Unfortunately, this is due to the way Eagle creates polygons.
Your best bet is to find some gcode editor that will allow you to reverse the line directions in the files that pcb-gcode creates.

Regards,
JJ

On 30 Jun 2019, at 12:47, harry0099@... wrote:

Hello comrades!

I use pcb-gcode.ulp for some time now and get decent results, probing the copper surface with OpenCNCPilot.

But I realized, that the edges of the second+ rounds are not as good as the first one.

That is easily explained, since the milling bit is rotating cw and the bit moves in the correct direction for the first run, but in the wrong direction ("climbing") for all sucessive rounds.

Now pcb-gcode.ulp has the option to select ?always climb“ to change the direction of the bit on its way, but this option changes the behavior for *all* rounds.

I wanted to implement another option which allows for turning on the ?always climb“ selection only for the second and all following rounds, but I can’t fiddle out where in the code I may attack.

So my question to all of you:

Can anyone give me a hint, where in the code I can start to look, or even better, where the decision between the first and all successive rounds is made?

Explanation:
I think I observed, that pcb-gcode does one first round in which it surrounds all traces once, so when stopping after this first round, one would have a functioning board with all traces as expected, but with too narrow trenches.
The following rounds only widen the already defined trenches to ease soldering.
All following rounds erode the "other side" of the trenches but this happens in "climb mode", resulting in not so good edges.
So I consider it as good practise to change the moving direction of those following rounds to make them again "normal".

Thanks in advance!

Harry