[jsantos]

I tried manually jogging the Z axis up and down about 5" at the reference point 20 times at 50ipm and there was no change in the X and Y at all.

I wrote the gcode to do a Z axis move from zero to 3" at 25ipm plunge rate 20 times. No change in the X and Y at all.

I wrote the gcode to do an X axis move from zero to 3" at 25ipm rate 20 times. No change in the Y or Z at all.

I wrote the gcode to do a Y axis move from zero to 3" at 25ipm rate 20 times. No change in the X or Z at all.

The gcode files have the same setup before moves as the circle and square other than I didn't add the spindle direction command because it does nothing on my machine anyway.

When the gcode files were running I put my finger on the large pinion drive belt gear of the X, Y, and A axis and felt no indication of any attempt to move. Not even a little quiver. Motionless. As in dead, dead.

When you move any of the axis, is there a shift at all in that axis? E.g., if you move the x-axis from 0 to 3 20 times, does it move from 0 to 3 everytime? If it does not shift, then could your problem be is when you do i-j moves?

[RomanLini]

That rules out hardware, no big surprise.

Looks to be in software. Back on about page 18 or somewhere I said it might be in your gcode where it tells to move down after cutting one layer and is somehow telling XY to move at the same time. It's probably a weird config thing where it is interpreting a "move down and repeat" as something like a "move down, compensate and repeat" where there is something that is compensating, adjusting, re-setting, XY etc in the process of moving down and repeating the cut pattern.

If it's related to config or setup that explains why someone else running your gcode may not get the same problem.

[CarveOne]

What if you remove the spindle commands in your original circle and square file?

The spindle command is the one for clockwise rotation. Since I have a wood router and not a spindle the command shouldn't matter, but I'll remove it. It's a default M code from CamBam.

[CarveOne]

When you move any of the axis, is there a shift at all in that axis? E.g., if you move the x-axis from 0 to 3 20 times, does it move from 0 to 3 everytime? If it does not shift, then could your problem be is when you do i-j moves?

I was looking at that. I put a mark at the 3" point on the X and on the Y tests and there was no shifting. It is as reliable as you could expect. The 3" travel is exactly right. I ran the same test a few times and looked for shifting of the pen from the zero reference point on each new test run.

I haven't made your square corner square test gcode yet, but I'll do it this morning so I can run it tonight. I'll leave IJ mode enabled first, then disable it and see what happens. That mode has been enabled from the start of this episode.

One other thing left to try is substituting your pport cable.

[CarveOne]

That rules out hardware, no big surprise.

Looks to be in software. Back on about page 18 or somewhere I said it might be in your gcode where it tells to move down after cutting one layer and is somehow telling XY to move at the same time. It's probably a weird config thing where it is interpreting a "move down and repeat" as something like a "move down, compensate and repeat" where there is something that is compensating, adjusting, re-setting, XY etc in the process of moving down and repeating the cut pattern.

If it's related to config or setup that explains why someone else running your gcode may not get the same problem.

Most of the settings in the General Config menu are still the default settings in a new Mach3 installation. I read some documentation somewhere that said to be sure to enable IJ mode, so I did. Maybe it needs to be disabled for this particular type of use when repeating the gcode at the same starting point.

Do you guys have it disabled?

[CarveOne]

I created a new square and circle gcode file that is the same as the first one but with a square corner square. (I'm a glutton for punishment)

Now a comparison can be made to the original file and I can turn off the IJ mode on both types of runs. The pen will plot a square corner but the cuts in wood will have a radiused corner due to the .25" diameter router bit.

[CarveOne]

The square corner square and circle file I created this morning does exactly like the original square and circle file as far as the step overs.

I'm going to have to give this effort a rest for a while anyway. I'll work on this problem offline as I can get time for it. I have a contract job that will require working at my home work shop for a few weeks. First order of business is to clean the place up tomorrow, reorganize things, and make some room available for a small production assembly line. Then study some CAD drawings.

Thanks to everyone for all the help and all of the head scratching on my behalf. I'll get back to it, and will be checking in here daily. The machine can make accurate parts as it is. I just have to set 0,0,0 each time I start another cut.

CarveOne

[ger21]

............and I can turn off the IJ mode

IJ mode isn't on or off, it's either absolute or incremental, typically, if it's set wrong, you'll get very large circles, or error messages. Not what you're seeing. The end point of a G2 or G3 arc doesn't change with IJ mode, so it won't cause a loss of position.

[CarveOne]

IJ mode isn't on or off, it's either absolute or incremental, typically, if it's set wrong, you'll get very large circles, or error messages. Not what you're seeing. The end point of a G2 or G3 arc doesn't change with IJ mode, so it won't cause a loss of position.

I'm not getting anything like incorrect size parts. The square and the circle measure as they were drawn in CAD. If I draw a 20" square it will be the correct size when it is cut. I wish I knew why the X/Y offsets are happening but it isn't a big issue for anything I plan to do with the machine. I'll be working on it in the background. I just have the need to make it a lesser priority while I work on something else that pays money instead of costing money.

[CarveOne]

"I wish I knew why the X/Y offsets are happening but it isn't a big issue for anything I plan to do with the machine." CarveOne

Ha! Ha! Famous last words.....

