Sweet progress on the circuit board mill at TON this week. I swapped in a pure Win98 laptop (booted to DOS) for the Win ME laptop I tried last week. Using pure DOS mode (vs a DOS box) allowed TurboCNC to run smoothly, no interference with either the laptop timer chip or parallel port access. Motors have that old smooth sound, no dropped steps, rock solid consistency of tool path. I hope to be cutting copper later this week.
More after the break…
Here I’m tracing out a circuit with pen again just to check tool path and the table for level. Looks good. The gcode generated from the Eagle-provided hexapod sample circuit is 1.4MB! (of text!) Barely fit on the floppy disk I use to sneakernet gcode to the mill driving laptop. That’s the only downside to Win98 for this application: haven’t got the Xircom PCI NIC working yet, so no net access. And 98 doesn’t grok mass storage, so USB drives are a bust too. Oh the dark ages of computing in the ’90′s.
Moved back to using this adjustable table I built a while back The 10-32 corner screws can be used to make fine adjustments in table height.
Lower left and centre, you see the wandering 1″ squares from Sunday. This is what happens when you drop steps on the axes: you don’t get back to where you started. I was running a simple square-drawing gcode file on 5-cycle repeats and getting these inconsistent results under Windows ME. Upper left you can see the rock solid result under Win98 booted to DOS. Cycled over that 20 times and had perfect registration.
Here’s a board I cut way back when I first built this mill. There are some issues with inconsistent table height, but you get the idea.