Background:
I’m doing a senior design project related to research on “flutter”. Flutter is the out-of-plane motion of a blade when it spins on something such as a wind turbine. This starts to occur at specific rpms depending on specific parameters and geometries of the blade. The blade can be modeled as a cantilevered beam with the hub being its fixed point. I’ve mounted a single strain gauge on the blade (blade is made of PLA) closer to the hub in order to obtain a sampling frequency >500Hz if possible from this device.
Goal:
I’m attempting to record microstrain in the blade when it flutters in a half-bridge configuration (another strain gauge will be mounted eventually, but want to nail down the quarter-bridge setup first). A micro SD module will also be configured into the setup to record the strain without connecting to a computer (this code has been written and tested, but not yet implemented with strain gauges).
Electrical components:
Arduino Uno with a “RB-Onl-38 Load Cell / Wheatstone Amplifier Shield” (SOURCE).
Quarter-bridge configuration with three 120 ohm resistors and one 120 ohm resistor strain gauge.
Problems encountered:
I plugged everything in and ran the program found on the “RB-Onl-38” website under ZIP Files --> Arduino Library. I deflected the blade with about 10lbf (with deflection of ~2 inches off of a ~12in moment arm) and the ADC value on the Serial monitor read a change of about 3. I adjusted the potentiometer completely clockwise otherwise the ADC values maxed out at ~670. The sensitivity to strain of the amplifier in my current setup is too low for me to see a small microstrain change. I’m unsure if I need to change the gain (default is 495 I believe) and how to change the gain on the “RB-Onl-38”? Or am I not using the amplifier (AD8426) correctly?
From my understanding, and research from other forum posts, the RB-Onl-38 was only tested using load cells, not strain gauges (SOURCE). However, it should be possible to configure a Wheatstone bridge and apply the same principles.
Please make sure, without your wheatstone bridge circuit connected, to adjust the reference voltage potentiometer of the channel you are using on the shield to obtain an ADC Raw value of around 338, which corresponds to about 1.65V of reference voltage for the amplifier.
Therefore, in theory, if the output voltage of the wheatstone bridge circuit with the strain gauge deflected is 1.638V, you will have a maximum ADC Raw value reading of about 669 :
1.65V-1.638V = 0.012V
0.012V x 495 (default gain of the shield) = 5.94V
Therefore, the on-board amplifier will saturate at its VCC which is 3.3V and corresponds to 669 ADC Raw value.
I changed the potentiometer back to its original starting position to obtain an ADC Raw value of ~338. From my interpretation of the math, the default gain of the shield is too high because the voltage difference becomes saturated (above 3.3V), upon which the shield is unable to interpret the voltage from a higher deflection. Does this mean I need to change the gain value on the shield? Upon doing the research, it looks as though I need to solder a resistor based off this (SOURCE p. 21) from “gain 1” to “gain 2” on the shield. Is that correct?
That’s correct, you can change the default gain by changing the gain resistor that corresponds to the channel you are using based on this formula : Rgain= 49400/(Gain-1)
More information in the gain section of this tutorial. Since the gain resistor is an smd component, please note that replacing it will require some smd soldering skills.
Hi! I’m looking to swap the smd resistor on the RB-Onl-38 with another smd resistors that gives the gain I’m looking for. What is the power rating (length/width) of the smd resistor that comes with the board? I’d like to order one that has the same power rating, just with the resistance value I need.
Hi, I’m having trouble understanding how voltages get mapped to the raw ADC value.
In my application, I have a load cell that measures weight in 1 direction. The wheatstone bridge voltage output (Va-Vb) is 0V at no load and .0066V at max load. This gets amplified by the gain of 495 to yield a voltage range of 0V-3.267V.
How is that voltage range translated to the raw ADC values?
Should I adjust the potentiometer so that I get a raw ADC value of 0 at no load and a raw ADC value of 1023 at near max load?
In the WheatstoneBridge library example code, I saw:
WheatstoneBridge wsb_strain1(A0, 365, 675, 0, 1000);
Why is the ADC value limited to 365-675 in that code example? Why not 0-1023?
The gain of the instrumentation amplifier IC which is set by default to 495 with the 100 Ohms smd gain resistor for the 2 channels. (this can be changed by changing the gain resistor based on the formula: Rgain= 49400/(Gain-1) ).
The reference voltage can be changed with the on-board reference voltage trimpot and that can be used to offset the output signal to a mid-supply voltage to be used with a single power supply ADC.
Let’s take your example:
As you mentioned, your load cell outputs 0V at no load and 6.6mV at maximum load.
The amplifier has a default gain of 495. Therefore the voltage output from the amplifier chip to the ADC of the Arduino at full load should be 495 x 6.6mV = 3.267V
The Arduino ADC has a 10-bit resolution, so over a range of 5-0 V DC, it would have a theoretical sensitivity of ~4.9 mV.
Using the reference voltage trimpot, you can set the output voltage from the amplifier chip to 0V (while no load on the load cell).
So with an output of 3.267V (at maximum load), you will obtain an ADC raw value of about 668.
To have a better resolution, you can connect the 3.3 V DC to the AREF pin and use this function (external) in your code. Also, please note that the Arduino boards do not include a decoupling capacitor on the AREF pin, so we recommend that you add one when you connect it to the 3.3 V DC source.
Therefore, an output of 3.267V (at maximum load), you will obtain an ADC raw value of about 1012.
The load cell’s excitation voltage which is 3.3V on the load cell shield (coming from the Arduino’s 3.3V). Some load cells require higher excitation voltage (5V-10V). In this case, a workaround would be to bend the 3.3V of the load cell shield and connect it to the Arduino’s 5V pin.
