Awkward imaging problem

[Jason]

I've got a part which is effectively an SO8 with a white body.
Problem is as I can't mask out the body with the threshold
like happens on a black SO8 so it measures the pin extremities
in one axis and the body extremity in the other.
I've killed loads of time trying to get the settings right by
doing a runtime debug but the only thing I can make work
is setting one of the dims tolerances to be at least 40%.
Any other ideas?

[Jason]

As always, sleeping on it helps with clarity. I'm going to create
a fake pad in the CDF which shadows the body shape and
hopefully it will then accept the extra width. Fingers crossed.

[phonoplug]

I would make a new cdf for the part, SO8White for example. Then you can set the body colour to be light in the CDF and that should solve the issue. When a package body is set as white, the imaging analysis expects to see the body and the leads so you should then be fine.

A good example of this is tantalum caps. You often get them in both white and yellow. The folded leads are narrower than the body so if you have your CDF definition set such that the body width and lead width aren't both within tolerance, you can do CDFs for each colour, and you can then fine adjust widths and tolerances for each.

[Jason]

Hmm, the body was already set to "Light" and it was still moaning
"component too big".

Ambient lighting has been a problem too. I need to get this set up
in a more consistent location

[Jason]

I never did get this fixed. I've been running with the huge tolerance to make it work
which isn't ideal. As of yesterday, I now have another component which suffers the
same problem.
Any takers?
I might try my idea of the fake pad approach tomorrow.

[phonoplug]

If you fancy sending one in the post to me I'll see if I can come up with something reliable. Hope its not expensive! PM me and I'll give you my address.

I presume its either an optocoupler of light sensor?

[Jason]

One is a Minicircuits RF mixer, the other is a Bournes trimmer.
PM on its way.

[phonoplug]

Aah like a 3214X series or something? I think I have some of these. Never needed to place them by machine though. But the SO8 could be useful to send.

[spiyda]

Ambient lighting has been a problem too. I need to get this set up
in a more consistent location

I'm very much a beginner at this but I have found that both fluorescent and LED lights cause more of a problem than ye olde incandescent bulbs or daylight.
It may be that the the dark perspex lets through some light frequencies more than others..
I have a dark cloth that I drape over the machine when I am working at night with artificial light.

I also added a self-adhesive LED strip stuck up under the top frame tube, powered from the PC 12V on a switch to help visibility when setting up. - works a treat and only cost pennies!

[Mike]

The acrylic is transparent at IR, and as the cameras are mono they probably have significant IR sensitivity. This would be more of an issue for natural or incandescent light than fluoro or LED. The latter may have significant AC content which may have some effect

[Jason]

My experience matches Mike's theory. During daylight I have major problems but during
winter dark hours with "sufficient :-) " LED and Fluor lighting I get no issues at all.

[lurowl]

We put garden gazebos over the top of our machines to keep the light out. I tend not to have problems when its dark, more so when it is light.