As mentioned in last section, PMT (photon multiplier tube) has a large active area with high capacity for photons thus can detecting strong signal with less saturation problem. PMT can multiply received photons thus has high photon sensitivity suitable for detecting weak signal at very low noise level. These features endow PMT with wide dynamic range and good SNR.

But in fluorescence microscopy, the total photon number is very small, less than 1000 photons per pixel time, and even smaller in confocal microscope, about 10-30 photons/pixel/µs due to the massive rejecting of signal by pinhole. The dynamic range can not be calculated from the ratio of full photon capacity / noise anymore, and solely affected by signal intensity.
It is determined by formula:
30 photon influx corresponds to only 5 bits grey level.

Similarly, the SNR is affected more by signal intensity (total photon influx) than by background noise level as mentioned in part 1 Optical section,
 Formula 4 describes different factors in SNR.

Where
N:  photoelectron number per pixel time.
S:  secondary noise from random variation of multiplication.
d:  dark current of PMT.
q:  laser noise.

When N is 1000 photoelectron per pixel time, if set S: 1.2, d: 100. q: 0.05, SNR is 25.
At 400 Hz scan speed, 512 format, pixel time is about 5 µs and N is about 150 assuming photon influx is 30 photons/pixel/µs, from formula above,  with S, D, Q unchanged, SNR is 7.89 only.

In formula 4, photoelectron number on the numerator has to be squared, it has more weight in the formula and results in a pronounced effect on SNR. This makes confocal microscope very vulnerable to weak signal. The reduced photon number not only weakens signal intensity, but also deteriorates image quality. Raising gain (voltage) on PMT can amplify weak signal but also raise noise, raising offset on PMT (threshold) can cut off background noise but signal is equally affected,  the SNR and image quality won't improve. In worst case, the structure details are buried in the noise, you even can't get usable data at all. To improve image quality, approaches which increase photon number has to be used, such as average, accumulation, slow scan speed, lower scan format, larger pinhole size, etc..