Re: MDC Luna EXP 7722FV handbook
handbook is currently 91 pages and I'm just starting the PCB programming, I yet have to include machine calibration, placement & vision test.
I don't own a dispenser so that part will be just copy / paste from the MDC manual (and honestly, I'm sure it "just works" as the difficulty isn't about programming this). I expect to have the document to be roughly finished in 10 days / 2 weeks, I work on it in the public transports when I go to work, a little bit everyday.
I've included a section for DIY improvised trays, various tricks about reels, most of them are simple things (and you'll probably yell d'oh at me...) but I wanted to include ALL level of expertise in the document. I also included most of the adafruit & freaklab tutorials from the wiki with proper credit.
glad to count one more user !
let's see if I can help about your part placement, please post exact refs so that I can see the package please.
as for getting out of vision, well, when the machine quits vision it's because it "thinks" the part is centered. If it's not the case, it's possibly because it "saw" the part centered.
things to consider :
- leads contrast and brightness : For bottom vision, most of the time it's the problem. For symetric chips checked as SOP in Lv=26 (small chips), the machine will do the centering by finding the 4 extreme pins. If there's an offset on them, you have the part off centered or not properly oriented.
- machine calibration. Mine was supposed to be but in practice, I'm pretty sure you have to run a calibration to be sure of what you're doing, otherwise, you'll spend the rest of your life adding nonsense offsets everywhere in part def.
bottom vision should not be corrected with offsets overall.
About the leads vision : especially with ROHS now, the finish of the leads can be problematic. You also have those shinny reflective plastics on certain part, despite they're black, they shine some light back and contrast with the leads is poor.
As I got explained by MDC, the threshold system for bottom vision uses min and max of the picture, so that you don't have an absolute gray scale level as a threshold. With poor contrasted parts, you might need to increase or lower the threshold with the Level parameter.
1 leaves it untouched, < 1 turns it lower etc. I had to create so far two 8 pin SSOP variants for my design (I have a total of 3 different SSOP on the one board) because one had a weird contrast. Pins are still incomplete because the leads finish is "wavy" but I was able to have the edges nicely.
You might need to experiment first with bottom vision lighting variants : add a 1 in front of the vision algo (126), you have a darker lighting, add a 2, you have it brighter (226). Lighting change is about 20-30%. Once you find the one with the best result, start playing with the level.
Real time level is possible when testing vision. Pick up a part (improvise a tray, create it in the machine, if the part is reeled) then manually pick the part, go to tray, edit (2-2), select tray, teaching mode (F8), nozzle mode (9), pick (R), vacuum on (V), nozzle up (R).
Test part with bottom vision with Shift-P, enter part #,0 (angle doesn't seem to work) and part will be taken to the bottom vision camera. Adjust the level with the + and - buttons in the live video window. Unfortunately, the level is overwritten by other centering info, but by clicking several times, you can see it before it's erased. Most of the time, I got the level adjusted between 1.2 to 0.6
I had an extreme case with my square inductor (that I was finally able to place with bottom vision as a CAP) because it's so shiny and reflective that I had to lower the threshold with level=0.4 so that anything that wasn't super bright (the lead) will turn black. As a result, my bottom pads/leads aren't completely seen "filled" but I don't care as only the edges are required for the CAP placement (yay)
If you get kicked out from the vision loop because the threshold ends being wrong, press escape, part returns to the tray. You can check it again immediately with another Shift+P process.
you can also return it to the tray with R+V+R (down, vacuum off, up).
then press F3 for a quick jump to part def, edit your part, validate with ENTER, then escape will return to the tray where you were, go in edit again, pick the part, test it. Wash, rinse, repeat :wink:
Overall, with a bit of obsession, it won't take you more than 2 days of work to create and test your main parts. New parts will be created much quicker once you know what to adjust.
The placement problem I got AFTER were something else and I went nuts. But you'll have to wait until the calibration section of the document is made to check it cause I'm not going to type it all twice :mrgreen:
your SOT-363 could also be check with side vision, but you'll need an offset correction. Once defined properly, it will work like a charm and quicker than bottom vision.
get us a bunch of screen caps here and let's use this thread to debug part placement or part definition. Include a screen cap of your current part def for the 2 problematic parts you mentioned in your post, I'll be happy to help. Screen caps are obviously easier to take when performing a vision check manually from the tray.