ADS1292R RA/LA Respiration

We are currently testing with the ADS1292R for both ECG and respiration, using the RA/LA without RL drive (Two-Terminal Measurement as per Figure 1 in the Ti document 'Respiration Rate Measurement Based on Impedance

Pneumography').

We've encountered an issue in which the respiration measurement can effectively invert, so we appear to measure a decrease in impedance as the chest cavity expands during inspiration. This change from inverted to non inverted measurement can occur during a single test or between tests. We understand (again from the above Ti doc) that this can occur when using a sine wave carrier and the phase shift is too large (>135 degrees), however we are using a square wave carrier (sine wave not possible on the ADS1292R). What conditions could cause signal inversion using a square wave carrier?

Similary, we suspect we may need to optimise the phase shift and frequency for our particular setup.  We are currently using internal modulation and demodulation at a phase angle of 112.5 deg and modulation frequency of 32khz as recommended in the ADS1292R datasheet. Is there a procedure to follow to find the optimal settings for our sensors?

See attached our current RESP/ECG front end.

Registry Settings:

reg_config1 = 0x02;  // continuous mode, sample at 500Hz
reg_config2 = 0xA0;  // internal ref, VREF 2.42V
reg_loff    = 0x11;  // AC lead off (actually unused)
reg_ch1set  = 0x00;  // normal, gain = 6
reg_ch2set  = 0x00;  // normal, gain = 6
reg_rld     = 0x00;  // no RLD
reg_loff_sens    = 0x00;  // all disabled
reg_resp1   = 0xEA;  // mod/demod on, 112.5 deg phase, all internal
reg_resp2   = 0x03;  // 32KHz resp control freq, internal RLDREF