[quote author="bcarso"]EDIT before hitting send, (composed before seeing mediatech's latest, but here it is):
Suppose we already have L and R signals that swing about rail-to-rail. Then this scheme doesn't help us, if we want to assure we never clip.
Why?
Well, take the two cases of L = R (even for a moment) and L = -R (again however briefly).
When L = R call the signal C. We drive our third wire with -2C---but wait---that overloads. So we have to drive with -2C/2 = -C. Meanwhile our L-R and R-L drives are each zero (C-C). The net can drive for each is just C, no different than we would have if the third wire were grounded.
When L = -R call that signal D and its inversion -D. We drive our third wire with -(D-D) = 0, meaning it is effectively grounded. Meanwhile wires one and two are driven with D-(-D) or 2D on the left and -D - D or -2D on the right. But this clips so once again we have to cut each by 6dB, so our net can drives are once again just what they would have been before resorting to this scheme.
What one could do: Use a crossover on each of right and left, do the scheme for low frequencies and not for high frequencies. Then one could realize the advantage of bridge drive and the approach towards ~6dB of bass enhancement, at the frequencies where localization is relatively unimportant. That is, at low frequencies we drive wires one and two with (L+R)/2 and wire three with -(L+R)/2, but at higher frequencies (L-R)/2, (R-L)/2, and -(L+R)/2, respectively. (EDIT so come to think of it only the wires one and two need the crossover treatment).
What would hang us here a bit would be some pathological bass mix that would undo our assumptions about monaural bass.[/quote]
If I thought they had a schematic in the owners manual I'd try to download one,,, right now specifics of this design have gone into my memory's dead letter dept.
Staring again from scratch we can ASSume worst case for clipping on sleeve is MONO signal, so lets reduce the drive on the common sleeve to -(L+R)/2 .
This makes L drive = L-(R/2) and R= R-(L/2),
Compared to 2 circuit only grounded sleeve, this gives us 1.5X for stereo or mono signals, more than twice the power.
Headroom is a bit of a mixed bag Mono will clip on the sleeve first, left only or right only will clip on the L or R line first but at the same amplitude.
Perhaps you could game this further by making assumptions about bass being mostly mono and HF mostly stereo but then you may not get the full 1.5X if Bass isn't mono or if HF is.
With this approach only real gotcha signal is if left is 180' out of phase or opposite polarity from right in which case clipping would occur in L and R legs at 2/3rds of 1.5x or roughly the same as a stock unit.
I'll take 2x power for only 50% more circuitry.
JR
