We were having difficulty with the MDC vision placement for our leadless parts. When we did a parts test, the vision worked fantastic:

{{:mdcpickandplace:test0.png|}}
{{:mdcpickandplace:test1.png|}}
{{:mdcpickandplace:test2.png|}}
{{:mdcpickandplace:test3.png|}}

But when it came to placement, the parts were always shifted or skewed :(

{{:mdcpickandplace:lgaassembly.png|}}
{{:mdcpickandplace:lgaassembly2.png|}}

I knew the part definition was OK - but during test, it would try 3 or 4 times to center the part. Whereas for assembly it would take one photo and take it from there. Much sadness.

Then I found out that you could make the assembly mode use the 'Closed Loop' vision system. Having taken a Signals and Systems class, I liked the sound of that!


  In assembly mode, we have feature so called closed loop which repeat measuring until the target tilt reaches the limit looks like shift+P.
  checking your system constant when shipped out, k11's remark starts as " LOse" and if you change it as "CLOse" it become closed loop. (actually first C  makes closed loop)
  And k11's first value is the angle limit for parts from tape or stick tube and 30 is approx one degree. current value is 5 and it may too small try 10  (0.3deg)
  (2nd value is for parts from tray but also 3 will be too small) 
  Also the last of remark of  k7' is 10 . This is waiting time for image capture but for closed loop you may better to increase to say 90.


OK so we went to the **System Constant** spreadsheet and change the **comment** of **k11** to "CLose..." (just the first character). We didn't change the angle limit (still trying to understand it) but we did set the **k7** remark constant to 90

The machine now uses closed loop vision, it works much much better for the leadless parts. Will report back on more info