Voltage V Current high voltage transformer rectification query

[Chris Wilson]

I have a 5.75 kVA 6.6kV single phase mains transformer. The output voltage is greater than my needs. It comprises two separate bobbins, see http://www.gatesgarth.com/transformer3/transformer3.html

The primaries are around the core, which is a full core, with laminations top and bottom, within the U shaped mounting channels. That's hard to see from the photos.

The secondaries are on the outside of the primaries, wound around plastic spacers over a half inch wide, forming channels for air to be blown through if the red temp switch triggers a blower fan.  In other words there little chance of arc over, secondaries to primaries. It's rated for 24 hours a day full output into an RF generator for drying wood glue used to make furniture, in seconds, not hours.

The "tap" between the two secondary windings, one per bobbin, is the insulated wire seen behind the Paxolin panel mounting the ceramic terminal insulators. As such, although not used as an actual physical tapping with a termination, is fully accessible. I have three  options that I am aware of.

1: Spur off the "tap" wire and now have effectively three terminations, one being a centre tap, and use a full (none bridge) wave rectifier and smoothing a smoothing array of 15 x 450uF 450 volt working electrolytics, (which I have built) , with bleeders, or two 10 uF 5kV oil filled in parallel with bleeders, (which I have got already).
 Should this give half the voltage, circa 3300 Volts AC, and DOUBLE THE CURRENT? Under test it does indeed give half the voltage and seems happy with no load like that, but unsure of what the current will be?

2: Split the secondary joining wire and parallel the two secondaries in correct phase. It's sat here right now in this format giving about 3600 Volts AC via a 5 megohm resistance voltage divider into an El Cheapo meter, so probably about 10% over loaded voltage via a proper meter. Seems happy....I then run a full wave bridge rectifier on the paralleled output, and a similar electrolytic capacitor array filter, or two 10 uF 5kV oil filled in parallel (which I have got already). This should give double the current of the standard series wired secondaries.

3: Split the secondaries completely into the two bobbins, two wires from each. Paying attention to phase ach bobbin feeds one full wave bridge rectifier array (I have some built from another HV supply project that will laugh at the load and voltages). The rough DC volts outputs from each bridge rectifier array are then joined in parallel at the filter capacitors. Idea behind this? It may help stop eddy currents in the transformer if each bobbin does not give an exact half share of voltage. I could put a low ohmic value high wattage resistor in each rectifier leg to help balance the levels if required. Again, this should give half the "as manufactured" voltage, and double the current.

I am aiming for a lowish (50% to 60%) fan blown duty cycle with 4300 to 5000 Volts DC at about 1.6 Amps. I can feed it at least 40 Amps from the mains supply here at home. My doubts are current multiplication and the best, rather than the cheapest, way to do this, short of having exactly what I want custom wound. In particular which method makes best use of the input supply cycles?

Thanks for looking! And yes, I know it's dangerous