Lab 2 · PythonElectrodes and interference
Goal. Rebuild the interference model of Figure 4.1 and find the common-mode input impedance, amplifier CMRR and right-leg-drive loop gain the front-end needs. Goes with Chapters 3, 4 and 12.
Run.
cd labs/python
python lab2_electrodes_interference.py # 50 Hz, 100 dB amplifier, RLD loop gain 100
python lab2_electrodes_interference.py --f 60 --rld 40 --cmrr 110
The model is a chain of phasors. The mains couples 230 V to the body through 3 pF; the displacement current leaves through the ground electrode (1 kΩ plus 51 kΩ‖47 nF) and sets the body’s common mode (biosig.body_cm_voltage). One lead carries the standards’ 51 kΩ‖47 nF imbalance, and each input forms a divider with the common-mode input impedance \(Z_\mathrm{cm}\), which turns common mode into differential interference (biosig.cm_to_dm). The amplifier’s own CMRR adds to that, worst case in phase, and a behavioural right-leg drive divides the common mode at the electrodes by one plus its loop gain. The script works through five steps:
- the body’s common mode, floating and with a ground electrode;
- the interference for \(Z_\mathrm{cm}\) from 10 MΩ to 10 GΩ;
- the \(Z_\mathrm{cm}\) that 90 dB of effective CMRR needs, with and without the right-leg drive;
- the 10 Vrms CMRR test of the ECG standards for the unboosted and the boosted front-ends of Chapter 10;
- a grounded against a driven shield on one metre of lead.
Expected results. With the defaults the script prints, among others:
| Quantity | Value | Where in the text |
|---|---|---|
| Body common mode, floating / with ground electrode / with RLD | 3.40 V / 9.0 mV / 89 µV | Worked Example 4.1, Section 12.1 |
| Interference at \(Z_\mathrm{cm}\) = 10 MΩ / 1 GΩ | 36.6 µVrms / 0.37 µVrms | Worked Example 4.2 |
| \(Z_\mathrm{cm}\) for 90 dB with a 100 dB / 120 dB amplifier | 1.88 GΩ / 1.33 GΩ | Worked Example 4.4 |
| CMRR the test implies; \(Z_\mathrm{cm}\) bound without / with RLD | 89.0 dB; 1.15 GΩ / 11.4 MΩ | Worked Example 4.3, Chapter 12 |
| Test residual, boosted front-end (200 MΩ), RLD loop gain 100 | 0.060 mVpp, 24 dB margin | Chapters 12 and 29 |
| Test residual, unboosted front-end (6.25 MΩ), RLD loop gain 100 | 1.83 mVpp, fails | Chapter 12 |
| Interference with a grounded / driven (1 % error) shield | 11.7 µVrms / 0.30 µVrms | Chapter 4 pitfall |
It also plots the interference against the imbalance for four values of \(Z_\mathrm{cm}\), and the effective CMRR against \(Z_\mathrm{cm}\) with and without the right-leg drive (out/lab2_interference.png), as in Figure 4.2.
Questions.
- Why does an amplifier with 120 dB of CMRR barely relax the \(Z_\mathrm{cm}\) requirement compared with one of 100 dB? Find the amplifier CMRR below which no \(Z_\mathrm{cm}\) reaches 90 dB.
- Run at 60 Hz. Which numbers change, and by how much? Is a 60 Hz country harder or easier for the front-end?
- Find the smallest right-leg-drive loop gain with which the boosted front-end still passes the CMRR test with 6 dB of margin. Compare it with the stability limit of Figure 12.2b.
- Replace the wet electrodes by dry ones (
biosig.z_electrode(f, "dry")) for the ground electrode and the imbalance. What happens to the body’s common mode and to the interference? - How many picofarads of input capacitance can the front-end afford, from pads, ESD diodes and wiring, before \(Z_\mathrm{cm}\) falls below the 11.4 MΩ that the right-leg drive allows?
Deliverable. The two plots, the CMRR the design needs, and a recommendation for the split between amplifier CMRR, \(Z_\mathrm{cm}\) and loop gain, with the margin each leaves in the CMRR test.
All labs · Previous: Lab 1. Biosignal explorer · Next: Lab 3. Devices in weak inversion