Friday, January 1, 2016

Focus tuning.

Happy 2016!

Here are the latest changes I’ve made, and their results.

First, when using Eagle and PCBGCODE, I noticed that the script was drawing too many traces around the tracks to isolate them. They can be seen in the picture below:


The isolation lines seemed to be a bit too messy, so I decided to try to make this better and changed some of PCBGCODE parameters. The first one I changed was the Etching Tool Size.


I noticed that this parameter must be smaller than the smallest track separation of your board, or PCBGCODE does not generate the traces, like this:


So I have set it to 0.1mm and tried again, getting my isolation traces back. But I was still getting too many traces around the tracks, which were making the etching messy:



After a few tries, I noticed that if I set the Board Isolation Maximum to zero, I get only one line around each object of the board. This is the parameter I changed:


But after many other attempts the best I could get was a line around each isolation track AND a line around each part of the GND polygon made to connect the circuit GNDs.


I tried, but I could not eliminate the lines around the GND polygons, so I decided to eliminate the GND polygon on my board and connect the grounds using tracks.


And, after running PCBGCODE again, with the new parameters, I got this result on OpenSCAM:


Which is what I was looking for since the beginning. And the pattern generated by the laser was much better:



But I was still having problems with the laser focus. Moving the focus to the Z axis opened up many possibilities for me, showing that the focal depth of the lenses that were shipped with the laser is very deep.

I used to search the laser focus by qualitatively measuring the amount of light that was generated when the laser evaporated the ink, but this method was not reproducible. So I wrote a function on YetAnotherGCodeSender to plot a focus scale, with one trace in each Z depth. Some of the results, follow below.


Under a USB microscope I could see that only two of the traces had almost all the ink removed. The others around had considerable amount of ink left inside the channels. After selecting the a good Z position I got the following results on the PCB:




Now it may be time to tune the ShapeOko2 screws and try to improve its mechanical positioning.

Thanks for reading! Ah, and please leave comments if you believe you can help or have questions.