Its obvious what its doing if you ignore terminal 7, U3 etc. - its a simple voltage doubler using C301, C302, U1, U2 and winding 6-8. The extra tap '7' on the winding derives a lower output voltage via U3 because it doesn't swing as high as '8'.
However the voltages marked on the schematic are as clear as mud,specifically the AC voltages between transformer pins 6 & 7, and 7 & 8. If the latter is actually 545V RMS, the HT rail from U1/C301 would actually be up around 1.5KV !!!
The rubric:
POZNÁMKY:
STEJNOSMĚRNÁ NAPĚTÍ MĚŘENA PROTI KOSTŘE NEVYBUZENÉHO ZESILOVAČE (BOD"7") PŘÍSROJEM AVOMET II. S OZNAČENÍM "+" NEBO ÚDAJE STŘÍDAVÝCH NAPĚTÍ S OZNAČENÍM "~" BYLY ZÍSKÁNY MĚŘENÍM NA VYBUZENÉM ZESILOVAČI NA JMENOVITÝ VÝSTUPNÍ VÝKON.
is some help as that (google) translates to:
COMMENT:
DC VOLTAGES MEASURED AGAINST THE FRAME OF AN UNEXCITED AMPLIFIER (POINT "7") BY AVOMET II. WITH THE CODE "+" OR AC VOLTAGE DATA WITH THE MARK "~" WERE OBTAINED BY MEASUREMENT ON AN EXCITED AMPLIFIER FOR NOMINAL OUTPUT POWER.
which clears up some of the confusion.
Assuming the no signal voltages (no suffix) reflect the peak AC voltages of the windings feeding their rectifiers, that gives 6 & 7, 125V RMS (177V Peak), and 6 & 8 177V RMS (250V Peak).
Probably the best bet for your negative bias supply would be a capacitively coupled charge pump driven by pin 8 (giving you -250V at no load), then regulate it down to approx -200V, with a string of six 33V Zeners setting the base voltage of a PNP pass transistor. Size the charge pump coupling capacitor so the unregulated voltage drops to about -210V @50mA load to keep the pass transistor dissipation reasonable. Ideally you should feed the Zeners from a current source, e.g a depletion mode N-MOSFET in the classic JFET current sink circuit, with a source resistor selected on test for 1mA through it.
If you want better regulation, replace the Zeners with a TLV431 shunt regulator cascoded with a HV N-MOSFET with its gate biassed from a tap in the '431's feedback divider.