Mirroring with Mach

[Cairns]

I’ve been trying to use the scale feature in Mach (and also via G51) to mirror a part and while I have had success mirroring a simple set of drill operations, when I try to mirror a (significantly) more complex part a whole bunch of extra arcs are added to the routine.

Anyone want to help a noob understand why this doesn’t work or have any advice on easiest way to mirror a part? Would it work if I went through the TAP file and inverted all of the X values?

[ger21]

I think the scale function was broken when arcs were involved in older versions. It may be fixed in the latest version, but probably not in the tormach versions.

[Pandinus]

The easiest I can think of is probably the unsafest... is that a word??? go into the config and reverse an axis (table homing and limits should work) But you really run a big risk of crashing things if you aren't paying attention, and also switch it right back afterward. If you do this homing sequences will most likely crash things as well!! so be careful.

[justgary]

Cairns -

Yes, you can edit the TAP file and switch signs for an axis. I always switch the negative ones first by substituting an unused letter, such as O, as a place holder. That way you can use a global replace and get them all. For example, to negate the Y axis, you could easily do it with three global replacements:

Replace "-Y" with "-O"
Replace "Y" with "-Y"
Replace "-O" with "Y"

Global replacement keeps you from missing one and wrecking your part.

Regards,

- Just Gary

[zephyr9900]

Just Gary, you can do that if the gcode is exclusively G0's and G1's. For the temporary placeholder I use something like $, because there are few letters that don't show up in gcode sooner or later.

When arcs are involved you also need to swap G2's and G3's, and the signs of the I values (when mirroring in X axis) unless you're using radius arcs.

Randy

[justgary]

Randy -

As usual, you are correct. In fact, the last time I swapped sign on an axis was about a year ago, and was actually to swap axes from Y to A to engrave on a cylinder using the rotary table. Using a bigger string than just one character will keep problems at bay, too. Something like -Y => $fred$ and then $fred$ => Y has more of a chance of uniqueness than just one character.

The bottom line is that you have to think about the swapping, and visualizing the result in something like NCPlot can help ensure that you got it right.

Regards,

- Just Gary

[zephyr9900]

You wouldn't believe the weird shapes I got the time I forgot to reverse the signs of the I's. You are so right in your admonition to independently check the gcode. I never send anything to the machine without running it through CutViewer first. Weird shapes on the screen are ha-ha, weird shapes in the workpiece as the cutter careens through the material are !*@$#&%@.

Randy

[keen]

Hi Just Gary - why dont you mirror the part at the part importation/creation stage and do the toolpath for L and R ?

[Cairns]

Originally Posted by keen Hi Just Gary - why dont you mirror the part at the part importation/creation stage and do the toolpath for L and R ?

It took me 8 hours to do the CAM work, doing it twice is a lot of effort for something that should be as simple as adding a G51 statement at the beginning of my file.

I tried a number of combinations of the file editing suggestions above with no luck. Inverting the X and I and messing with the G2 and G3 gives me the same result that G51 does in Mach. I need to break it down and create a very simple part and experiment with getting it right, I haven't had time to mess with that yet though.

I didn't stumble across a way to invert the axis in the Mach config files. If someone's got a slightly more detailed suggestion on that I'm willing to try it.

I did make a copy of my completed SprutCam project file and inverted the model to see what the results looked like. It was pretty good except that the 2D contouring and the hole machining had to be adjusted. It's close enough.

It sure seems like G51 is something that people would use enough that it would function correctly.

I'm tempted to just swap the phases on the X stepper. I'm kidding of course.

[zephyr9900]

Originally Posted by Cairns Inverting the X and I and messing with the G2 and G3 gives me the same result that G51 does in Mach.

Cairns, I suspect something might be irregular with your gcode. I took a moderate 2.5D gcode file produced by SheetCam containing both lines and arcs and did the above operations to it. It reversed successully, as viewed by CutViewer Mill. (Unfortunately, CutViewer doesn't seem to have G51 support so I couldn't try that to confirm.)

Randy

[Pandinus]

Again things will go completely backwards, but you might want that.....
you can also tell mach to home in the opposite direction and change its "home offset" to reduce the likelihood of a crash, but its more to change back later.
KEEP YOUR HAND ON THE ESTOP!!!!
I don't know if you have access to it if its a Tormach version of Mach3....they might be trying to make the software "FOOLPROOF" Well here's your fool

[zephyr9900]

Pandinus, you are correct that the Tormach version of Mach3 doesn't have the Ports and Pins menu available to the user, which is why in general we must go through the machinations described above (I say "we" even though I'm running an unlocked/unrestricted copy of Mach on my machine. I have done the text-edit reversal of gcode files myself to avoid the hassles you described in altering the homing behavior.)

Randy