I would be very interested to see what effect the capacitors used actually have on ripple voltage. What this means is obtaining several sets of capacitors from various series and testing each of them in the same PSU. To make sure aging doesn't skew the results, they should all be brand new. For completeness, a few different test units may be desirable.
My thinking is that the capacitors used in many if not most units are chosen not to get the lowest ripple practical but instead as a "good enough" cost/performance compromise. After all, in the business world, marginal savings are taken way too seriously.
Please show the ripple waveforms in detail along with listing the peak-to-peak voltages.
