workpiece touchoff getting lost (v2.6.4)

tommylight wrote:

Check the tool table in fusion360.

i went over my tools used in the program above and double checked that the tool number, pocket number and offset number are all the same.

tommylight wrote:

andypugh wrote:

tommylight wrote:

Check the tool table in fusion360.

This shouldn't be the problem unless Fusion has been told to apply tool offsets. (I don't actually know if Fusion can be told to apply tool offsets)

I do not know that either, but it sure looks like it, and from the looks of it it is doing the first pass with a 6mm diameter tool and the second with 10mm, or vice versa, as it was not mentioned if the error was +2mm or -2mm.
Then again, all cam software have settings for tools and tool offsets and tool types.

i do not tell fusion to apply tool offsets at all... in the code you can see that it outputs g43 to use the tool offsets stored in linuxcnc.

what happened is that the touch off gets about a 2mm bump on the z axis, i've seen this several times now, and last time when i used the machine it seemed like the touch off on x or y was bumped by about 2mm. it seems to be kind of random, so i'm still wondering if the start of the program above somehow "deletes" or modifies the touch off i performed, and if i should be touching off each time i load a different program even if the tool is still the same ...

touching off each time i load another program is very annoying and problematic since i want to keep the exact same x,y touch off ...

andypugh wrote:

I think we need to get to the bottom of whether this is a tool offset, a work offset or an actual physical drift.

So: Touch off the Z in your normal way, but "lie" about X and/or Y so that the program will execute well away from the workpiece.

Make a note of the machine Z position in both relative and absolute coordinates. Also look at the Tool and G92 offsets (There is a version of the DRO that displays all these in Axis, available from the menu)

Now run the program, but stop it immediately after the tool change.

Jog back to the top of the work and compare all the offsets.

Now run the program in air. Then jog back to the reference point and look at the offsets again.

Report back with the numbers.

(Edit) I am assuming here that you have checked the G-code, and that the G-code origin isn't a stock-box point 2mm above the surface of the work.

thanks, will do this ASAP and report back.

today I ran different programs all using the same 10mm tool.

i tried something differently - execute m6 t1 on the MDI prompt before running the program i loaded, and before touch off.

i then ran my program, loaded the next one (same tool) and verified the touch off positions and found that the z offset had shifted by 2mm (i used a piece of paper to touch off to the top of my fixture, as soon as the tool grabs the paper i touch off z and enter 0.1mm).

i kept going and loaded several programs, each time making sure to repeat touch off for x,y, and z just to be sure.

when i stopped working for the day i moved my spindle to where 0,0,0 should be and then found that there was a 4.2mm offset (roughly) whereas for the z axis there was a 2.2mm offset (roughly).

keep in mind that for each program i stopped the program as soon as it was done with the first pocket in the fixture since i only had one piece of stock in the fixture i wanted to mill (2 pockets can hold 2 pieces of stock).

i stopped by hitting the e-stop button. this is also how i change tools - press the estop button before i touch the spindle, to make sure nothing crazy happens as I put in the new tool.

i really don't get why after running a program, there is now a 4.2mm offset on the y axis whereas on the z axis there is a 2.2mm offset, and the x axis is still alright.

it seems to be the same issue each time. i was thinking there could be a problem with the stepper motors (connected via belt with the ballscrew spindles) but then surely the offset should be different each time???

anyway, i'll do what you told me to do and report back tomorrow.

Pressing the e-stop can cause lost steps, especially if an axis is moving when pressed.

The e-stop does a couple of things, that can lead to lost steps in a stepper system, 1st if anything is moving, step pulses just stop, with no deceleration, causing possible overruns and loss of position. 2nd if your drives are connected to disable on e-stop, they will be powered down and can free wheel, causing loss of position.

Technically on a open loop stepper system the machine should be re-homed anytime an e-stop is triggered. If everything was already stopped, and your system isn't set up to enable/disable the drives, then this probably isn't the source of your problem. Look for lost steps elsewhere.

Todd wrote:

Pressing the e-stop can cause lost steps, especially if an axis is moving when pressed.

The e-stop does a couple of things, that can lead to lost steps in a stepper system, 1st if anything is moving, step pulses just stop, with no deceleration, causing possible overruns and loss of position. 2nd if your drives are connected to disable on e-stop, they will be powered down and can free wheel, causing loss of position.

