Schaublin 125-CNC retrofit.

Thanks for the link.
On the Maho a friend of mine who is a professional programmer wrote the gearbox comp. Although a wrote the spec and drew a flow chart of what it needed to do, I suck at coding, and can't really follow it to modify it , and my friend got too busy.

I kind of want to see if I can do it in classic ladder, as it seem more intuitive to me than C.

Thanks for your help guys, and especially to Andy for those lathe macros. Brilliant.



Cheers,
Mark

Mark,

I'm sure you're knee-deep in MotoGP, but when you come up for air here are a few thoughts about the EMI issue.

• Quick & dirty fix: add some debounce to the X-input in HAL
  • Retest and see if the problem goes away
  • Home your X-axis and see if the debounce delay is acceptable.  You aren't using the home switches as precision locations anyway (i.e they aren't encoder index pulses)
  • Do the same thing for all the other halls and test them too.
• Your 24v-5v relay board has optoisolators on it, so even if the contactors snubbers aren't working properly you aren't going to get overvoltage alarms on the Mesa outputs  - i.e. the first sign you've got a dead/missing contactor flyback diode.
• Your variator contactors appear to be the only mains voltage components turned on & off while the machine is in use.  Your spindle contactor is turned on and the VFD does the on/off thing until an estop.  So that one hasn't potentially caused a hall sensor spike when LCNC is monitoring the sensor.
  • Step 1 - disconnect the A1/A2 wires from both FWD/REV contactors and manually trigger them.
    • Don't forget that your limits will be disabled when doing this test
    • If it still faults, you've confirmed that it's the power side of the contactors causing a problem and not the control side (contactor coils doing something wierd through the 24v side)
  • Step 2 - switch the VFD's contactor with one of the variator's contactors and see if the same thing happens
    • If it still faults, it's unlikely to be a 'bad' contactor or dirty contacts and you can rule out those components as defective.
• Assuming you've narrowed it down to power side and the contactors themselves aren't defective, the motor winding field collapse is the likely source of a voltage spike in the hall wiring.
  • You mentioned in the YT comments that Schaublin used shielded cables for all the motor leads.  Are the shields still connected to ground? If so, where?
    • It looks like all the mains wiring is unshielded and in the same raceway as the signal wiring (left-most raceway on the door).
    • What did you do with the shields when you reattached the door?
  • Is it always joint0 (X-axis I presume) faulting?
    • If I recall, your halls cables are all running through the same conduit from the saddle. This suggests that the X-hall cable is picking up the spike inside your cabinet and not outside.
    • Are you using terminal blocks for your halls, or are they direct to the Mesa?
  • The two above items suggest (to me) that the 3rd test should be to physically separate the signal wires from the variator mains wires.
    • Disconnect the X-halls inputs from the Mesa and pull the cable out of the raceway all the way to where it enters the enclosure.
    • Reconnect the cable to the Mesa, but keep it out of the enclosure and as far away from everything as possible. Use some tape if necessary and tape the slack (you've got some slack, right?) well away. If you don't have slack, add a jumper cable so you can get the cable 8-12 freedom units away from everything.
    • If you're feeling frisky, scrounge an external shield braid from your rat's nest of spare wiring and slip it over the halls cable.  Connect the Mesa end of the temp halls shield to the nearest ground point.
    • Bonus points: do the same thing for the mains wires.
  • Re-test.  If this fixes it, you know what to do:
    • Re-wire the variator mains using shielded cable all the way from the motor to the contactors (or add an external shield)
    • Same-same for all halls/sensor cables
    • You still need to connect your tool-changer and that has (potentially) a few sensors in it which may also be affected by the variator contactor spikes. 
• Additional solutions
  • Get rid of the contactors and use a VFD to control the variator.  This will stop the on/off contactor spike during operation, and you won't need a complicated EMI suppression system because you aren't running it variable speed - just on/off through the VFD - and Tommylight's ferrite ring suggestion will probaby be sufficient on the VFD output.
  • This has been covered ad nauseum in the forum and YT comments... but get rid of the variator and install a pulley set that gives you an appropriate speed range.  Just use the gearbox and VFD.  Or disconnect the variator control and manually set it at a ratio that makes you happy.  Just let it wear a groove in the pulleys and ignore it until it breaks and then replace the whole thing.
    • How much torque do you really need?  You don't need buckets of torque at high rpm.
    • Vastly simplified control system - especially for CSS
    • Simplified mechanical system

Sorry for the long-winded post.  Lot to think about.

