he tell that there is wiggle in the waveform due to the Signal to Noise ratio. What is the signal here and what is the noise?
The scope has a wave generator which injects signal into the LDO input via the Pico accessory. Then one scope channel probes the LDO input (10:1 probe) and the other channel probes the output (1:1 probe).
It automatically generates signals over a range of frequencies, then measures the PSRR which is Vin/Vout in dB. So if it injects AC signal into Vin and creates a 100mV ripple on the input, and there is a 10mV ripple on the output, then the PSRR is 100/10 = 10 = 20dB. The ratio is between AC amplitudes, ignoring DC.
What it displays at 5:20 is not the actual signal, it is the PSRR versus frequency curve. I don't know what algorithm the scope uses to compute the AC amplitudes before it does the calculation to get the PSRR, but if it is simply RMS (which would make sense) then any extra noise generated by the scope itself or the LDO will also sneak into the result and manifest as noise on the PSRR graph... ie, as wiggles. That's what you can see on the graph.
Now, since PSRR measurement is Vin/Vout, trouble can come from noise on Vin, or noise on Vout. Usually the AC voltage on Vin is large enough to be properly observed, but if the LDO has good PSRR, AC output voltage will be tiny so it can disappear in noise, especially if the LDO has good PSRR but it is noisy.
He also mentions that it can be corrected by increasing the amplitude above 200mV. How is this possible?
As long as the LDO operates in linear fashion, increasing AC amplitude at the input also increases AC amplitude at the LDO output, so the signal we want to measure is larger, better signal to noise ratio results in a cleaner graph.
However, increasing input voltage ripple too much may cause the LDO to behave in a nonlinear fashion (he shows this around 7-8 minutes) which is not what you want. PSRR is supposed to be a small-signal, linear measurement. When input AC voltage is high enough to cause non-linear distortion on the output, this is no longer small-signal conditions, and while something will be measured, it will no longer be "PSRR". You could call it "RMS AC voltage on the output with such and such AC amplitude at the input"... but not "PSRR" since the "R" means "Ratio" and thus linear conditions.
Around 10:00sec, he mentioned that there is something called as inflection around 2.5MHz. Can someone help me understand what is inflection that he is talking about and why does it occur?
It's the dip in the curve inducated by the yellow triangle.
It is either due to the output capacitor self-resonance, or it is the point where the output impedance is no longer dependent on the LDO, but just on the output cap.
