Precision Low Power: Understanding CMRR and RLD in Biopotential Signal Chains

Precision Low Power: Understanding CMRR and RLD in Biopotential Signal Chains

In the last blog post we showed a DC coupled biopotential configuration using the AD4130-8 and referenced a third electrode used to bias the body to midsupply.  We mentioned this was not a true Right Leg Drive (RLD) and that this may be acceptable for battery powered solutions.  Today’s post will provide some clarity to why that is and what are some of the benefits of using three vs. two electrodes for a single channel biopotential measurement.

Input Biasing

First let’s talk about the third electrode’s use in biasing.  Since biopotential signals and interferers are fully differential, ideally the circuit measuring the electrodes would need to be biased somewhere near mid-supply.  The circuit’s common mode input range should also be taken into account.  In a two electrode solution, the body is floating to some unknown potential, so resistors must be added to provide the DC bias to the inputs as well as an input bias current return path.  This results in a reduced input impedance of the circuit measuring the electrodes.  With the addition of a third electrode, the inputs can be set to this same DC bias via a low impedance path through the body without the need for extra components, thus maintaining a high input impedance.  See this comparison in Figure 1 below.

Figure 1 – Biasing inputs with 2 and 3 electrodes

Common Mode Rejection Ratio (CMRR)

In a bipotential signal chain, the typical common mode interferer of concern is the 50/60Hz that comes from the AC Mains.  The body acts as an antenna and this 50/60Hz signal can show up at the electrodes.   While the CMRR specification found on a datasheet for a component like an Instrumentation Amplifier is important, it is only one part of the System CMRR.  This datasheet specification is a figure of merit that compares the gain of common-mode signals (ACM) to the gain of differential-mode signals (ADM):

Ideally, the common-mode gain is very small and the CMRR should be high (100dB for example).  Make sure to look at the Typical Performance Curve (TPC) as well to see the CMRR vs Frequency (See AD8237 example in Figure 2).  Sometimes the datasheet specification table only provides the CMRR at DC.

Figure 2 – CMRR vs Frequency of AD8237 for different gains

System CMRR

Other factors that affect the overall system CMRR include common mode (CM) to differential mode (DM) conversion and isolated vs non-isolated solutions.   We touched on CM to DM conversion in the second blog when discussing AC coupling at the very frontend.  When making a differential measurement, anything that touches the inputs needs to be balanced.  This includes bias resistors, RFI filters, cables, connectors, the pc board, and even the electrodes themselves.  Figure 3 shows an example and how CMRR vs Frequency can be affected. 

 Figure 3 – Example of mismatch effects on System CMRR vs Frequency

To model the 50/60Hz injection into the body, a simple capacitive divider can be used and represented by a capacitance from the body to the AC mains (Ct) and a capacitance from the body to earth ground (Cb).  A non-isolated circuit would have a direct short between system ground and earth ground. Isolated circuits such as battery powered solutions have some stray capacitance (Cstray) between an isolated ground and earth ground as shown in the Figure 4 models.  You can see the benefit of the third electrode (Ze3), as it provides a direct path from the body to ground, shunting around the capacitive divider and reducing the 50/60Hz voltage picked up at the measurement electrodes (Ze1, Ze2). Otherwise, in the 2-electrode solution, the 50/60Hz current path is through the Zc1 and Zc2 common mode input impedances. This leads to a larger common mode voltage at the input electrodes which can then be converted to differential mode due to mismatch.

Figure 4 – Models for Isolated 2 and 3 Electrode Biopotential Measurements

Right Leg Drive

The name Right Leg Drive comes from the location of where the third electrode was typically placed (furthest from the heart) for an ECG measurement.  The third electrode is not limited to this location however.  This circuit is built by sensing the common mode at the inputs (Vcm), buffering, inverting (usually with an integrator circuit) and driving back into the body.   This creates a feedback loop, where higher loopgain at 50/60Hz improves common mode rejection.  See example circuit and integrator transfer function recommended for the AD8233 in Figure 5.  The integrator crossover frequency is set at ~1kHz.  Pushing to higher frequencies can increase risk of instability, so there is a tradeoff here.

Figure 5 – Example AD8233 RLD Integrator Circuit and Transfer Function

Note that this circuit will be more effective for a non-isolated solution or larger values of Cstray (such as AC mains powered solutions with isolation).  An isolated solution will also improve system CMRR as Cstray becomes smaller, allowing the battery powered measurement circuit to move up and down with the common mode.

Next time we’ll talk about electrodes including their models, and the challenges of making good contact with dry electrodes. 

[1] B. Winter. & J. Webster, “Driven-Right-Leg Circuit Design,” IEEE Transactions on Biomedical Engineering, vol. BME-30, no. 1, pp. 62-66, Jan. 1983.

Anonymous