Supercaps (was PET composite video adapter)

Hi, I have designed products using 150F 2.7 volts caps, and like a lot of these things the devil is in the details. The supercaps take a while to charge -- we limit the current to 2 amps. The charge is not like most capacitors -- they have a sort of dielectric soak effect, where charge doesn't reach the remote areas for a while -- maybe 10's of minutes. We use them down to 0.5volts, and would like to go lower, but it's hard work. Simply turn joules into watts for x seconds to work out the energy available. In practice we get 300 joules out of the 550 stored in the above capacitor -- the rest is lost to self discharge, the bit below 0.5volts, convertor losses when the voltage gets low, convertor losses stepping up to 3.3 volts etc. The normal discharge and charge curves are quite different on these supercaps -- due to all this hidden energy. VoltageMinimum)/2 * LoadCurrent * Time. In my case (2.7+0.5)/2*0.05amps*1800seconds = 144 joules needed. Then Capacitance = 2 * VoltageWorking * LoadCurrent * Time / (VoltageWorking^2 - VoltageMinimum^2), in this case 41 Farads. Factor in a whole lot of losses and I needed 150 - 170 Farads to give 50mA for 30 minutes. The really neat trick would be to have put two in series -- CV^2 helps here, because (obviously) the capacitance halves, but the voltage squares. I played around for ages trying to balance the charge on the two caps -- isolated chargers -- the whole bit, but never got anything to work whose self discharge current wasn't more than the gain from having 5.4 volts. Intrigued by this -- any answers ? Regards, Gavin Melville.