Multiple parallel path machining

More
16 Mar 2018 22:49 #107452 by Zig
Consider a gantry machine with table moving in X direction and cutting head mounted along Y axis.

The head consists of a number of cutters ( say ten ) each being individually controlled.

The NC file is generated from a raster like cutting tool data file.

Tool path lines are all equidistant.
Each line may contain either a single or a series of multiple sequential cutting instructions.

Since there is no way to mechanically align the ten tools it is necessary to measure the tool tip of each of the ten tools and use that information ( tool separation and tool departure from straight line ) when pre processing data file and assigning cutting paths to each of the tools.
How would LinuxCNC handle the task in terms of data file being aligned in terms of starting point with tools being at some known ( measured )random departure from a straight line?

Please Log in or Create an account to join the conversation.

More
18 Mar 2018 22:49 #107521 by andypugh
Do all the tool tips follow the same path?

Please Log in or Create an account to join the conversation.

More
19 Mar 2018 04:30 #107532 by Zig
Andy,

Thanks for Your interest.

Let me add a furthar explanation.
I am presently using a single laser diode to expose a photo sensitive surface ( draw soldermask layer onto Liquid photoimageable solder mask).

While it produces great results, the process takes for ever.
I would like to speed it up.

Modern machines use micro mirrors and higher power sources to increase through put.
I too would like to hurry up things a bit.
I am proposing a bank of laser diodes side by side at some known distance apart.
The distance between adjacent diodes would be an integral number of spot diameters.

The tool path consists of a raster like series of lines each representing a full stroke line through the soldermask data.

Lets say the solder mask data raster is on a 0.2mm pitch.
Lets say the diodes are on a 20mm separation.

Every tenth raster line would then be used in controlling the apropriate diode.
Each raster line would consists of multiple G code commands.
Each raster line starts at same position along the raster axis.

If each laser diode is assigned an axis and if each raster line corresponding to a laser diode is "mixed" with the other lines corresponding to remaining diodes then a very simple solution is possible.

By "mixing" i mean overlapping the set of scan lines such that a resulting G code expression is derived which identifies when each of the laser diodes needs to be modulated along a common path.A slight misalignment can be tolerated.

This would result in a motion which features increased level of top start events but at the same time other many would be processed in parallel.

Please Log in or Create an account to join the conversation.

More
19 Mar 2018 10:04 #107537 by andypugh
I have looked at this in the past, and I think that G-code is the wrong way to do it.
I think that what is needed is a component that takes a bitmap image as input and outputs a laser intensity based on the (X,Y) coordinates of the laser.

Then the G-code just needs to create a raster pattern.

In your situation you would have 10 of these components running in parallel, but using the same bitmap data.

Of course this hypothetical component doesn't exist. I did make a start on it once, and got part-way there (It needs to be two components, one user-space component with file system access which loads a shared-memory area with the data, and a separate real-time module which reads the bitmap data and calculates the laser intensity)

I did, however, once make a start on this,
forum.linuxcnc.org/forum/24-hal-componen...al-power-laser#22961

Please Log in or Create an account to join the conversation.

Time to create page: 0.149 seconds
Powered by Kunena Forum