Boulder, CO (October 2024) - This experiment compared the tradeoffs between using a single-ended and differential pair measurement for an analog signal. Single-ended measurements are interpreted relative to the reference plane. Using this method, only a single wire is needed for the signal, but the measurement can be a victim to reference plane noise since any noise on the reference plane will transfer noise to the measurement. To mitigate this, differential signaling may be used. Instead of comparing the signal to the reference plane, it is compared to the reference signal at the sensor. By doing this, the voltage drop over the length of the wire or reference plane noise impacts both signals; using the difference between these two signals, instead of using the difference between the signal and a potentially noisy reference plane, gives a more accurate measurement. However, using differential measurements requires an additional wire to be routed and consumes an additional Analog-to-Digital Converter (ADC).
Figure 1: Hardware setup connecting a TMP-36 temperature sensor to the ADS1115 16-bit, 4-channel ADC. The ADC is connected to the Arduino UNO via I2C.
A temperature sensor (Tmp-36) was used to explore the benefits of the two analog data transfer techniques. The analog signal representing temperature was read by a 16-bit, 4 Channel ADC (ADS1115) which passed the digital signal to an Arduino UNO for analysis. Figure 1 shows the prototype circuit setup. Figure 2 compares the difference between the single-ended and differential signals. The two signals are similar with a stable reference plane over several inches of wire, both indicating between 23 and 24 degrees Celsius, as determined by Equation 1, where T is the temperature in degrees Celsius and Z is the measurement in volts. C and V represent the units on the constant terms.
Next, a 0.5Hz, 20V peak-to-peak signal was applied across the reference plane. Figure 3 shows the significant impact this reference plane noise has on the single-ended measurement when compared to the stable differential measurement. The single-ended measurement varies between 21 and 26 degrees Celsius while the differential measurement remains steady at 23 degrees Celsius. This indicates that using differential measurements reduces noise substantially when compared to single-ended measurements. This requires additional hardware and signal routing which must be considered when determining the technique to use for a specific application.
Figure 2: Shows the single-ended measurement in green and the differential measurement in yellow bounded by a minimum and maximum expected value in blue and orange, respectively. With a stable reference plane, the noise is minimal, and the two measurements are nearly identical.
Figure 3: Shows the single-ended measurement in green and the differential measurement in yellow bounded by a minimum and maximum expected value in blue and orange, respectively. With a 0.5Hz, 20Vpp signal applied across the reference plane, the significant impact of reference plane noise to the single-ended measurement can be seen.Â