P.S idea

Hi All,

My idea is like as follow;
-In general, tube amplifier P.S. units contain a big electrolytic cap after choke trans. This electrolytic capacitor is connected in serial between OPT and tube anode. Sound flows from this cap. It is effected on the sound quality, especially on mids and highs. I thought that, if i use another small choke filter bobbine serial to this big electr. cap, then only low (Bass) freqs. can pass on these circuit. For the mids and highs, i can connect two different valued PIO caps as parallel to this circuit. Mid and high freqs. will pass easily with less distortion trough these PIO caps. Cap values on the drawing are only for example. Any comment?

Unfortunately is not that simple.

For starters, a capacitor in series with an inductor forms a resonant circuit that will shunt the signal at the resonant frequency.
Additionally, all the other inductors and capacitors will also modify the overall frequency response.

If you have access to any sort of simulation software, please perform an AC simulation with the values you plan on using, and you'll be surprised at the oddball frequency plot.

Be sure to include equivalent series resistance of both the inductors and capacitors when you run the simulation.
p.s. unit

The second choke and it"s cap is not for the cancellation of ripple. Their job is to pass only lower freqs. for example below 1khz. Over 1khz, other two PIO caps will play role. Anyway, i am concern like fernando_g about the low pass filter it will work as i wish or not. I mean they will play role as band pass filter or low pass filter.
The idea is terrible because it comes from a brain that thinks in terms of false audio metaphysics rather than true circuit physics.

If you want to play games with LCR networks and win, you have to learn the physics and the math related to these LCR systems (the real rules of the game).


diyAudio Moderator Emeritus
2002-10-24 10:19 pm
I think that what Eva was trying to say in his/her usual inimitable style was that you may have caused more problems than you've solved. You have a problem which is largely imaginary, and the means you have used to solve the nonexistent problem will almost certainly upset the dynamic stability of the supply. Run some simulations, and be sure to see what happens when you step the load current- I bet you'll see some astonishing ringing.


Paid Member
2002-08-14 1:47 pm
SY said:
You have a problem which is largely imaginary

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


  • ph3ps.gif
    37.9 KB · Views: 260
PS unit

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!
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 :D:D:D 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...


2006-12-30 11:55 pm
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.
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

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:
analog_sa said:

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 ;)