Monday, January 24, 2011

Vision troubleshooting tips debated

Under the title of “Troubleshooting Vision Systems,” (January 2011,) John Sprovieri of Assembly Magazine shared some ideas about how to deal with machine vision systems that are being a little awkward. The article has some great content but there are two points with which I disagree. Let me take you through them.

  • Defocus the lens to produce a blurry image.

    I find this a good method of eliminating small scratches and marks that might otherwise be detected as edges and blobs, but there is a problem: a well-intentioned engineer or technician unfamiliar with the setup will almost certainly want to bring the image into focus. Now given that factory vision systems run 24-7, how do you ensure that everyone who works on the system knows precisely how unfocussed the system should be?

    Much better, in my humble opinion, to achieve the defocusing in software. That way it’s one the same way every time and is unlikely to be changed by the most helpful tech.

  • LED’s suffer no drop-off in output

    Well that’s what the manufacturers claim, but I am not so sure. First, it’s been my experience that many LED lights don’t get the claimed life, probably because they run too hot. Second, I think there is some degradation, due to aging of the encapsulating material, if not the actual LED, and then add in the effect of dirt. Lights will build up a film of dust, so even if the LED output doesn’t change, less light will actually reach the target.

    Now don’t misunderstand me; LEDs perform much better than fluorescent and incandescent lights, I’m just saying that you should never assume their output now is the same as it was two years ago.

So that’s two points from a good article that I think are open to debate. But don’t take my word for it. Read the piece and decide for yourself.

1 comment:

Spencer Luster said...

Regarding LED light output, they definitely degrade. The Illumination Engineering Society recommends specification of LED liftime as the point at which the output has degraded to 70% of the original output.

BTW: There are many LED links available, but one of my favorites is Don Kliptein's site:

He has old data, keeps up with new goings on, and he does testing himself. A nice resource.