Autoleveller generates bad gcode

[Olovskos]

Hello,

I have a problem with autoleveller.

I probed a gcode file for a pcb where I set the target depth to 0.02mm. I don`t know if this is deep enough to remove the copper but I wanted to test it.

But if I run Autoleveller with the gcode and probe the new file and start milling/isolation it goes almost 1mm!!! deep...

So whats the problem? I did pcbs before with autoleveller where the milling depth was 0.1mm and everything was fine.

Any ideas?
Maybe the calculation is wrong?

[daedelus]

My first guess would be that there was a spike during probing. By spike I mean a probed value which is much bigger or smaller than any of the other probed values. This happens occasionally to me and can be caused by dust on the board or on the probing tip. Try to give the board a good wipe and reprobe. A spike can cause incorrect results in calculations and you would see deep cuts like the 1mm ones you are seeing.

If you are using an RPF you can visually check the Z column to see if any of the values look oddly different to the rest. If so, either manually tune that value or reprobe to make a new RPF.

This would be my first accusation anyway.

[Olovskos]

I use mach 3 and the pcb is clean.
I also use a very very slow plunge rate to probe the pcb .
The values of the z axis are looking good to me so I really don`t know what is wrong but I will investigate

[joeaverage]

Hi Olovskos,
I had the same problem. When I looked closer at the probe file I found one entry
with a Z of -1mm, clearly an aberration relative to the rest of the board.
I did not notice it before I committed it to run and had a disaster much like you had.

I like the visualisation feature of latest AL7 version but it does not appear to be implemented
in AL8 yet.

No matter whether visualisation is operable or not I go through the probe file very carefully
now before I commit it to levelling code.

Craig