Singe triode driving a conventional high-ratio output transformer wired backwards.
 Additionally, a single-ended amplifier must have an idle current at least equal to the peak current into the load divided by winding ratio in the case of a step-down transformer and multiplied by the winding ratio in a step-up transformer. Our 8 mA now requires an idle current of 200 mA for the triode! This is a good amount of current and would certainly saturate the output transformer, as the implicit air gap present in transformer due to sloppy construction would prove insufficient to allow such a high unbalanced current flow through the transformer's secondary. The solution is to use a high quality single-ended, i.e. air gapped, interstage transformer with a center tap and a low winding ratio of 1:1 or 2:1. This transformer will be expensive as this type is fairly rare.    Alternatively, if a center-tapped choke can be found, it can be used as an inductive load for the output tube and by connecting the center tap to the B+ connection provide a balanced output. This was a technique that was used back in the days when tubes cost more than chokes.
 This arrangement will result in a stepping up of the plate voltage swing by the transformer's winding ratio, usually between 14:1 to 30:1. So if the winding ratio is 25 (nominally a 5,000 ohm to 8 ohm transformer), then the plate need only swing a peak of 20 volts to force the secondary to the desired 500 volt swing.    The math looks good, but it is unlikely that the output waveform will. Several problems arise with this approach. The first is that the output transformer does not step up voltage as nicely as it steps down voltage. Why the discrepancy? The capacitances internal to the winding of the transformer are largely shunted in normal use by the 8 ohm load, which is primarily resistive in nature. But in this application, those capacitances are only augmented by the capacitive load represented by the headphones and are shunted by only the relatively high plate resistance, all of which adds up to a very poor frequency response. Connecting the output transformer to the triode's cathode only decreases the rp by the mu plus one, which is still an order of magnitude greater than the load impedance represented by a loudspeaker.
 Unfortunately, most chokes are intended for use in power supplies and are usually not as well made as audio transformers, the result usually being a poor frequency response and odd oscillations.
 One solution might be a single-ended transformer, if one can be found with a center tap on the primary, as its primary could be used in place of the center-tapped choke, while its secondary is left floating. (Actually, a sweet dual use headphone amplifier might be made with the secondary switching from no load when driving electrostatic headphones and then switching to dynamic headphones when the need arises.)
 Singe triode driving a conventional high-ratio output transformer wired backwards from its cathode.
 pg. 3