I have a new project that will need step downs and it looks like my new machine is pretty much useless until I find out what is causing this and fix it.

My buddy and I reconfigured yet another computer this morning so I can try Linux/EMC2 on it today. I finally got a new EMC2 setup to work properly as far as the X/Y/Z axis is concerned but the A axis just will not run as slaved to the X axis. I can make it run by itself, but not in combination with another axis.

So I'm going to disappear again and see what I can find on the EMC2 forum here and on other sites that confirm that slaving is or is not currently possible in EMC2. One post that I previously found said that he had it working. The post left out some of the important details.

CarveOne

[contactirfu]

carve one there is no need to run it as slave - you can pull it out on a single pin also -read this thread

http://www.mechmate.com/forums/showthread.php?t=1232

[CarveOne]

carve one there is no need to run it as slave - you can pull it out on a single pin also -read this thread

http://www.mechmate.com/forums/showthread.php?t=1232

Thanks! I think my C11G BOB has 74LS245 devices on it. I'll look at it tomorrow.

CarveOne

[contactirfu]

then your single pin will be able to drive 2 203 drives. just invert a pair of wires on the stepper motor.

[CarveOne]

then your single pin will be able to drive 2 203 drives. just invert a pair of wires on the stepper motor.

Yes, that looks simple enough. I printed a copy of Mike Richards two posts with the details and the buffer circuit schematic.

Thanks for leading me to that info. If this works with the C11G board then I can see if using EMC2 does the 45 degree step over thing I'm having problems with. I will also try this with the Mach3 computer also.

CarveOne

[joecnc2006]

I would not really recommended running two drives from a single output, Yes you can do it I have in emergency's, But you may want to put in a homing routing which will need the two motors to be Independent from each other to ensure the gantry is sq. I guess you could always do that manually, but I would not, and ever so often you may loose steps either computer error or any number of things can happen. this is from experience.

[Pandinus]

In EMC i believe you don't need an "A" axis .. you can put Xstep on two different pins and Xdir on two different pins

[CarveOne]

I would not really recommended running two drives from a single output, Yes you can do it I have in emergency's, But you may want to put in a homing routing which will need the two motors to be Independent from each other to ensure the gantry is sq. I guess you could always do that manually, but I would not, and ever so often you may loose steps either computer error or any number of things can happen. this is from experience.

Yes, the posts that Irfan pointed to says the auto squaring is defeated by doing this. At this point though I just want to see if EMC2 gives me the same problems with the same file(s) that I try cutting on the new CNC machine that has two motors on the X axis. Auto squaring is a nice feature to have but is not a high priority for me right now.

I already tried two different Windows Vista and XP computers with Mach3 and get the same results. The problem followed the computers over to my first CNC machine that runs fine with EMC2. Now I want to try EMC2 on the new machine in an effort to eliminate the software and computer side of the system. I was going to do that before building up a minimal XP Pro/Mach3 computer from my EMC2 computer. EMC2 did not have the stepover problem with the square and circle gcode on the Xylotex controller and a single motor belt driven dual screw X axis. Yesterday I built up a third computer to have EMC2 capability again. If driving two motors from the same BOB output pins doesn't solve the step-over problem, I can at least say that I have eliminated something in Mach3 as being the source of the problem.

It may well be the control box layout and wiring. But, why did the computers do the step-over thing on a different machine with NO control box? The gcode isn't suspected because three other people say that it runs fine for them.

All of this effort and it seems that I'm no closer to resolving the problem.

I'll be working on this all day again today.

CarveOne

[CarveOne]

In EMC i believe you don't need an "A" axis .. you can put Xstep on two different pins and Xdir on two different pins

I did that yesterday. Starting with Stepconfig I chose the gantry stepper configuration and then ended up with the following in order to move the correct axis as it is wired to the motors.

Pins
----

1. Unused
2. Xstep
3. Xdir
4. Astep
5. Adir
6. Ystep
7. Ydir
8. Zstep
9. Zdir
10. thru 17. are Unused

Even if I set it up as above with pins 4 and 5 also selected as Xstep and Xdir it still does not slave the drives. I'll work on this some more today though.

CarveOne

[CarveOne]

EMC2 will indeed slave two drives. (Of course everyone knows that but I had to be convinced.) Documentation on how to do it is sparse to nonexistent. It is working now. I know I tried this before but may not have seen two motors move at once when I saw the wrong motor moving. Because it racks the gantry I haven't been moving the A motor much distance.

Pins
----

1. Unused
2. Xstep
3. Xdir
4. Xstep
5. Xdir
6. Ystep
7. Ydir
8. Zstep
9. Zdir
10. thru 17. are Unused

Y direction is inverted, all others not inverted.

Now I can check that the travels are calibrated and run the gcode files to see what happens. More later today.

Thanks Pandinus for getting me to try it again now that the other motors are operating as they should.

[Pandinus]

Oh I think I remember when you are in the STEPCONFIG it would only move one of the X motors for me when "Testing the Axis" but it worked fine after saving and then actually running the machine controller portion. Glad you figured that much out