Making this resistor larger in value runs the risk of increasing the noise level.
  Some have a preference for bringing the feedback to the cathode, as was done by Futterman in his OTL. While others do not like the idea that the phase will be inverted by any amplifier. (Of course, they must be putting a great deal of faith in the recording process, the mix-down process, the player's internal circuitry, and their own preamp or line stage not having already inverted the phase.) For these tube enthusiasts, the amplifier can be reordered.
  By reversing the outputs of the Split Load phase splitter, the amplifier will no longer invert the input signal's phase. This will leave the input tube's cathode as the inverting input to the amplifier. Thus, by bridging the output to this cathode, a conventional feedback amplifier can be built. One liability to this approach is that the output stage's output impedance will greatly increase, although the open loop gain will increase as well, which will bring the output impedance down. A second liability is the potentially dangerous DC offset that results from connecting the feedback resistor to the output from a cathode that is several volts positive. In the large Futterman amplifier, this was not much of a concern as the loudspeaker can easily withstand half a volt of offset and its low DC resistance shunts away much of the offset voltage to ground. Headphones, on the other hand, have much higher DC resistances (32-600 ohms), which will not voltage divide as deeply as the speaker and they are much more sensitive to the DC offset. One solution is to capacitor couple the feedback resistor to the cathode. The value of capacitor needed for this task is found by the following formula:
         C =  159155/R/F (in microfarads)
where R equals the impedance of the feedback resistor and F equals the low frequency cutoff. Still, this capacitor will be on the order of 30-100 microfarads, which probably will mean more electrolytic capacitors in the signal path, as a film capacitor would be huge.

  Although the choice of output tubes is fairly wide, one tube stands out. An EL86 (AKA: 6CW5, 8CB5, 8CW5, 10CW5, 15CW5, 45B5, 45CW5, N119, N379, PL84, UL84, XL86) would work handsomely. This is a real sleeper of a tube. It is a low voltage, high current pentode that is linear. With only 50 volts on its plate it is able to conduct over 100 mA's of current, which means a B+ of only 100 volts would work with this tube. This 100 volt power supply would also supply the heaters of two 45CW5s in series (plus one 100 ohm resistor).
  (The EL86 makes an excellent pass device in a series regulator. It also works well as the shunting device in a shunt regulator, if the output voltage is not too high: up to 250 volts.)   
  Alternate tubes are the 6AQ5, 6L6, 6V6, 12BY7, 12HG7 (a major sleeper, this tube has a frame grid that gives it a transconductance of 32 mA per volt), EL84 (6BQ5), and the SV83. All of these tubes will yield only about 30 mA of peak output current, which although sufficient to drive just about any headphone to ear-aching levels, headroom is always welcome.  Possibly the coolest choice would an amplifier entirely populated with 6BM8s. This tube has a big following and it would work well in this application because it comes with its own triode for the Cathode Follower that is needed to feed grid 2.

pg. 7

<PREVIOUS

NEXT >