Understanding noise in transimpedance amplifier

Question

So I've been studying on TIAs for the last days and understood everything except the term of "noise".

Let's take this TIA as an example: https://www.ti.com/lit/ds/symlink/opa380.pdf?ts=1645280932536&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FOPA380

On page 12-13 noise specifications are mentioned and examples are made. I understand that the capacitance of the photodiode as well as all capacitances of the TIA itself and the feedback capacitance take a big role in calculating the frequency, at which gain is decreasing (\$f_{-3dB}\$) as well as the expected noise.

Basically you should cut of your bandwidth to the required bandwidth thru the feedback capacitor in order to decrease noise on large bandwidths. Is this correct?

Also revert biasing a photodiode decreases it's capacitance and also helps in reducing noise and gaining bandwidth (figure 10). Correct?

On page 12 "TRANSIMPEDANCE BANDWIDTH AND NOISE" they are writing about how noise increases with the square root of the feedback resistor in comparison to the linear increase of the signal. Gain should appear in the "transimpedance stage"

What is the transimpedance stage? Is it the frequency below \$f_{-3dB}\$?

Therefore one should also amplify one's signal as much as possible -> Aren't there any downside?

Last and perhaps most important question: What is the difference between "Integrated Output Noise" (figure 9) and " Output Noise" (figure 8)?

Figure 8 shows the output noise (resulting noise AFTER or BEFORE gain?) in nV/√Hz. Let's say I have a bandwidth of 10kHz, therefore my typical output noise in this exact bandwidth would be somewhere between 80 nV. Is this correct?

But what does figure 9 then want to express in comparison to figure 8? Is it the exact output noise for the equivalent circuit and gain that is established with \$R_f=10M\Omega\$ on figure 6 a-c?

Answer 1

What is the transimpedance stage? Is it the frequency below \$f_{-3dB}\$?

A "stage" is a single amplifying circuit within a larger circuit that cascades several amplifiers together.

The transimpedance stage is the circuit you are building with the OPA380, in contrast to a second voltage gain stage you might consider adding to the output of this circuit.

It is not a range of frequencies.

Therefore one should also amplify one's signal as much as possible -> Aren't there any downside?

Increasing the gain will reduce the achievable bandwidth.

Also the amplifier has a maximum 4.4 V output, so you shouldn't increase the gain above the point where you get 4.4 V output for your maximum optical input.

Last and perhaps most important question: What is the difference between "Integrated Output Noise" (figure 9) and " Output Noise" (figure 8)?

Figure 8 is showing the spectrum of the noise at each frequency. If you squared the value here, you'd get a value proportional to the power spectral density of the noise.

Figure 9 is showing the total noise power from DC to a given frequency. This is the integral of the curve in Figure 8.