Designs that require high bandwidth from a large area detector with relatively high transimpedance gain benefit from the low input voltage noise of the OPA818. This input voltage noise is peaked up over frequency by the diode source capacitance, and in many cases, becomes the limiting factor to input sensitivity. Figure 8-3 shows the transimpedance circuit with the parameters defined in Section 8.2.1.1. To use the Excel calculator available in What You Need To Know About Transimpedance Amplifiers – Part 1 to help with the component selection, determine the total input capacitance, C_TOT. In the calculator, C_TOT is referred to as C_IN. C_TOT is the sum of C_D, C_DIFF, and C_CM, which is 7.4 pF. Using this value of C_TOT, the targeted closed-loop bandwidth (f_–3dB) of 24 MHz, and the transimpedance gain of 100 kΩ results in the need for an amplifier with approximately 2.68 GHz GBWP and a feedback capacitance (C_F) of 0.092 pF. Table 8-2 shows the calculator results. These results are for a Butterworth response with a Q = 0.707 and a phase margin of approximately 65°, which corresponds to 4.3% overshoot.

With a 2.7-GHz GBWP, the OPA818 is an excellent choice for the design requirements. A challenge with the calculated component results is practically realizing a 0.092-pF capacitor. Such a small capacitor is realized by using a capacitive tee network formed by C₁, C₂, and C_T (see Figure 8-3). The equivalent capacitance, C_EQ, of the tee network is given by Equation 2:

The tee network forms a capacitive attenuator from input to output with C₁ and C_T, and from output to input with C₂ and C_T. With the value of C_T being higher than C₁ or C₂, only a fraction of the output signal is seen by C₁. This network results in a much smaller shunting current provided to the input through C₁, and this reduced shunting current effect is equivalent to how a much smaller capacitor behaves. At a fixed frequency, a smaller capacitor has a higher impedance, and thus reduced current. Keep the same level of attenuation from input to output, and vice versa. To find the appropriate capacitor values for the tee network, chose an arbitrarily low but practically realizable and equal values for capacitors C₁ and C₂, set C_EQ = C_TOT, and use Equation 3 to get the value of the tunable capacitor, C_T. The values of capacitors C₁, C₂, and C_T in Figure 8-3 are determined using this process.

Figure 8-4 shows the TINA-TI™ simulation software closed-loop bandwidth response of the circuit in Figure 8-3. The circuit is designed for f_–3dB = 24 MHz and the simulated closed-loop 3-dB frequency is 24.6 MHz with approximately 0.1‑dB peaking. The OPA818 TINA-TI software model models the input common-mode and differential capacitors that are not added externally when simulating in the TINA-TI software. Figure 8-5 shows the noise simulation of the TIA circuit. The output-referred voltage noise shows on the Y-axis to the left. The input-referred current noise, which is essentially output-referred voltage noise divided by the transimpedance gain of 100k, shows on the secondary Y-axis to the right. The simulation results are fairly accurate because the OPA818 TINA-TI software model closely models the voltage and current noise performance of the amplifier. The flat-band output voltage noise is 41 nV/√Hz that is equivalent to 0.41 pA/√Hz of input-referred current noise. The noise in relatively low frequency region where the noise gain of the amplifier is 1 V/V is dominated by the thermal noise of the 100‑kΩ resistor (40.7 nV/√Hz at 27°C). At mid-frequencies beyond the zero formed by R_F and C_TOT, the noise gain of the amplifier amplifies the voltage noise of the amplifier. The amplifier noise starts to become the dominant noise contributor from this frequency onward, before the output noise starts to roll off at frequencies beyond the 3-dB closed-loop bandwidth. When looking at the integrated root-mean-square (RMS) noise, mid-frequency noise is potentially a significant contributor. Therefore, use a 2.2-nV/√Hz low-noise amplifier, such as the OPA818, to minimize total RMS noise in the system.