Thanks a lot for that, it is an excellent check list. That really helps me.



Did you watch the MotoGP? Both the Moto3 and the MotogGP races were excllent. Real cliff hangers. I was really glad to see Aleix take another victory. Brilliant.

I figured it best to time for all the comments to come in and see where things shake out. Then I went for a run and did a bunch of honey-Do's, so I still haven't started the trouble shooting.

I have LinuxCNC set up to home to those hall switches, back off, slow home, and then finishing homing on the next encoder index, so the idea about adding debounce could be a last resort if I can't suppress the spike.

From the motors back to the electrical cabinet terminal block strip at the bottom, it is original shielded wire from Schaublin. In the cabinet, I crimped a pig tail onto the shielding, and ran that through terminal blocks to the main start ground.
There is no shielding in the electrical cabinet raceways, and I did run these signal wires in the same race way as 3Ph. I will do the contactor check you recommend, and also reroute the signal wires around a different raceway regardless.

Yes, the hall sensor lines  for both axis, and the tool changer are in the same (shielded) cable. At least to the door terminals. It is always X Joint faulting, but that could be due to having no limits on Z. For a lathe, with the variations of chucks, tailstocks etc, it seems Schaulin was no along in only using soft limits on Z. So the Z hal sensor line might have those spikes, but LinuxCNC is not looking at that sensor except during homing.

Removing the variator is work
Thanks a lot for giving me a cooking recipe. I will report back.
Cheer,
Mark

Motor shielding doesn't do you much good if the conductors from the terminal block to the contactors are radiating like mad in the raceways... but I suspect you know that, too.

If you're lazy (like I am) before you touch any wires set up a limit in HAL on Z too and see if that faults when the contactors are deactivated.  Then do the debounce and see if it fixes both when manually activating the contactors.

Can do both sitting down...

MotoGP - I stopped watching sports on TV about 30 years ago, and stopped watching TV entirely about 20 years ago.  I've got friends involved in car & bike racing I'm so still interested in the technology... not so much the events.  Quite seriously, I'd rather go to the dentist than watch a sporting event (unless one of my friends is competing and I can grab a wrench).

i was thinking about the variator, you could setup a stepper and probably as a joint or something, then use a macro to control the steps.

or you could write a hal component, which, if you know the steps to speed ratio, it'll be fairly simple.


I'd be happy to help with a hal component if you want one

I appreciate your offer thanks, I will see if I can acheive what I am thinking in ClassicLadder, but if that is beyond my ability (highly likely), I will take you up on that.
Since there is so much overlap in control authority between the variator and VFD, I should only need very rough speed control over the variator. So rough, that if I use open loop, timed dead reckoning between 4 "gear ratio's", that should be plenty adequate to cover the entire range without gaps. Basically the idea would be an "8 speed gearbox" (Variator plus backgear), and the VFD running closed loop over the top.
Adding a stepper would be a lot of work, and would seem a pretty big step down from the high quality Schaublin Variator drive.

 put a encoder on the motor  use it  as a  servo, is the variator drive proportional control or just contactors ?

I have zero familiarity with maho machines

This Maho is a Schaublin It is a little baby 3Ph motor. Very cute.

RotarySMP wrote:

I appreciate your offer thanks, I will see if I can acheive what I am thinking in ClassicLadder, but if that is beyond my ability (highly likely), I will take you up on that.
Since there is so much overlap in control authority between the variator and VFD, I should only need very rough speed control over the variator. So rough, that if I use open loop, timed dead reckoning between 4 "gear ratio's", that should be plenty adequate to cover the entire range without gaps. Basically the idea would be an "8 speed gearbox" (Variator plus backgear), and the VFD running closed loop over the top.
Adding a stepper would be a lot of work, and would seem a pretty big step down from the high quality Schaublin Variator drive.

Have not been following along but here is a rough component for  a gearbox that may help.  Its based on a 2 belt gearbox and I used a digital tacho to measure the ratios. There is a matching panel I wrote (with massive help from Chris) which is part of qtdragon examples as there is no way to sense which gear it is in. now. linuxcnc.org/docs/2.9/html/gui/qtdragon....ed_custom_vcp_panels