Technically on a open loop stepper system the machine should be re-homed anytime an e-stop is triggered. If everything was already stopped, and your system isn't set up to enable/disable the drives, then this probably isn't the source of your problem. Look for lost steps elsewhere.

i didn't know that ... i feel stupid not having thought about this earlier. I'll see what happens if i just stop the program and look at the touch off data again.

in my setup the e-stop is connected to the controller box, which has a benezan motherboard in it which controls the leadshine drivers, connected to the steppers. the benezan board has an e-stop signal which goes over the parallel cable to the pc if i remember correctly.

i didn't do a lot of tweaking to ini/hal beyond just running stepconfig and adding a probe screen by verser.

i have to say the acceleration on my machine is only set at about 300 mm/s2 if i remember correctly, so maybe that is not enough and i should increase this to 600 mm/s2 (both the supplier of the machine and benezan have mentioned higher acceleration but i set it slower because it was scary when i ran the test in stepconfig)

BigJohnT wrote:

bschiett wrote:

i stopped by hitting the e-stop button. this is also how i change tools - press the estop button before i touch the spindle, to make sure nothing crazy happens as I put in the new tool.

Make sure you have VOLATILE_HOME = 0 in each axis section of your ini file if you're going to use this method. You will of course have to home after each loss of power which is the correct thing to do with a stepper. If you're worried about the spindle taking off put a switch on the spindle.

JT

thanks, will check ... with loss of power do you mean when the controller box is switched off? or do you mean when the e-stop has been pressed?

the spindle is hooked up to the VFD. I guess i could also switch off the VFD before a toolchange but pressing the e-stop is just more convenient and shouldn't be a problem if the spindle is not moving?

BigJohnT wrote:

bschiett wrote:

i stopped by hitting the e-stop button. this is also how i change tools - press the estop button before i touch the spindle, to make sure nothing crazy happens as I put in the new tool.

Make sure you have VOLATILE_HOME = 0 in each axis section of your ini file if you're going to use this method. You will of course have to home after each loss of power which is the correct thing to do with a stepper. If you're worried about the spindle taking off put a switch on the spindle.

JT

i checked my ini file and there is no mention of VOLATILE_HOME
so i guess the default value of 0 is in effect?

my machine does not have position feedback. assuming i want to press the estop button to keep the spindle from taking off when i do a manual toolchange, can i just leave it as VOLATILE_HOME = 0 or should the value be set to 1?

ok, so here is what i did:

start axis, home machine, load program, touch off to corner and top of fixture (no material in fixture) like i usually do. I touch off using a slip of paper which i assume to be 0.1mm thickness. I put my 10mm tool against the fixture with slip of paper in between and touch off x,y this way (enter -5.1mm).

z rel/abs = 0.1 / -84.567
G92 and TLO read 0,0,0

1) start and stop immediately after tool change popup (using stop button in software, not estop), jog back to top of fixture in roughly same area, lower onto fixture with paper in between

z rel/abs = 0.081 / -84.586 (diff with above is 0.019 / 0.019)
G92/TLO still reads 0,0,0

2) run program in air, stop halfway, go back to check offsets

z rel/abs = 0.168 / -84.500 (diff with above is 0.068 / 0.067)

3) run program in air, let run until done, check offsets

z rel/abs = 0.162 / -84.505 (diff with above is 0.062 / 0.062)
double check if x and y touch offs are still ok - they are

4) press e-stop while machine is not running, depress estop and power machine again (F2), check offsets

z rel/abs = 0.116 / -84.551 (diff with above is 0.016/0.016)

5) let program run, stop while spindle is moving in X direction (no Y or Z movement), using the estop button rather than the stop button in software, go back to check offsets

y, z are OK but x touch off now has a difference of 0.326 compared to its original touch off position!

6) rehome machine, check offsets again

y, z touch offs are still ok but x touch off needs to be redone (makes sense since it was already off in 5)

conclusion - i guess the problem lies with me hitting the estop button all the time to stop the machine midway programs and not rehoming or repeating touch off ... i guess i should use the ESC button on the keyboard or just click on the stop button instead of using the estop for something it's not intended for ... i don't have a pendant yet or control box with a big stop button, maybe i should look into getting one so i avoid using the estop button for what it's not intended.