50uH is even lower reactance than the 90uH coil at the frequency of interest, so you are coupling noise back into the power supply through L1 and L2. as i mentioned above, a Royer oscillator uses a single transformer, and all of the inductors in the circuit are windings on that transformer. i used to use this circuit all the time to drive flyback transformers from color TV sets (the "old style" flybacks that had no built-in rectifiers, but a large "pancake" winding). you mentioned in your first post you had some difficulty getting this circuit to oscillate. usually when it takes a lot of work to get an oscillator circuit to oscillate, there's either something missing (like adequate feedback) or something backwards (like something non-inverting when it should be inverting). using RF chokes for L1 and L2, and having L3 separate, instead of all 3 inductances coupled on a common core is where your feedback path has gone missing. you probably had to make drastic changes to all of the inductances until you finally got oscillation. it looks like you are relying on the diodes to provide a feedback path. instead the feedback path should be through L1/L2 inducing current in L3 with L1 and L2 being a center tapped winding on the same ferrite core as L3. the diodes might not even be needed. change the values of R1 and R2 to something that won't slam the gates on (something like 20K rather than 100 ohms).
L1, L2, and L3 should probably be about 10 times the values you currently have, and the capacitor one tenth or smaller than what it is now
if you need to know how to make L1, L2, and L3 into a transformer in LTSpice, you add the following SPICE directive:
K1 L1 L2 L3 0.95
this makes a mutual inductance "K1" with L1, L2, and L3 included, and sets the coupling factor to 0.95 (you CAN use 1.0, but you will never get 1.0 in any real-world transformer). once the Spice directive is set, LTSpice will add a polarity dot to L1, L2, and L3. you will want the polarity dot on L1 at the end of the coil that goes to M1-Drain. on L2 the dot should be at the end of the coil that connects to the power supply. on L3, with the dot on one end of the coil, the circuit will oscillate, if you flip L3 around the other way, it won't oscillate. unlike the circuit in post #1, you will find that once you get the circuit oscillating, that there's a wide range of resistor values for the gate circuits that will work, all that really matters is the ballpark of the gate bias. in an oscillator with adequate feedback and no inversion errors, the oscillating condition prevails for wide ranges of component values.
if you look at the circuit diagram in post #4 you will see that L3 should actually be smaller (fewer turns) than L1 or L2, and not connected between the drains of the transistors, but only providing gate drive for the transistors. if you look at post#4, you will see that the output is taken from a secondary winding of the transformer. in that case you would add L4 as an output winding, and add it to the K1 spice directive, simply by placing L4 into the list of inductors, so it would then read "K1 L1 L2 L3 L4 0.95"