[gernoff]

Done quite a bit of searching....... I have a 24" Mahr Motorized Height Gauge that I would like to retrofit to LinuxCNC. The existing glass scale is damaged and I cannot get repair parts from Mahr. Very low current draw on the DC motor, the original system would run for 8+ hours on 4 D cell batteries. Anyway, what would be the easiest servo implementation via LinuxCNC? The retrofit will include a 24" linear encoder (TBD) and we are shooting for sub micron resolution. My intention is to get the system driving via the software then start building macros to be able to duplicate the gauges original measurement routines.

[alan_3301]

It should be pretty simple, you may need some hardware for encoder counting and to give the +/- 10V for the servo amp (I'm guessing it's analog)
But once you get going it shouldn't be too difficult.

Can't you just find a scale to replace the broken one though? You could find one that will work even if not a direct replacement.

[alan_3301]

How does the original control work?
Does it drive the height gauge down and stop automatically when it touches the part?
What kind of motor electronics does it have already.
Maybe it just needs a DC motor drive that can sense current and stop the motor when it reaches a certain point.
Then just have a DRO connected to the scale to show the height.

Maybe using linuxcnc would just overcomplicate things, you don't really need a pc or an interface for this do you?

[gernoff]

Your questions got me to dig into the (very basic) users manual. I found that Mahr claims a constant 1N gauging force. The small DC gearmotor is attached to the metal band drive via a slip clutch.
Based on this paragraph:
"When contacting the test piece, the height measuring
instrument generates a constant force which acts upon
the stylus until it stops bouncing and delivers a constant
measuring value. As soon as the DX 1 accepts the value,
the measuring force is deactivated."
I am now assuming that the controller simply drives the probe into the measurement feature and watches the encoder output until it stabilizes.

I am still interested in using LinuxCNC to build an interface to replicate the original measuring abilities of the gauge. Below is a link (my google drive) to a PDF outlining the controllers built in measurement routines:

CX1 Measurement Routines

A retrofit of the glass scale will probably not work. It seems as though the onboard memory of the controller is corrupted. whenever I attempt a measurement, I get the same 'wierd' reading. Forget what is is but something like -9654,8867. Most of the glass scale looks to be OK but in the first inch of its travel, all of the etched lines have been worn off. Try to set a zero elsewhere on the scale then begin a reading, alway get the -9654,8867 reading......
I am taking an O-scope home this weekend to look at the output of the scale as well as the input of the motor.

[alan_3301]

If you can find a pinout of the glass scale connector, it shouldn't be too hard to tell if it's working.

All of those probing routines should be easy to make gcodes for.
If the slip clutch lets the motor keep running, with 1N force, while the control waits for the measurement to stabilize, you could make that work.

[andypugh]

Anyway, what would be the easiest servo implementation via LinuxCNC? .

Interesting question. I think that this could end up being a HAL-only config, possibly with a custom HAL component to halt the motion on contact.
(I don't see anything to be gained from having any G-code as part of the system in this application)

I suspect that a PID component driving the existing DC-motor directly, then tripping out when the PID-error term gets too large would work.

[gernoff]

Had more time to think about this........ Here is what I'd like to do:

I too think it can be HAL only. I'd like to build a spring loaded strain gauge contact point. I can then easily calibrate it to an exact touch force. Spring load allows distance for the system to decelerate. That is, sys drives the probe at high speed until it senses strain buildup, slows and PIDS it way to a specific (adjustable) strain reading, then displays measurement.