In Pressure sensor miswired, I talked about getting a lot of low-frequency noise from the amplifier, that was not related to simple AC coupling. I tried recording the signal using the Arduino, rather than looking at it on the oscilloscope, and I saw even larger noise than before (the equivalent of about 200µV of noise at the sensor—way larger than the signals the pressure sensor is intended to read). I tried reducing the length of the twisted pair wire from the sensor to the amplifier and saw essentially no improvement.
After this morning’s meeting, my co-instructor suggested that it might be power-supply noise, since I was using the Arduino power supply to power the amplifier and the pressure sensor. This made some sense to me, particularly if running the Arduino ADC increased the noise, so I tried powering the sensor (but not the amplifier) off of a battery.
Poof! the noise was gone (reduced to 1mV at the output of amplifier, or about 1µV at the input to the amplifier). With the long twisted pair (about 2′) put back in, the noise increased, but stayed fairly low (about 5–10mV at the output of the amplifier). The remaining noise still seems to be digital in nature and looks like it is coming from the Arduino—perhaps now by capacitive coupling rather than directly from the power supply (or it could be noise from the power supply through the amplifier, since I’m still using the Arduino power supply for the amplifier).
[Correction: I wasn’t using the Arduino power supply, but powering the Arduino over USB from the Macbook Pro, so it is no wonder that the power was so noisy. If I try powering the sensor off the same power supply as the Arduino (whether the linear regulator on the Arduino board or a battery) I get 5–10mV noise at the output: more than if just the sensor is powered by the battery.]
If the amplifier gain is 1072, and the battery output is 4.026V, and the Arduino voltage supply is 5.52V, then the 1024-count full scale reading corresponds to about 5mv, and each count is about 5µV at the sensor, or 0.25 mm Hg pressure change, which is plenty of resolution.
When I tried getting as high and low a pressure as I could just by putting my mouth to the sensor, I read a max of +83 mm Hg (1.6psi) and -46.5mm Hg (-0.9psi).
I now think that the pressure-sensor interfacing with an instrumentation amp is a reasonable lab exercise, as long as they use a clean power supply for the sensor.