Measuring the RTO noise and then calculating the RTI noise of the amplifier circuit is easy by knowing the NG. Figure 2-4 shows the measured amplifier versus the simulated RTO noise, RTI noise, and the measured noise floor of the HP4395A spectrum analyzer.

Figure 2-4 The Measured Amplifier Versus the Simulated RTO and RTI Noise

In Figure 2-4, the measured RTO noise curve of the amplifier matches the simulated RTO noise very well. The measured RTO noise of the amplifier is also 10dB above the measured noise floor of the spectrum analyzer (as required), which achieves a measurement accuracy of greater than 95%.

Similarly, the calculated RTI of the amplifier, from dividing the measured RTO noise by the NG, matches the simulated RTI noise well. The measured amplifier RTI noise is acting as the new (improved) noise floor for the LDO noise measurement.

Before making noise measurements of the LDO (DUT), TI recommends verifying another measurement to confirm the amplifier performance. A flat white noise generator like the NC6105A was used to generate a 10nV/Hz (-147dBm) flat noise across a BW of 10Hz–10MHz, then that noise was measured using this design. This very low-noise level is slightly lower than the noise floor of the HP4395A. Figure 2-5 shows the measured flat noise of 10nV/Hz.

Figure 2-5 Measured Flat Noise of 10nV/Hz Across a BW of 10Hz–10MHz

A lower noise level can be measured using this amplifier (as low as 1nV/Hz and potentially 0.9nV/Hz), however, the input noise to the amplifier is lower than the noise of the actual amplifier (op-amp 1/f noise), below 1kHz, as shown in Figure 2-6.

Figure 2-6 Measured Flat Noise of 1nV/Hz Across a BW of 1kHz–10MHz