And that's why it keeps turning off the 220v going to the vfd when I close linuxcnc to pickup the new settings I'm working on in the HAL and ini. For me it's a lot of brute-force and ignorance working in there sometimes and trial & error to get things working. (And a whole lot of help from you guys, thank goodness)

I'll hot-wire the SSR to keep the vfd from having to reboot every few minutes until I get this somewhat sorted out. I think I've at least figured out what's causing my gear function, that'll be the next series of questions lol
Thanks again
M

Servo Tuning advice needed!

Attached are the oscilloscope reading from our 4 axis mill, we are using mesaboards to retrofit a bostomatic machine. Any tips on tuning the servo parameters would be great.  The readings are all from the X axis we had it doing 2 inch movements at the same velocity. If anyone has pointers on PID values based on the readings that would be so helpful.  Also all the other values are 0 unless otherwise written in the slide. 

 

Probe in Sim works okay... now comparing the various files.
Thanks for the suggestion.

Forgot some important safety stuff...

M254 also sets an M5, as well as turns on spindle.N.inhibit.

M255 also sets an M5, and turns off spindle.N.inhibit.

I got caught out by this during testing when I left an M3 command active and switched to C-axis.  When I switched back to spindle mode and gave it an Sxxxx RPM value it immediately started rotating.  That was a pants-fudge event...

As you go through my HAL file, you may notice some partly-related but (perhaps) unclear functions/logic.  One of these is a spindle disable/freewheel feature which I use when indicating a part or toolholder.

My particular spindle motor (SZGH) does not permit freewheeling the motor in velocity mode if it's enabled.  Simply disconnecting the drive enable whenever you want isn't good enough, because the C-axis freaks out with a following error once you start rotating the chuck by hand.  So the disable function has a bunch of checks to make sure C-axis is paused, and there are no other backstage things trying to force the motor to spin the moment you turn the freewheel function off.

No, the machine starts up in C-axis mode.  Actually, it may be more useful to think of it as the C-axis is always 'on', but paused when the spindle or orient is engaged.

• Start LCNC
  • C-axis is not paused, encoder disconnect not active
• Home all axes
  • axes-are-homed triggers C-axis homing
• C-axis is-homed signal triggers M255 via one-shot and halui.mdi
• M255
  • Checks if C-axis is homed, abort if not homed
  • G0 C0
  • Disconnect encoder from c-axis joint/axis and PID
  • Disable C-axis PID
  • Change stepgen settings to spindle values
• Now in spindle mode, all automatically.

To get back in C-axis mode:

• M254
  • Checks if C-axis is homed, abort if not homed
  • M19 orient spindle to 0 position
  • change stepgen to C-axis values
  • reconnect encoder
  • enable PID
• C-axis is on, and positioned at C0

Minor tip: one thing that frequently trips me up when copying/using other people's files is forgetting to change the M-code (and other) path(s) in the INI file.

Thanks! If I understand correctly you load the machine with spindle in spindle mode, rather than Caxis mode? That makes sense and is one of the things I would've wanted to achieve too, I'll be definitely taking notes from your files Once I figure out why the mcodes don't seem to work for me even though the directory and files seem ok, probably some other silly thing I overlooked. Gotta rtfm. I will have a chance to play around with the lathe in a few days and will report the results

Hi PWC  and many other folk. After many months I am having another go. I took the advice from this thread, and installed new proximity sensors, and added home positions. The issue that  now have is that it homes the z axis correctly, homes the X axis correctly. The y axis homes in sync and in tandum (y1 and y2 gantry motors) as seen in the first photo. It completes the four part homing procedure as seen in the second photo. It shows a capital H for X, Y and Z axis and it also shows the correct Home locations. However it also shows an error - exceeded negitive softlimit on joint 2 . So even though it seemed to of homed correctly , it disables the machine, and I can do nothing with it. 

Any suggestions are greatfully received.  

PS: I have tried endless permutations. The current positive and negitive values are the only ones that make the four step process work. The software suggests that I jog the axis off the joint but its part of the gantry pair so wont do it.

Some think that circles and ellipses are polygons as well.  So I added a ellipse pocketer subroutine  (which does circles too) to the program, and here it is, I also "sharpened up" the  vertices of the polygons a bit. See the screen dump for how well the "half ellipse" fits with the square.

RotarySMP wrote:

That forty year old hall sensor module is already cracking, no way I will risk mechanical contact with it.

More powerful magnets may trigger the halls even with a bit of distance.  Try triggering them with a good (N50) neodymium magnet and see how much spacing you can get away with.  You might be able to stay with a fixed magnet and nothing touching the fragile hall ring.

If they need to be closer than the fixed position allows, you could cut a protective surface out of plastic sheet, glue it to the sensor ring, and let the magnet rest on that instead of the halls.  Basically reinforce the hall ring - which it probably needs anyway - and let the reinforcement take the wear.

With a light spring there shouldn't be any real up/down impact forces on the hall ring... just the magnet sliding along.

That forty year old hall sensor module is already cracking, no way I will risk mechanical contact with it.

And another thing...

The existing trigger magnet is a really old ceramic magnet.  You could potentially use a newer, much smaller, neodymium magnet with a square shape to get a cleaner rising edge from the hall sensors.

Or stick with a round one and fiddle with the alignment and possibly carousel.comp's decel-time setting.