Autoleveller

Using version 8.3 in two places on the board, it carved through the board. The board is 1/32" thick, in several areas, I am removing large areas of copper with a 1.0mm end mill. The mill depth is set at -0.1mm. The board is held down with vacuum on a MDF board. Most of the milling is just fine, the depth looks good, but at two places, the mill carved through the board. The following are the Z depth commands leading up to the the deep carve; (I've rounded up) 0.034, 0.034, 0.054, 0.088, -0.275, -0.639, -0.460, -0.44

When I look at the log file for the probe, X43.76, Y42.45 Z-.9975, the Z for the X, Y either side of this spot are 0.1975, 0.1800 .

How can there be such a great variance the board isn't even 1mm thick, and clearly the carving through the board shows that the probe for this location is way off.

Haven't seen this problem before today.

Ken

Hi Ken,

I once had the same problem where one of the points in my log was 'way off'. For this reason I would always do the following before etching / milling:

• Give the surface you want to probe a wipe to remove any dust or dirt.
  Probe again.
  Check the log file for any obscurities in the numbers.
  Repeat if you find any odd points.

See if this helps.

daedelus wrote:Hi Ken,

I once had the same problem where one of the points in my log was 'way off'. For this reason I would always do the following before etching / milling:

• Give the surface you want to probe a wipe to remove any dust or dirt.
  Probe again.
  Check the log file for any obscurities in the numbers.
  Repeat if you find any odd points.

See if this helps.

For the probe to be off by a full millimeter is to me way beyond the bounds of a clean board. I wipe off both sides of the board and vacuum the platen prior to putting the board down.

True, it shouldn't be off by that much.

Try running the same probe gcode file and check the log file produced anyway, see if any points are off by that much this time.

In reality any probed values are down to your controller and AL has no control over the results (AL merely assumes the probe points are correct). One solution for a later version of AL might be to probe each point 3 times then take the median average of these values. This would discard the values which are way off in most cases.

"One solution for a later version of AL might be to probe each point 3 times then take the median average of these values."

BUT if you do this, make it an option if you can because on a larger sized board, this could make the probe time quite extensive.

No problem Art,

I was going to make it some sort of advanced option, with the default to probe once as usual.

daedelus wrote:No problem Art,

I was going to make it some sort of advanced option, with the default to probe once as usual.

Are you saying that this option already exists or, will be added in an up coming version?

Ken

It doesn't exist yet, Just an idea at the moment.

If you generate the probe file separately, i.e. by checking the 'probe only' option, you can probe as often as you like until your probe file looks right.

The other option at the moment is to manually edit the raw probe file and change the point that reads -0.9975 to a close neighboring point (say 0.1975), save it and read it back into AL.