He is probably quoting me there.
Ok, so you have a super cap, and you need to charge it before use. So, you have to have a current limited supply to charge it otherwise bad things will happen. So, you need a current limited supply, ok, so the cap charges up and you connect the load. So now the load is connected to the current limited reg and the super cap. So the load draws some current. Where does that current come from? The super cap or the current limited power supply? I don't know either, maybe one, maybe both. If its one or both, how is it really any different from powering the load with a regulator with a cap on the end? is there some magic about the supercap? I don't know either. Ahh, you say, but the super cap can provide more current in a pulse than the current limited supply. Well, maybe so, but exactly what load is going to need that big a pulse of current, and no matter what the ESR of the super cap, it needs to get from the cap to the load, so there will still be the resistance and inductance of the connecting leads. So just how big is the pulse going to be that can't be supplied by a normal cap? Ok, lets assume there is a load that does draw that big a pulse? How often does it do it? If its more often than some value, then the current limited regulator that's feeding the supercap is not going to have enough power and the supercap will eventually discharge.
The Berrisford supply is what I described, a simple voltage regulator, followed by a simple current limiter followed by a super cap. Is it better than a cheap voltage regulator? Yes, the cap hides the noise of the reg. Is it a general solution? No, because its current limited in what it can supply by the current limiter to stop the thing blowing up on startup. Is it better than a good voltage regulator with a small cap on the output? I don't know, to my ears no, but I am biased.
So what if we get more tricky and say, ah, we will use a bank of the caps, charge one off line, then switch it to the output, let it discharge while the next one charges? Sounds good, except, the voltage from the cap will reduce as the cap discharges, so we will get a saw tooth wave out of the thing as it switches from cap to cap. Ah you say, we can reduce the problem, by switching faster do the discharge is less. Ok, so now you are just adding more HF noise to the output, one of the points you were trying to avoid at the start. Ok, so you get clever and make it switch at the point the voltage drops a certain amount, so it only switched as fast as it needs to. So still got the sawtooth, but now you have a load dependent HF noise source in series with the output. I know, we can avoid the sawtooth, by switching the caps as before, but send the output via a low noise regulator to remove the sawtooth. Hang on, isn't that just the same as using a cap that charges at 100Hz, and filtering the output from that with a low noise regulator. Hmm.
I am biased as I said, but to me that looks like a solution in search of a problem.
Little known fact, coherent thought can destructively interfere with itself leaving no thought at all, that’s why I prefer incoherent thought.