Strain gauge touch probe for CNC digitizer

[harvie]

How to wire Strain Gauges without canceling each other?

Hi,
i am experimenting with various CNC touch probe designs and i've came across following setup using straingauges CA-glued to elastic frame holding the probe tip. I will probably use aluminum or acrylic as base for the strain gauge.




I wonder if there is a way to wire this using single instrumentation amplifier channel. (i am going to use HX711 inamp with integrated ADC).
Problem is that i am afraid that under certain angles of side load, the strain gauges can cancel each other when placed in single wheatstone bridge. Unlike personal scale (which has all strain gauges loaded in the same direction by person standing on top of it) these touch probe strain gauges will be loaded in both compression and tension randomly based on direction of touch. If this happens, the CNC will continue moving towards the probed object and break the probe. I need this to be as sensitive as possible without being affected by stepper motor vibrations and intertia of probe tips.

I don't really need to use the 3 beamed star topology showed in image. I can use any other strain gauge placement as long as it's at least kind of symetric. But i want to produce strain gauge based needle that will be sensitive to mechanical touch from any direction (both axial and radial). I can even use some kind of multiaxial strain gauge. But i am bit lost in the choices. Also they are usualy much more expensive...

I need this to work with sharp tip to have enough resolution. But i don't want to puncture the scanned object. I've tried with cheap chinese weighing scale. It's based on load cell and sensitive enough to react when i press my finger against the tip of the needle needle without puncturing my skin. So i think the strain gauges should be fine for this application. But i need it to work in all posible probing directions.

HX711 has two channels. I might use both. But i don't know if it's good idea to use 1 strain gauge in one channel and 2 gauges in other. That might introduce inconsistent results when touching from different directions. Also i might use two and two gauges in "X" configuration, but then again i would be afraid that the can cancel each other out...

Any ideas on this one? Thanks.

 

[ZASto]

What about piezo disc?

 

[FvM]

I presume the output of the shown sensor arrangement can be transformed to cartesian xyz coordinates by a linear operation. A signal that represents force in any direction requires however nonlinear operations, e.g. determine the magnitude of a vector.

 

[ZASto]

Again,have you considered piezo disc?

This is a principle of operation of Renishaw touch probe


from
https://resources.renishaw.com/en/download/user-guide-touch-trigger-probe-systems--99347

The metal rods and serially connected balls could be replaced with piezo disc and suitable high gain opamp+comparator.
But, 3 wires , as your idea, two for power and one for output
Renishaw only has two wires, acts like a NC switch.

 

[harvie]

> What about piezo disc?

Yes, i've considered this. Problem with piezo sensors is that it only reacts to change in pressure, not the pressure itself...

> I presume the output of the shown sensor arrangement can be transformed to cartesian xyz coordinates by a linear operation.

I don't need that. It will be mounted to cartesian robot and it will be moving in Z until touch is detected. Once touch is detected it will record Z height and move to scan another point in XY plane.

- - - Updated - - -

This is interesting: https://hackaday.com/2019/01/27/quartet-of-smd-resistors-used-to-sense-z-axis-height/

 

[harvie]

These guys even designed load cell / strain gauge into the PCB design:

https://www.duet3d.com/DeltaSmartEffector
https://reprap.org/forum/read.php?178,783399



> This is a principle of operation of Renishaw touch probe

I already have one such probe based on this ball-cylinder design. But i am not completely happy with how it turned out for me. Don't get me wrong. It's cool probe and i can reccomend it for most purposes. But i would like to experiment with different designs.

I've made it using PCB and some hardware The rods are from inside of terminal blocks. Balls are heads of screws. Two boards are held together using 3 adjustables riser screws (preloaded by the main spring). But in hindsight i would use 4 screws for easier adjusting in 2 axes. With 3 screws it's hard to adjust, because once you get one axis right, you will screw it up while trying to fix the other...



I have following goals:

• Has to be able to scan non-conductive surfaces (rules out simple continuity testing)
• No moving parts (rules out renishaw design)
• Has to stay active if still touches after machine stoped (rules out piezo-based design, except for self-oscillating)
• Need at least 0.1mm precision. (have to use sharp probing tip, but i cannot use lot of force, because such tip would scratch my delicate surface. this is real problem i had with spring in renishaw design)
• Have to be easy and cheap to make at home. (PCB based designs are cool!)

 

[harvie]

BTW the renishaw design is based on something called "Maxwell Kinematic Coupling"
https://en.wikipedia.org/wiki/Kinematic_coupling#Maxwell_Coupling