"Floating" is an unfortunate choice of words in this case. Floating heaters WILL cause hum, but this supply is not floating. It's biased above ground by the voltage divider R3 and R4. As a matter of fact, it's really biased too high above ground. Rafa used the same values for the voltage divider as found in the original MB-1; but note that in the MB-1, the pickoff for the biasing voltage appears after a tube rectifier and two 1.5K resistors. In Rafa's scheme, the solid-state rectifiers, the small resistor and the choke mean that the DC at this point will be higher than in the original design.
Maximum positive heater-cathode voltage for the 12AU7 is 100VDC or 200V total DC + peak signal. The MB-1 cathode follower's cathode sits at 163V. So we'd like to bias the heater to about +100V above ground. (Pultec chose 115V). So the resistor values may need to be altered from those in the original schematic, but I'm not going to do the math for you :wink: If you need to increase the division, better to increase the value of the upper resistor. The heaters are not drawing current from this voltage divider, so the impedances can be pretty high, to avoid loading down the plate supply unnecessarily.
You should avoid biasing the heater supply much more than +100V above ground because you don't want to exceed the negative heater-cathode rating of the first tube.
Eliminate R2. Its value is too low and it's not needed anyway because the heater biasing voltage divider will act as a bleeder. Also, to achieve full cathode temperature and emission from the tubes, insert a rectifier diode between the 12V regulator and the "low" side of the heater circuit (anode to "com", cathode to the heater low side) to bring the heater voltage up to 12.6V.