Automatic calibrate down camera / needle offset

[Spikee]

It is quite important to regularly calibrate the down camera / needle offset.
I found out that this can change when needles are replaced, you did something stupid and the needle shifted a bit.

So how to do this ? ... lets just use vision!

step 1:
move down camera above up looking camera:


step 2:
Use vision and filters to center the down camera in the up camera's view:




step 3:
Run the already in place needle calibration function


step 4:
Now you can automatically calculate the camera / needle offset X and Y.
You could even include the wobble factor of the needle at said rotation (if that is not already done in the current software)

In this case it is:
Down camera in up camera view: x 2.706 y 87.375
Needle middle in up camera view: x 79.559 y 120.918

Result: x = (79.559-2.706) = 76.853
y = (120.918-87.375) = 33.543
In reality it is more ~80 x 33.543 so there must been a small error in my calculation / measurement. This is all hand aligned

---------------
Alternative:

-Do it manually by making the filter as I did in step 2 and just have a button that can center on one circle that is in view.


-Now run the needle wobble calibration function.

-Calculate offset by hand


These pictures are from me running RMOD software but it is intended for the regular branch.

[JuKu]

> It is quite important to regularly calibrate the down camera / needle offset.
> I found out that this can change when needles are replaced, you did something stupid and the needle shifted a bit.

This shouldn't matter. The initial needle calibration establishes the up looking camera physical location and the needle-down looking camera difference with no correction applied. I would assume that the up looking camera position vs the homing mark stays constant. When you change a needle, the up looking camera location does not change. The down looking camera (which is by definition, the machine location) vs needle position does change. The needle calibration routine then measures this change at various rotation angles. Even though the new needle might not rotate around the same circle that the needle used initially it doesn't matter; the correction is calculated as X-Y difference.

[Spikee]

After a bit of testing with the new liteplacer release (26 oct 2015) I must agree.
I have not had any more issue's this far. The "misalignment" seems more of a camera noise issue because of low light in the room.

Might try the camera upgrade found in other topics to try to combat this.

[Spikee]

I did all the calibrations right but I still see a (significant (0.1 mm or so)) error when changing needles.
Not sure whats going on ...

I have found somewhat a sweet spot which works ok for most needles but still not perfect.
With smaller parts 0402 and 0603 this error can sometimes be significant.