Stabilizing the high voltage

. woensdag 18 mei 2011
  • Agregar a Technorati
  • Agregar a
  • Agregar a DiggIt!
  • Agregar a Yahoo!
  • Agregar a Google
  • Agregar a Meneame
  • Agregar a Furl
  • Agregar a Reddit
  • Agregar a Magnolia
  • Agregar a Blinklist
  • Agregar a Blogmarks

I planned to stabilize the HV output with a big capacitor, got 8x 4000Vdc, 8uF from surpluscenter.
They were quite big, one pin was connected to case (would become very dangerous to put them in series) and they looked rather damaged.
Surpluscenter didn't make any problem about giving the money back.
Looking at these very big capacitors it made me think of ways to stabilize my power with the 3nF/35kV capacitors.
I want to put all the high voltage stuff under the hemispheres, to make it as safe as possible.
I don't know the frequency, output impedance, induction, parasitic capacitance's etc. of my power supply, I will measure that when I have it.
For simulation I made some assumptions, two 10kHz out-of-phase sine waves of 12000V peak with both 5k impedance, two TVR30 diodes and 20pf capacitance to ground simulate the power supply. A 600k resistor simulates the load and 4x 3nF make the 12nF output capacitor.
This simple simulation gives a (completely unacceptable) ripple of 60V pp:

It should be possible to disconnect the ground of the secondary windings of the flyback transformer and make an active ripple rejection circuit:
The circuit on the left is a simple feedback circuit which amplifies the ripple, the emitter of Q1 is connected to the ground side (now disconnected from ground) of the secondary windings of the flyback transformer.
The signal on the Q1 emitter is the inverted amplified ripple signal.
Zooming in on the output shows that the ripple is now only 1.6V!

The ripple depends on the amplification of the feedback circuit.
Making the amplification higher would make the ripple lower but also creates a higher risk of parasitic oscillations.
I will do the math later, this just shows that active stabilization is possible.
Also, the circuit is not save, it will need zeners between the emitter of Q1 and ground which can handle an output short-circuit without sending 12kV sparks to the 200V supply (a bjt amplified zener can do the job). The gate of M1 also needs to be protected with zeners.

1 reacties:

Sky Bird zei
Deze reactie is verwijderd door een blogbeheerder.