Did you check the frequency dependence? What you built is a 1:10 transformer and it transforms the voltage across the primary. As the primary is an inductor, it's voltage may be about 10x less at 3 MHz than at 30 MHz. With the sampling circuit as load the frequency dependence is probably less. Maybe one can run the ciruit with a near short on the secondary, like a 1 Ohm resistor.
How much signal do you need to monitor the transmitter?
Can't you just use a resistor? If you use a 5 KOhm resistor to feed the 50 Ohm sampling line it will also get 0.5 W at 50 W RF out. Sampling signal will be 0.5 Vrms, roughly frequency independent. You can split the resistor into two parts to reduce capacitive feed-through.
Regards, Dieter