane will have effect to x axis aligment. The camera will rotate....
Thanks to Jet's exe file with adjustable backlash, the machine has no offset anymore shifting between several locations. The standard 0.4 backlash compensation is NOT enough. Please make this adjustable in a next update!
Re: My build
Very strange, probing on a fiducial is always spot on:
But not on a part:
What is happening?
- fid.PNG (325.17 KiB) Viewed 8032 times
- part.PNG (552.45 KiB) Viewed 8032 times
Re: My build
Solved it!
Redid the down camera nozzle calibration and now it is picking and placing the parts spot on.
I think things go wrong if you do minor corrections to the nozzle calibration on a actual pcb ( measured location, probe down), if you do it at component level everything is correct.
BTW:
The downloadable tape measures do not print correct in Adobe reader, only numbers no lines.
Redid the down camera nozzle calibration and now it is picking and placing the parts spot on.
I think things go wrong if you do minor corrections to the nozzle calibration on a actual pcb ( measured location, probe down), if you do it at component level everything is correct.
BTW:
The downloadable tape measures do not print correct in Adobe reader, only numbers no lines.
Re: My build
0.5 pitch BGA placed with loose part pickup:
Did this 10 times, always spot on. 
- ad.PNG (661.49 KiB) Viewed 8014 times
Re: My build
After assembling several boards I observed a lot of misaligned part. Producing the last board I discovered the 0,5 pitch BGA placed as first ( and was spot on) was shifted after the last components were placed. I am using Loctite GC-10 paste and discovered that is not very sticky. So parts are vibrating from there position. Upping the placement pressure made it better.