I suggest you take a PL509 (or PL519 or EL509 etc.) and set it up on your bench and characterise it. You need to know how much voltage swing you need on the control grid before you start designing circuits.
As you've sold off your HT variable supply, for the DC supply, feed an isolating transformer from a Variac, into a bridge rectifier followed by two banks of reservoir caps and a large choke (one built like a transformer) between them with a power bleeder resistor calculated to have a time constant with the total capacitance of no more than 2s so it will totally discharge in no more than 10 seconds. For the PL5xx filaments, you need 40V at 300mA, and hopefully you have a bench supply that can do that. Otherwise use a transformer and a rheostat, and adjust to 300mA RMS when hot.
You need to know the minimum anode voltage with zero grid bias to get your maximum output current, and should also test at 400mA to have some margin for loss of emission with ageing and to allow for differences in individual valves. Put an ammeter that will read up to 1A and down to <1mA between the cathode and ground for this test. Let the valve warm up at a moderate cathode current - say 100mA for ten minutes before making any measurements.
You also need to know the negative grid bias required for cutoff, at the maximum anode voltage you valve could ever see e.g with the PSU unloaded and the mains input 10% above nominal. You'll need to decide what minimum current you'll accept at cutoff, and will need to make sure your final design can always sink that much current right down to your minimum output voltage. If in doubt, measurements at 1mA and 0.1mA would probably be useful. Measure the current as voltage drop across a 10K resistor.
You may need over -250V on G1 to get full cutoff, but you only need a few mA current so to get the negative grid bias supply, capacitively couple another bridge rectifier to the transformer secondary, feeding a decoupling capacitor then a resistor (47K 5W wirewound would be good) and another decoupling cap, then use a physically large 500K pot rated at least 1/2W between the negative supply and ground. Decouple the pot wiper with 0.1uF to ground, and feed G1 from it. Start with maximum negative bias and crank up the Variac slowly till the anode voltage reaches the maximum design voltage (allowing for 10% mains over-voltage) while watching the voltage across the 10K cathode resistor. If it rises over 10V, your negative bias supply cant supply enough voltage. Then reduce the bias until you get 1V (100uA) and 10V (1mA), measuring the G1 voltage for each. The voltage drop across the cathode resistor adds to the G1 negative bias, so if you have -100V for 10V across the cathode resistor, your actual Vg1k bias voltage is -110V.