g33.1 problem

Good morning. I wanted to ask for clarifications on a problem that I encounter both on a real machine and on sim/axis. The version is 2.9 pre0.
If I command m3 s800 and after g33.1 z-50 k1, the tapping is performed correctly, the spindle turns in m3, stops with deceleration after z-50 and returns to m4 until it exceeds the initial z in deceleration, and then returns to the initial z.

If I command m3 s1000 and after g33.1 z-50 k1, the tapping is not performed correctly, the spindle turns in m3, the z reaches -50 and no longer returns to z +, I send back with the spindle rotating.

If I command m3 s2000 and after g33.1 z-50 k2 the tapping is performed correctly, the spindle turns in m3, stops with deceleration after z-50 and returns to m4 until it exceeds the initial z in deceleration, and then returns to the initial z.

If I command m3 s3000 and after g33.1 z-50 k2, the tapping is not performed correctly, the spindle turns in m3, the z reaches -50 and does not return to z +, I return with the spindle rotating.

This value the spindle are on real machine, on sim have value for spindle velocity much hight, but problem is the same .
on linuxcnc 2.9.0 pre0 is the same at 2.10.0-pre0

On fanuc system, the spindle have 2 acceleration value.
One for velocity mode, when work in m3 or m4 and the second accelleration for tapping.

i have test change encoder resolution and mesa card , but the problem not is hw.
some test:
if change acceleration the spindle and is much fast , the m3s1500 is ok.
The acceleration value the spindle change reaction on return the z axis .

because ?

papagno-source wrote:

If I command m3 s1000 and after g33.1 z-50 k1, the tapping is not performed correctly, the spindle turns in m3, the z reaches -50 and no longer returns to z +, I send back with the spindle rotating.

 

Do you mean that Z makes an unsynchronized move to -50 and then just sits there?  Does the spindle reverse?

Have you checked the spindle-at-speed signal/pin during the improper tapping process?

I'm interested in your problem as I (and others) have noticed some issues with spindle synchronized moves.  Mine has cleared up somewhat from my first tests, but I suspect that LCNC's code for spindle synchronized moves (G33, G33.1, G76) do not calculate the appropriate spindle angular position for the start of synchronized movement.  There may be other problems, too.

Unfortunately, I'm not able to critique (or really understand) LCNC's trajectory planner code for tapping/threading - I've no real programming experience.  But a review of old github issues and threads on the Developers' mailing list indicates this may be a old - but still present - issue.

Here are a couple of recent threads you might be interested in reviewing:
forum.linuxcnc.org/38-general-linuxcnc-q...pping-problem#319620

forum.linuxcnc.org/38-general-linuxcnc-q...es-with-speed#317088

If you can offer any insights I'm all ears.

Thanks for support.

The problem can is test in sim/axis, is the same on real machine.
Set axis ini.file in mm and increase max velocity general and axis at 300 for example.

Start linuxcnc.

Commnad m3 s2500
after g33.1 z-50 k1.

the spindle start rotation at 2500 rpm .
The g33.1 start when index encoder is reached.
the momachine move at z-50 with correct velocity.
The spindle decelleration , but Z not move in decelleration ( not correct)
The spindle ramp in m4 , but Z axis not move ( not correct)

I have test on real machine and sim/axis, if ramp acelleration spindle is very short, the problem disappears with m3 s2500 and it could reappear at 4000 rpm. If you further reduce the spindle acceleration, making it almost instantaneous from 0 rpm to 3000, the problem becomes evident with the m3 s5000 command.

But these are just attempts, to understand where the problem is, in reality the spindle must have its own acceleration and cannot be instantaneous, in order to preserve the mechanics and electronics, especially in deceleration, since it will have to download the energy of the drive's DC BUS onto the braking resistors.

Interesting - almost sounds like the spindle is 'decoupling' from the synchronized z-axis when it reaches the programmed z-value (-50).

This is not the exact behavior I was seeing, but I also wasn't trying G33.1 at such high spindle speeds or accel values on my real machine.

It may help troubleshooting if you can set up a halscope that shows the z-axis position command and spindle velocity during the decel-stop-accel phase (M3-M5-M4) at different spindle accel settings.  Maybe we can see the point where the z-axis stops doing what we expect it to do.

Hopefully someone more experienced than I am can help diagnose, or confirm that this is a code problem rather than a config issue we're both having.

Side note - the behavior you describe is almost identical to my Emco 325-II lathe running Emco's version of Fanuc 21TB.  Emco's manual claims the lathe can rigid tap, but after a number of broken taps and some investigation using a high-speed camera, I discovered that the control stops synchronization of the z-axis when the spindle stops to reverse.  It doesn't 're-couple' the z-axis until the encoder index signal is triggered during the M4.
So depending on where the spindle stopped there could be up to 359 degrees of spindle rotation with no Z-movement while the tap is still in the hole.  I've had to switch to tension-compression holders to stop breaking taps.
Ironically, Emco's version of Siemens 840d and Operate rigid tap fine...with the exact same hardware.