Using an index pulse with stepgen

So, I'm running a CNC with 3 linear axes and a spindle. The spindle is hooked up to the encoder input on the board, so I run it in velocity mode with a PID. Problem is, I would like to home to the index pulses on the encoders for the other axes. I've got 24V index signals all set up and connected to an 'encoder' hal component, and then to the axis.N.index-enable pins. My home sequence should go like this:

1. Go the the home switch.
2. Back off the home switch to the next index pulse.
3. Stop and set the machine coordinate to 0.

but I'm getting:

1. Go the the home switch.
2. Back off the home switch to the next index pulse.
3. Move to machine 0.

Essentially, the machine coordinates are not being reset, so when the index-enable signal is released, the coordinate is updated with the old value. I think I'm having a problem that is similar to this thread: homing and encoders

So, my question is: how do I use an index pulse to home the axes without actually closing the loop with the encoder? All I have is the index pulse, and what I need to do is reset the stepgen/machine coordinate system.

I think your error is in your ini settings for the axis in question.

If you want the 1st index pulse point to be set as the machine zero, and stop there, the ini home and home offset settings should be
(This is assuming all of your other ini settings are correct.)

If you need more assistance, or have more questions, it will help if you share your ini file.

Todd, thank for the feedback!

The INI file is correct, but here it is for your viewing pleasure, along with a video of the procedure, including some watched HAL pins.



Hope that helps.

What happens if you restart Linuxcnc?

My one machine that I home to index on, isn't running right now for me to compare behavior. (Took it down to move and relocate it in the shop, and haven't finished that project yet.)

Why isn't the encoder.counts changing? Something is goofy.

Could you share the hal file as well?

What hardware are you using to read the encoder?

Thanks, everyone, for your input.

I've concluded a few things:

1. When the use-index setting is YES, the encoder.N-index-enable pin is primed when the home sensor is tripped.
2. When the use-index setting is NO, a tristate_bit hal component is needed to duplicate this function.
3. The home sequence numbers are different for both settings of the use-index (INI file) parameter. With the use-index, the final home state is 21, without, the final home state looks like 14.
4. When the use-index setting is YES, the encoder.counts is not reset on the home sensor. Hence, unless the axis is set up with an *actual* encoder, the index-enable pin will not do anything. The encoder hal component does not connect to the stepgen automagically.

With that said, the tie-in of the index pulse and the home sensor works very nicely. I've got a comp.N hal component (to determine the home state), the index-enable signal, and the actual home switch set as inputs for the lut5 component, for which the correct value was 0xFFFFFFF8. The behaviour of the homing process is now:

1. Go to the switch, prime index-enable.
2. Back off until the switch has released, and go to the first index pulse in the same direction.
3. Reset encoder counts and drop index enable at the first index pulse.

There are three duplicates of this system set up, so each axis has its own.