SBAA666 February 2025 AMC0106M05 , AMC0106M25
A series of tests was performed to quantify measurement noise and the corresponding effective number of bits (ENOB) at different oversampling ratios and load currents. The ENOB was calculated from 1200 consecutive samples measured at the center of the 10kHz PWM over 1200 PWM periods. The PWM vector was kept constant during the test. The constant PWM vector produces constant phase voltage and DC current. The ENOB was calculated as follows:
where noiseRMS equals the standard deviation of the 1200 measurements for a given OSR. Figure 4-3 shows the histograms for OSR 32, 64, 128, and 256 measured at a 3A DC current. The resulting ENOB as a function of OSR is shown in Figure 4-4.
At OSR=32 the design already provides a resolution of 11.1 bit. There is no measurable difference between zero and 3A or 6A of current. As the OSR increases, the impact of system noise at larger phase currents becomes visible. For example, the ENOB at OSR 128 increases to 13.7 bit for smaller currents while slightly degrading to 13.5 bit at 6A. Table 4-1 summarizes the ENOB, noise, and latency measured at 3A with different OSRs.
| OSR | 32 | 64 | 96 | 128 | 192 | 256 |
|---|---|---|---|---|---|---|
| Noise(mARMS) | 18.4 | 5.93 | 4.14 | 3.46 | 2.94 | 2.51 |
| ENOB (bit) | 11.1 | 12.8 | 13.3 | 13.5 | 13.8 | 14 |
| Latency (µs) |
2.4 | 4.8 | 7.2 | 9.6 | 14.4 | 19.2 |
A sinusoidal phase current measurement is shown in the next chapter using a high 100kHz PWM frequency to demonstrate high measurement accuracy even during PWM switching.