Supplier eBooks

distortion, see Zero-crossover Amplifiers: Features and Benefits). However, the offset of the amplifier is corrected through internal periodic calibration, so the magnitude of the offset transition and the crossover distortion is greatly diminished. Figure 2 shows a comparison of the offset between a standard CMOS rail-to-rail and a zero-drift amplifier. How zero-drift works Chopping zero-drift amplifiers' internal structure can have as many stages as continuous-time amplifiers the main difference is that the input and output of the first stage has a set of switches that inverts the input signal every calibration cycle. Figure 3 shows the first half cycle. In the first half cycle, both sets of switches are configured to flip the input signal twice, but the offset flips once. This keeps the input signal in phase but the offset error polarity is reversed. Figure 4 shows the second half cycle. Here, both sets of switches are configured to pass the signal and offset error through unaltered. Effectively, the input signal is never out of phase, remaining unchanged from end to end. Since the offset error from the first clock phase and second clock phase are opposite in polarity, the error is averaged to zero. A synchronous notch filter is used at the same frequency of switching to attenuate any residual error. This principle continues to be in effect throughout the amplifier's operation across its input, output, and environment. In essence, TI's zero-drift technology delivers ultra-high performance and outstanding precision owing to this self-correcting mechanism. Auto-zeroing requires a different topology but results in similar functionality. The auto-zeroing technique has less distortion at the output. Chopping results in lower broadband noise. Noise in zero-drift amplifiers In general, zero-drift amplifiers offer the lowest 1/f noise (0.1Hz–10Hz). 1/f noise (also referred to as flicker or pink noise) is the dominant noise source at low frequencies and can be detrimental in precision DC applications. Zero-drift technology effectively cancels slow varying offset errors (such as temperature drift and low frequency noise) using the periodic self-correcting mechanism. Figure 5 shows the 1/f and broadband voltage noise spectral density for a zero-drift (red) and continuous-time (black) amplifier. Notice the zero-drift curve has no 1/f voltage noise. 9 Figure 2: CMOS and Zero-drift Input Offset Voltage Comparison Figure 3: First Half-cycle of Internal Structure Figure 4: Second Half-cycle of Internal Structure Figure 5: Voltage Noise Comparison Table 1: Input Offset Voltage and Drift Comparison Table 1: shows a comparison of VOS and dVOS/dT of a continuous-time and zero-drift amplifier. Notice that the VOS and dVOS/dT are three orders of magnitude smaller on the zero-drift amplifier.

• Cover