9.2.1.2.1 Antialiasing Filter

Converting analog-to-digital signals requires sampling an input signal at a rate greater than or equal to the Nyquist rate. Any higher frequency content in the input signal beyond half the sampling frequency is digitized and folded back into the low-frequency spectrum. This process is called aliasing. Therefore, an external, antialiasing filter must be used to remove the harmonic content from the input signal before being sampled by the ADC. An antialiasing filter is designed as a low-pass RC filter, for which the 3-dB bandwidth is optimized for noise, response time, and throughput. For dc signals with fast transients (including multiplexed input signals), a high-bandwidth filter is designed to allow the signal to be accurately set at the ADC inputs during the small acquisition time window. Figure 40 provides the equation for determining the bandwidth of antialiasing filter.

Figure 40. Antialiasing Filter

For ac signals, the filter bandwidth must be kept low to band limit the noise fed into the ADC input, thereby increasing the signal-to-noise ratio (SNR) of the system. Besides filtering the noise from the front-end drive circuitry, the RC filter also helps attenuate the sampling charge injection from the switched-capacitor input stage of the ADC. A filter capacitor, C_FLT, is connected across the ADC inputs. This capacitor helps reduce the sampling charge injection and provides a charge bucket to quickly charge the internal sample-and-hold capacitors during the acquisition process. As a rule of thumb, the value of this capacitor must be at least 20 times the specified value of the ADC sampling capacitance. For this device, the input sampling capacitance is equal to 15 pF. Thus, the value of C_FLT must be greater than 300 pF. The capacitor must be a COG- or NPO-type because these capacitor types have a high-Q, low-temperature coefficient, and stable electrical characteristics under varying voltages, frequency, and time.

Note that driving capacitive loads can degrade the phase margin of the input amplifiers, thus making the amplifier marginally unstable. To avoid amplifier stability issues, series isolation resistors (R_FLT) are used at the output of the amplifiers. A higher value of R_FLT is helpful from the amplifier stability perspective, but adds distortion as a result of interactions with the nonlinear input impedance of the ADC. Distortion increases with source impedance, input signal frequency, and input signal amplitude. Therefore, the selection of R_FLT requires balancing the stability and distortion of the design.

The input amplifier bandwidth must be much higher than the cutoff frequency of the antialiasing filter. TI strongly recommends performing a SPICE simulation to confirm that the amplifier has more than 40° phase margin with the selected filter. Simulation is critical because even with high-bandwidth amplifiers, some amplifiers may require more bandwidth than others to drive similar filters.

9.2.1.2.2 Input Amplifier Selection

Selection criteria for the input amplifiers is highly dependent on the input signal type and the performance goals of the data acquisition system. Some key amplifier specifications to consider while selecting an appropriate amplifier to drive the inputs of the ADC are:

• Small-signal bandwidth: Select the small-signal bandwidth of the input amplifiers to be high enough to settle the input signal in the acquisition time of the ADC. Higher bandwidth reduces the closed-loop output impedance of the amplifier, thus allowing the amplifier to more easily drive the low cutoff frequency RC filter at the ADC inputs. Higher bandwidth also minimizes the harmonic distortion at higher input frequencies. In order to maintain the overall stability of the input driver circuit, select the amplifier bandwidth as described in Equation 2:
Equation 2.
where
• GBW = Unity-gain bandwidth
• Noise: Noise contribution of the front-end amplifiers must be low enough to prevent any degradation in SNR performance of the system. As a rule of thumb, to ensure that the noise performance of the data acquisition system is not limited by the front-end circuit, keep the total noise contribution from the front-end circuit below 20% of the input-referred noise of the ADC. Noise from the input driver circuit is band limited by designing a low cutoff frequency RC filter, as explained in Equation 3.
Equation 3.
where
• V_{1/f_AMP_PP} is the peak-to-peak flicker noise in µVrms,
• e_{n_RMS} is the amplifier broadband noise,
• f_–3dB is the –3-dB bandwidth of the RC filter,
• k is the Boltzmann's constant, and
• T is absolute temperature in kelvin.
• For symmetrical feedback, β = R1 / (R1 + R2) = R3 / (R3 + R4).
• For details on noise analysis, refer to the technical brief Analysis of fully differential amplifiers (SLYT157)
• Settling time: For dc signals with fast transients that are common in a multiplexed application, the input signal must settle to the desired accuracy at the inputs of the ADC during the acquisition time window. This condition is critical to maintain the overall linearity performance of the ADC. Typically, the amplifier data sheets specify the output settling performance only up to 0.1% to 0.001%, which may not be sufficient for the desired accuracy. Therefore, always verify the settling behavior of the input driver with TINA™-SPICE simulations before selecting the amplifier.

The OPA316 is selected for this application for its rail-to-rail input and output swing, low-noise (11 nV/√Hz), and low-power (400 µA) performance to support a single-supply data acquisition circuit.

9.2.1.2.3 Reference Circuit

The analog supply voltage of the device is also used as a voltage reference for conversion. TI recommends decoupling the AVDD pin with a 1-µF, low-ESR ceramic capacitor. The minimum capacitor value required for AVDD is 200 nF.

Table 4. Design Parameters