Electronic – Troublshooting noisy smps

The PWM range of the device you list is from 0% to 50% (theoretically) but I'm assuming, for the sake of this answer that it's from 3% to 45%. (this is a more guaranteed range)

For a decent load range of (say) 5 watt (basic internal switching losses) to 50 watts, the PWM has to "push" its duty cycle around a fair bit to maintain the output voltage at a constant level (even on a fixed input voltage of say 150 volts).

Given that the input voltage range of 3:1 reduces the duty cycle range of 45:3 to 45:9, it means that 5:1 of the PWM range is all that remains to regulate the output voltage against load power changes of maybe 10:1 (50 watts to 5 watts).

Without knowing the load range you have to cover it's impossible to say whether 60 Vac will be OK or not.

Another consideration is that the bridge diodes, capacitor and switching transistor have to be rated so much more powerfully at lower AC supplies because this type of switcher takes a constant power from the AC to provide the load power - you reduce voltage by 3:1 and current taken from the AC rises by 3:1 - if you want a 4:1 range then this has to be accommodated. Bear also in mind that the flyback transformer will also have to be made bigger to avoid saturating at the peak currents expected when operating from lower AC voltages.

I'd say it can be done but tread carefully.

The error amplifier will not always have a negative output and not always have a positive output. This is the point of the error amplifier.

Because of the negative-feedback characteristic of the op-amp it will try to keep 2.5V at its negative pin, as you say yourself. So if the incoming signal over the potentiometer P1 drops below that it will swing up the output to start trying to pull it up. When the signal incoming goes above the target 2.5V it will swing the output low.

There is your feedback, that signal gets propagated through the current sense amplifier, which is yet another kind of error amplifier, but for the primary current, as a protection, into the control circuitry.

The P1 potentiometer allows you to adjust "how violently" it reacts to errors in the output. The lower its value the more heavily the error amplifier will react to small errors on the red arrow.

The point of all the capacitance in the feedback, is that it is not at all a DC signal. The switching regulator switches the output on and off with a frequency, likely between 20kHz and 200kHz (I have not looked up this exact chip, but those are common fly-back outer frequencies). So, the capacitances in the feedback loop allow the error amplifier to filter out frequency responses that are too low or too high, by adjusting the feedback impedance depending on the frequency it tries to amplify.