Wow. This is actually pretty ingenious.

(ThantiK) #1

Wow. This is actually pretty ingenious. Using large SMD resistors as the strain gauge elements. Now…the problem is that I don’t see where I can get an HX711 in a smaller package (or at all for that matter)… I seriously want to replicate this.

(Ray Kholodovsky) #2

I don’t quite understand how this works. Does the flexing cause the resistance to change? Wouldn’t that be bad on the solder joints over time?

(Michael K Johnson) #3

I would expect that the copper layer would be more likely to delaminate from the FRB than a good solder joint fail.

(Ryan Carlyle) #4

Solder fatigues but you may get enough life for the application.

(Alex Wiebe) #5

I believe this is just clever “circuit bending”, right?

(Richard Karlson) #6

@raykholo the flexing is not an issue with metal fatique as the actual travel distance is probably only in the thousands of an inch. Also if they are set up like the O-haus scales I’ve worked on, only one of the 4 legs of the Wheatstone transducers are subject to the load thus giving you a differential reading.

(ThantiK) #7

@raykholo yeah - flexing of the resistors causes the resistance to change ever so slightly. By using some clever amplification circuitry you can detect enough of the change to measure it. This is a typical strain-gauge circuit but used in a clever way; given that the resistors themselves are manufactured with such high tolerance and on a mass scale.

(ThantiK) #8

By the way, there is not enough strain on this circuit to ever cause a fatigue issue. The “Smart Effector” made by Duet3D uses the same effect for their touch probe. I’m even adopting it at my office for ours. SeemeCNC uses the same for theirs. I’m looking at this though, and it may simplify my design. Instead of meandering traces on a PCB, I may simply be able to create some cutouts on the PCB and bridge these resistors across the cutouts - freeing up valuable space for connectors, lights, shielding, etc.

(Ray Kholodovsky) #9

So the resistor body itself being flexed is what causes the change. Got it. Cool stuff.

(Brent Crosby) #10

I wonder if this could work for your amp:
I2C interface, and 24-bits.

(Gary Tolley - Grogyan) #11

Flexing these types of devices, even as slight as suggested, will cause the device to fracture.

Best to use the right device for the job. There are flex strain resistors which would be better

(Ryan Carlyle) #12

I’d like to see some cycle testing. For example if it can do 10,000 probe cycles before failing, it’s probably reliable enough.

(Stephanie A) #13

Increasing the pad size would probably help. You can likely use a ceramic capacitor instead of a resistor too (measure frequency instead of resistance). Though I’m sure that would come with its own complications. I saw someone suggest a cutout under the resistor, but that can cause strength issues.
Lots of methods to experiment.

(Duncan Gunn) #14

@Stephanie_A It might work better, or fractures in the ceracmic capacitor could cause a sudden failiure. Those things are quite brittle.

(Alan Lord) #15

Take a look at Precision Piezo. I’m using a piezo disc and getting +/-10 micron probing accuracy.

(Raptor Swire) #16

thats quite interesting

(Griffin Paquette) #17

@raykholo its a Wheatstone bridge setup. The strain causes the resistance to change and therefore the potential across the bridge changes as well.

(Griffin Paquette) #18

@ThantiK I wonder if you could just use a standard OpAmp with a high gain to amplify this? Then there would be no need for the 12C interface.

Edit- This can definitely be done using an op amp and just triggering a transistor to pull the endstop to ground. Interesting… Might have to make one of these to try out.

I have a few machines at school here so we could figure out how many cycles it takes to see any degradation. I bet your nozzle wears out before the bridge does.

(Stephanie A) #19

@Griffin_Paquette you wouldn’t even need to send the analog signal. An op-amp and a Schmidt trigger would make it just like a limit switch.

And You could likely just use the pcb and omit the resistors all together. Trace resistance.

(Griffin Paquette) #20

@Stephanie_A I was thinking that as well with the PCB instead of the resistor. My concern with the resistor setup is that if a resistor is misaligned at all, there could be serious deviation.