*@mulveling* , thanks for providing the comprehensive response to the question posed by @jw944ts. In doing so you saved me a lot of time, and I of course agree with what you wrote.

JW’s calculation reflects a commonly held misconception that a "voltage db" and a "power db" both correspond to 10x the logarithm of the ratio between two voltages or two power levels, respectively. But they don’t, at least in terms of generally accepted usage, if not etymology as well. 10x is the proper multiplier to use when computing the difference between two power levels in db, but 20x is the multiplier that should be used in computing the difference between two voltage levels in db.

A db is a db. It is not either a voltage db or a power db. The numerical result will be the same regardless of whether the number of db is calculated from voltages or from power levels, assuming impedances are the same in the two cases. That is a consequence of the fact that if impedances are the same power is proportional to voltage squared, as mulveling indicated. And therefore squaring a 2x increase in the voltage provided into a given impedance corresponds to supplying 4x the power, not 2x. (In the interests of simplicity I’m putting aside effects that occur when the load impedance is not purely resistive).

And assuming the speakers and the rest of the system are operated within limits that allow them to perform in an essentially linear manner, a gain of 66 db corresponds to a voltage multiplication of about 1995x, that commonly being rounded off to 2000x. (20 x log(1995) = 66 db, where "log" is the base-10 logarithm). And if everything else is equal that results in approximately 4,000,000x (2000 squared) as much power being delivered to the speakers, and correspondingly to an approximate SPL increase of the same 66 db. (10 x log(1995 x 1995) = 66 db).

Regards,

-- Al