Hi,
If you put a resistance in series with a capacitor and drive it with a square wave and measure the voltage across the cap, a pure capacitance will show an exponential curve that is rounded except where the square wave transitions from low to high or from high to low. At those points the curve will change from concave up to concave down or vise versa, but during that tiny portion of the wave the amplitude will not change, so here are only two parts to the wave.
Now if you add a second small resistor in series with the cap and measure across both the small resistor and cap rather than just across the cap, that forms a voltage divider, so that when the input wave transitions we'll see an immediate voltage increase that does not take any time to change, followed by the typical capacitor curve as before.
Since this small resistor represents the ESR, that is the new wave shape we see with a given ESR equal to the small resistor value. The ESR can then be determined by the amplitude of the part of the wave that goes straight up when the square wave drive transitions.
At higher frequencies the inductance will start to show an effect too though. Now we have a resistor and an inductor in series with the cap, and when we look on the scope w see the effects of both the inductor and the ESR during the transition. The inductor adds the new effect where the wave shoots up quickly and then may take a little time to come back down. That will interfere with the reading we get from the ESR somewhat.
Depending on the ratio of the component values and the frequency, we may or may not get a good reading on the ESR this way. This is when we have to switch to a method that tests for the energy loss in the circuit and deduce the ESR from that, because a pure inductance and pure capacitance does not absorb energy while a pure resistance does.
That's probably as deep as you need to get but then there is radiation. If the circuit radiates significant energy then that starts to look like increased energy loss, and then the measurement also starts to depend on frequency.
So there's no perfect rule, except for one, and that is where the cap is actually placed into the end application circuit and the entire circuit tested as whole. This means that the cap will be operating at the required frequencies of the application and that's the best test. We could then look for signs of higher than normal ESR, or in some cases even lower than normal ESR which could mess up some circuits like boost regulators.
After all is said and done though, i would think for capacitors 1uf and above you should be able to get usable results using the square wave and scope method because the ratio of capacitance to inductance should be high enough. If you have doubts, we can roll some numbers and see what we get. The circuit is simple enough, a resistor in series with a small inductor and small resistor and in series with a pure capacitance, and take measurements across the inductor and small resistor and capacitor.