Almost placing

Started by Switch, May 21, 2011, 05:00:30 PM

Previous topic - Next topic


Hi everyone,
I am experiencing strange behaviour with my RV4s: it picks-up a component, moves over the position to place, than the head goes down half-way (almost placing) but aborts the placement and dumps the component into reject-bin. After which tries again with the same result and goes to pickup the next part just to do it again, and so on until I stop it.
I am setting up the PCB height every time. What can cause this to happen? What I am missing here?
Needless to say, I am currently self-learning to operate it.
Will appreciate any help.
Thank you.


Sounds like the component has failed the vision analysis. It takes time for the PC to work through all the possible rotations to check whether the component can be identified within the limits for that component, so after taking the snap-shot of it the arm keeps moving to where it should be placed while the PC does the analysis. By the time its got to the location on the board and is half way to putting it down, the system then checks the result of vision analysis which has by that time completed. If it failed analysis, it retracts the dumps the component which is what you are seeing. Try doing a 'run-time test' (tick option within CDF editor) for that package and you'll probably see the problem then.


Yes, that's standard 'vision failed' behaviour.
Assuming it's showing a reasonably clean image, look at the top-right of the screen when it happens and you should see an error message - usually component too big or small, look at the measured sizes it shows and (assuming the image looks OK) tweak the sizes in the CDF to match.
If it isn't getting a  clean image you'll need to tweak the vision parameters - some parts need offsets, e.g. for different body colours.


Start by setting the blob size to be 1/3 of the smallest dimension of the part.. 1206 60 thou so 20 for blob..  then set the threshold to 0 and run time test. then increase in steps until the graphical image becomes the grabbed on. make a note of the number... typically 80 or 90 with res 1206 then decrease the threshold until you get so much data that the device fails the vision test typically -80... then you look at the spread lets say 70 to -50 which is 120 then half it. and subtract from the highest figure. so you end up at 10 for the threshold. works with a range of lighting levels and wont reject the parts..  :) btw if you do an SO the blob size is the 1/3 the width of the leg as thats the smallest dim.

That process will work for any device.


The RV is bad at dealing with varying light levels, so you want to keep ambient lighting as consistent as possible - in particular if it's near a window, fit a blind.
Be aware that the tinted plexi covers are transparent to IR, so IR-rich sources like incandescent or natural light will not be significantly attenuated by them.
The 'typical' threshold values given above will vary depending on the camera's calibration setting, but there shouldn't be a reason to change this except maybe due to illuminator LED ageing.
Simple rectangular parts are genarally no problem (although you may want different thresholds for Rs and Cs so worth having seperate CDFs and parts in your library).
You will find some parts that are very fussy, particularly where you need to discriminate metal leads from body edges.