Page 1 of 1

Strange X-Offset when picking components

Posted: Mon Feb 15, 2021 11:08 am
by ibbkoeln
Hi,

I'm facing a strange issue when picking up components.

The starting point:
I calibrated the machine very careful. The x- and y-travels are verified with a precise ruler, the nozzle/camera-offset is set up. When aiming a point with the camera and I jog the amount of the nominal nozzle offset, the nozzle perfectly hits the point. Nozzle calibration is disabled.

Now I move to the tapes. I'm testing with an empty black plastic tape, because that helps to see, if the nozzle hits the hole under the part which is the center of the part. I set up the first hole and let liteplacer show me the first part. This is working perfectly, either. The camera shows exactly the center of the part. When using one of the "Probe ..." buttons, the nozzle hit the center of the part, as expected.

And now my problem occurs: When placing the part in the context of a job, the nozzle is not anymore hitting the center of the part. There is a slight but annoying x-offset which is enough to place 0603 parts off target.

Does anyone have an idea, what could cause this issue?

Btw: Where can I find the description how to calibrate the x- and y-travel? It's no witchcraft, but maybe I overlooked something here...

Thanks a lot!
Oliver

Re: Strange X-Offset when picking components

Posted: Wed Feb 17, 2021 2:48 pm
by JuKu
I took the x offset discussion to emails, as it might take a few back and fort messaging to solve. I'll post the solution here, when found.

> Where can I find the description how to calibrate the x- and y-travel?

Short answer: 40mm per revolution is fimne, no need to further calibration.

Long answer: In most cases, using the nominal value of 40mm per revolution is just fine. Any deviation will be so small, that it will not have any noticeable effect. After all, everything is either set directly or measured by vision. We don't really care if up camera or nozzle position is exactly (say) 40mm from the origin. If the machine thinks a nozzle is at 40mm and in real life the distance happens to be 40.02 or 39.98mm, so what? Even when measuring the PCB, it doesn't matter if PCB manufacturer's mms are exactly the same as our machine's mms; the fiducial measurement and the following calculations handle it.

However, if you happen to have a very accurate ruler, you can put a ruler under the camera, move a certain nominal amount, such as 320mm (8 revolutions), measure, how much movement you actually got and calculate the true mm/rev. I happen to have Schaedler rulers, that are accurate to a few um's (https://www.schaedlerprecision.com/products.htm), so I've made the correction, just because I can 8-) , not because it would matter. (Now when you took this up, I should measure my machine again; it would be interesting to know if the belts stretch over time. I'll do that, but my reference rulers are not in the same town as my machine right now.Stay tuned.)