You need to take a look at what you are expecting.
You want about 35 W out at 500V. That's 500V into a load of 7000 ohms.
Given 3.2 A at 24V, you are putting 77W in to get 35 W out. Your efficiency is about 58%.
Your smaller power supply can 2.1A * 24V, so that's 50W going in.
But, wait. You are reducing that to 19V so you've got closer to 40W going through the transformer.
Assume 45W to the transformer. You would need an efficiency of at least 80% in the boost section to get your desired output.
That 80% needed efficiency is a long ways from the 58% you are getting.
Why are you introducing losses into the system by reducing the voltage going into the transformer?
Limiting the current is the wrong solution.
You need to be more efficient.
Lose the buck converter, drive the transformer directly from the 24V source.
Check the signal driving the FET. If it has a slow rise time, you will lose power.
Ditch the 7805, and use a buck converter to power the processor. The 7805 is throwing away power that could be going to the booster. It may only be a couple of watts, but you seem to be pretty close to the edge with your power requirements.
Use a better FET. Check the Rdson for 5V, and see if there are FETs available with a lower "on" resistance.
Your "schematic" shows no resistor between the processor and the FET. That might cause ringing on the gate, which will also reduce your power output.
I'm no expert on switching power supplies - I have never built one.
The pointers above are based on the other questions about switching power supplies that I have read on this site.
If I have gotten something wrong, I hope someone will point out my mistakes.