Japanese style spindle orient, how to?

I have a dead Yaskawa spindle drive, the machine has a lame old control. Time to upgrade. I plan on replacing the spindle drive with a Hitachi, problem being the original drive has an “ orientation card” that provides spindle orient. The card watches a magnetic sensor which yaskawa calls a magneto. The spindle has a magnet mounted on it. 

  How would one recreate this as a hal component? It is not an encoder. Obviously the drive runs the spindle at a slow rpm and waits for the sensor to change, but there is more to this, there is no encoder and the original drive has some complex circuitry that allowed for adjustment for the stop angle.

There is no way I see to add an encoder, the gears can shift at any given place. 

Anyone whip this one yet ?




 

It will not be trivial to integrate the current magnet and sensor into linuxcnc. In fact, your original control doesn't know about it either. It simply sends a digital signal to the drive and the drive performs the orientation on its own.

You can keep the drive and sensors as they are and use it the same way with linuxcnc.

If you are going to replace the drive, you should go for a proper spindle drive with closed loop control and an orientation function built in.

If you're planning to use a WJ200, it has an internal homing (and simple positioning) function along with native encoder input terminals.  You'll need an encoder, but no add-on encoder board.

Maybe a shot-pin & disc on the spindle?

LCNC sends a digital signal to the Hitachi to drop the max current (to bare minimum), fires an air-powered, spring-loaded shot-pin, and commands a slow speed.

Another sensor (proxy?) sees the shot pin drop in the place and LCNC stops commanding the Hitachi to rotate.

All of the above can be set up in HAL using digital i/o (motion.digital-out.nn), and those are available via G-code.  A macro/subroutine named Mxx could run the g-code to orient the spindle.

JR1050 wrote:

There is no way I see to add an encoder, the gears can shift at any given place. 

 

You got pics or a diagram?  Maybe the hive-mind can come up with an encoder mount.

I had to look up what a magneto was in a sensor context. looks kinda like it can be used like an encoder depending how it was orientated

cakeslob wrote:

I had to look up what a magneto was in a sensor context. looks kinda like it can be used like an encoder depending how it was orientated
 

Is it another term for a resolver?

I'd guess the 'magneto' at the actual sensor level is either just a reed switch, or a hall effect sensor.
The complicated bit would be getting the spindle to stop under control at the required angle, especially if there's gearing involved.

If it's a simple case of stopping at the magnet, you just need to rotate at a slow speed and stop the spindle motor.
If however you need to stop at x degrees after the magnet, then things get quite a bit more complicated, but should be doable.

 

The magneto is not a resolver. It’s a magnetic sensor.

Replacing the drive with another drive that uses an orientation card is not remotely economically feasible. In addition to using a magneto for position sensing, the original drive used resolver feedback for speed sensing. Replacement equivalent is a 626 M5 , about 15k. I’m presently repairing the mk2 drive, again... 

While there is no place to put an encoder on the spindle itself, I can put an encoder in place of the resolver on the motor. This would give me the ability to count pulses , either the magneto could be used to find the initial orient position or it could be replaced with a hall type switch . After the switch has been first sensed, the drive can be instructed to run until the control counts the amount of pulses, which would be an offset amount, from initial position sense to actual position , then apply the brake. Since it’s an incremental amount, the encoder doesn’t need to be on the spindle itself. 

JR1050 wrote:

 After the switch has been first sensed, the drive can be instructed to run until the control counts the amount of pulses, which would be an offset amount, from initial position sense to actual position , then apply the brake. 

If your magnetic trigger doesn't work for you, I've had good luck with a plastic ring on a spindle using a set screw to trigger an inductive proxy.

I think you'll be better off leaving the homing inside LCNC even if your VFD has an internal homing function.  Assuming you set things up with motor encoder and spindle index sensor, I suspect you'll need a mux component to select from a number of different encoder count offsets depending on which gear you're in at the time of homing.

I think there are a number of gear indicator or change schemes described in the forum.  I'm not sure of the details, but I think LCNC can be set up so that the encoder counts are logged between index pulses, and a look-up table is consulted to determine which gear ratio has been engaged.

Once the current gear ratio is determined, a second look-up table is used to set the appropriate encoder count offset from the index signal during homing.  I suspect that the sequence would require the spindle to turn at least two revolutions to home: one to calculate the gear ratio, and a second to do the index-then-offset move.