A conventional 1-MΩ transimpedance amplifier has at least 130 of output-noise density at room temperature (Figure 1). You can consider the 130 nV as the theoretical noise floor limit of the amplifier because that is the noise density of the 1-MΩ resistor itself. Any noise in the op amp can only make things worse. Cooling the resistor to 77.2K, the temperature of liquid nitrogen, quiets it to 65 provided that it survives, but is that the only option? Can you beat the 130-nV theoretical noise floor without cooling?
Figure 2 shows one way. IC1, a Linear Technology LTC6240, provides an overall transimpedance gain of 1 MΩ, but it has an output-noise density of only 43 about one-third of a conventional 1-MΩ transimpedance amplifier at room temperature. It achieves this figure by taking an initial transimpedance gain of 10 MΩ and then attenuating by a factor of 10. The transistor section provides voltage gain and works on a 54V supply voltage to guarantee adequate output swing. By achieving an output swing of 50V before attenuation, the circuit maintains an output swing to 5V after attenuation. The 10-MΩ resistor sets the gain of the transimpedance-amplifier stage and has a noise density of 400 After attenuation, the amplifier’s effective gain drops to 1 MΩ, and the noise floor drops to 40 which dominates the observed 43 To achieve this noise performance by cooling requires a temperature of 33K, much colder than liquid nitrogen. Note also that the additional benefit of this method is that it divides the offset voltage of the op amp by 10. The worst-case output offset for this circuit is 105 µV over temperature. Bandwidth is 28 kHz.
Click here for Illustrations:
Figure 1
