IIRC, the point of the expression is just exactly that ... If you increase the output current, the ripple is increased. The point of that is that one must know what the tolerable ripple at the input is. That means that one has to know what the unregulated input looks like at the desired load. It's a somewhat confusing thing to consider, but, given all the correct information, it produces a reasonable estimate of the minimal input capacitance.

It also tells you that if you can tolerate larger input ripple, a smaller input cap will do the job adequately. What it doesn't address is the heat (power) dissipated in the regulator if you have so much "overhead".

When I first had to wrestle with these issues, 5-volt output was generally the desired result, and any raw input voltage, at rated current, in excess of the required 7.5 volts was considered excessive. Of course, we needed between 10 and 20 amperes for a sizeable system back then.

Times have changed, but not the requirement for input filtering.



Well, I recall numerous discussions of just exactly that matter. I don't recall the details. After all, we were trying to come up with a standard way of sizing input filter cap's. How NSC stated their recommendation wasn't as important at the time as was their syntax. Perhaps it should have gotten more attention.

BTW, the bulk of the capacitance generally was an electrolytic capacitor, though NSC, at that time, liked small (2.2 uF) tantalum caps.

RE