It is kind of confusing.
The shunt regulated power supply for the Paramount driver is currently 350 volts. It is protected with a string of Zener diodes, until the shunt regulator tube warms up. In the new design there is a change that exposes the plate load transistor to a possibly greater voltage during the start-up transient, so we had to drop the regulated voltage to 300 volts. This is the effect of the resistor change PB describes. The shunt regulator tube now also has 300 volts on it, so it operates a bit more conservatively.
In the current production, the driver uses either an LED or a shunt-regulator chip to provide fixed bias without electrolytic bypass capacitors. However, that makes the range of actual plate voltage wider than would be the case with ordinary cathode resistor bias. That range is typically 140-220 volts, but can vary a lot with different tubes. There are two problems:
1) With the lower regulated voltage, some 12AT7s will run out of voltage headroom (compliance) when driving a 300B
2) The direct-coupled 2A3 is very sensitive to power line voltage variations, and the driver plate voltage should really be optimized for the particular power line voltage in your home. Even at the design value of 120vRMS from the power line, different tubes will have different plate voltages and can be uncomfortably off optimum.
For these reasons the new board has an adjustable bias voltage, so that the driver plate voltage can be set precisely. Normally it would be around 175-200v, same as the current design - only the C4S headroom (compliance) has been reduced, not the actual driver plate voltage. The adjustable bias arrangement we have implemented will not permit a bias below 2.5v, so the 12AT7 has too much gain and we went to the 5670 - which is also more linear and widely available in premium versions. The adjustable bias does make it easier to use a different tube; I plan to change my personal amps over to 6SN7s to reduce the excess gain.
