P.S idea

[noyan]

=============================================
You can tell this to ARC as well. They might even have a patent for this.
=============================================

Thank you very much! I am so sorry because i am late Someone had already thought this and have a patent! . Anyway, what i am thinking was right and i can arrange my powersupply unit in same way. Of course i will check the frequency response of the low-pass circuit and whole circuit by using an oscillator.
Thank you again analog_sa!
Noyan

[Eva]

I analysed ARC circuit.

Current up to 5Khz comes mainly from the 470uF electrolytic capacitor. Between 5Khz and 80Khz, current comes equally from both electrolytics. Between 80Khz and 5Mhz the 2uF capacitor provides the current. The 10nF bypass is a bad joke. Actually the whole circuit is a bad joke and a marketing tale because the electrolytic capacitors are the ones supporting audio frequencies.

Supply impedance contains a large resonant peak at 93Khz and another bigger one at 6Mhz, but both disappear if L1,L2,R23,R24 are removed The 93Khz peak will probably compromise amplifier stability.

If you want a really low and non-inductive suppy impedance, parallel several medium sized electrolytics with the help of a double sided PCB, like four 330uF or six 220uF. For low inductance it's important to chose radial types with 10mm or less terminal spacing, but classic tube amplifier wiring is going to be 20 times more inductive than the capacitors themselves, so who cares...

[jnb]

Why don't people like electrolytics? Isn't it because capacitors increase their impedance at high frequencies, by themselves, without needing the choke?

Do you think the electrolytic would sound different in place of a PIO if the impedance is the same?

What is your main goal here? Mine would be to have the impedance of the supply equal at all frequencies. Your PIO will help above frequencies where the electrolytic has a higher impedance. Wherever that is, it doesn't matter much as long as the impedance remains balanced.

You can now buy electrolytics that work well at high frequencies.

[TechGuy]

You can reduce Ripple using a "Coupled inductor". Use a bifilar winding. Use one wire in series with the voltage source and the output load. Connect the other wire, with the voltage source and a cap in series to ground.

V+ -------^^^^^---------- V out
-------^^^^^-+
|
Cap
|
GND

The ripple is coupled with the second winding and is filtered with through the Cap. Ideally a Cap with a very low ESR should be used, and you want tight coupling between the two winding, if you need very low output ripple.

White Paper:
http://www.hamill.co.uk/pdfs/azrtatad.pdf

[analog_sa]

Originally posted by Eva
I analysed ARC circuit.

Without knowing the parasitics of the particular capacitors used?

[Eva]

Originally posted by analog_sa



Without knowing the parasitics of the particular capacitors used?

I took the parasitistics of the capacitors that I usually employ, which are in the low side. Higher parasitistics would just shift resonances down and make things worse. Capacitor parasitistics didn't make much difference below 1Mhz anyway, the 93Khz resonance and the 5Khz crossover are mostly inherent to L1 and L2 and capacitance values.

Systems with several L and C usually exhibit counter intuitive behaviour. If you do the homework and analyse the circuit, you may reach some interesting conclusion or even learn something. If you just complain, you won't