I have a Teensy 3.2 MCU (datasheet) hooked up to an ADXL326 analog accelorometer breakout board (datasheet). Visual examination of the board reveals a few SMD capacitors, which are 0.1uF.
I have successfully used a scale value to convert a value to a G value using advice from the manufacturer's forum. For example, when the accelorometer has no force on an axis, it will read 0G. When I apply gravity force, it will change to a ridiculous value like 7G. When the force is removed, it will not go back to 0G — it will go to a random value from -4G to 4G.
Research has revealed that the accelorometer has an impedence of 32kOhms. See the below image for details on the MCU impedence:
Is there anyway to reliably go from a analog input to a G value using this setup? This is going to be used in a vehicle for data recording during impacts. I have spent $18 on this analog accelorometer, and would not like to purchase a digital accelorometer unless absolutely necessary.
Disclaimer: The truth is, I have no clue what any of this impedance stuff means. I just determined this was the most likely issue from here. They stated that the likely cause is difference in impedance combined with interference from the MCU clock(s). I am not an electronics engineer, I am a software engineer with some basic electronics experience.
Best Answer
The output impedance of 32 kohm can be a significant problem when attached to the ADC input on the MCU. Table 24 entitled "16-bit ADC operating conditions" states that the input impedance is typically 2 kohm and this will screw up any measurement from a source that has an output impedance of 32 kohm.
You might get away with a parallel capacitor - the sensor data sheet specifies there must be a minimum of 4.7 nF on the sensor output so have you done this? Alternatively you might need an op-amp buffer.
The table in the question suggests that you might be using a PGA in front of the ADC - please confirm that you have enabled this because it makes things easier but you will still need a 4.7 nF capacitor and normal decoupling across the power rails.
Reading further down the PGA table suggests that the analogue source impedance should be typically 100 ohm and this is way lower than what the ADI device is. I think a buffer might be a good idea.