Rectifier ideas

[Osvaldo de Banfield]

Hi all guy's. I have a question (Idea) this night it come to my brain. It is oriented for all that loves simulation tools (not my best job) or theory.

Suppose I have a transformer with two secondaries (Each one having 2 * n2 and 2 * n3 turns in each one of them), and a primary of n1 turns. Both center taps are grounded, perhaps at the same earth potential, or not.

One of the secondaries (Say n2) is connected to a rectifier full wave, say a 5U4, 3DG4 or any other kind of rectifier, tube or solid, it doesn't care. This is LC filtered, with choke input filter. This gives power to a circuit, say the final stage of class AB or B amplifier.

Second secondary (n3) is wired to a second rectifier, also full wave, say a 6X4 or also any other rectifier (My original idea was vacuum tube rectifiers), and uses capacitor input filter, loaded by another load, which may be the previous stages, and with CRC filter cell.

Suppose also that I calculate the L of filter at n2 in such a way to resonate at line frequency with C of the filter on n3, scaled them by the turns ratio between them, say n2/n3, to reflect L and C via the transformer to resonate. The R in the second filter plus loading resistances in both filters are clamping the Q of such resonance. Add in L all the parasitic inductances in the transformer (Magnetization and leakage).

It also can be imagined with a only one secondary having a 2 * n2 turns, and the secondary having 2 * (n3 - n2) turns in each winding (My first idea).
If any other secondary is in the same core, all of them are resistively loaded, so no reactances are present in them.

Which would be the situation in the primary current (Pure sinusoidal?; In phase with line voltage?) and voltages in each secondary under this conditions? (Peaked?)

Thanks for reading this paragraphs and help me reasoning this situation.

[Elvee]

You should draw a diagram to illustrate your idea more clearly (and perhaps add some simple equations to make what you mean more obvious).
I think you want to make some kind of passive PFC.
It is possible to achieve some level of compensation, but I don't think it could be perfect, and more importantly it will be critically dependent on the matching of the power drawn on each secondary (plus the initial matching of the reactive components).

If you come up with actual values like voltages, capacitor values and current drain, simulation is very easy, but if we have to guess what you aim at, things will be more difficult.

[Elvee]

Here is a first tentative, based on the info you provided, and reasonable assumptions for the components like inductor, transformer, etc.
For the sake of simplicity, the secondaries are identical.
The transformer is a 100VA, and the supply is designed to provide 300V DC output, with a 4 to 1 power ratio between the outputs.

First pic is the situation when the power ratio is 1 : 1

Second pic is the normal, full power situation (4 : 1)
In this case the rms primary current is 446ma, an apparent power of 102VA.
The active power is 97.5W, which is good despite the ugly current waveform.


In the 1 : 1 case, the rms current is 249mA, apparent power = 57VA and active power is 52W.


3rd pic shows the currents through L7 and C2

And finally, the LC filter alone. The waveform is the cleanest of the three. Rms current is 358mA, Papp=82VA and Pact=75W.

In summary, the presence of the capacitive filter slightly improves the overall power factor, but the effect is very small and load dependent.

[Osvaldo de Banfield]

Originally Posted by Elvee

You should draw a diagram to illustrate your idea more clearly (and perhaps add some simple equations to make what you mean more obvious).

Yeah, but this wasn't the original idea, but certainly, it may act as a PFC in some manner. In a true PFC the L and the C are in the same circuits and don't resonate, here my idea was to be resonant and in separate circuits although coupled by the transformer. And the idea was an attempt to compensate
I think you want to make some kind of passive PFC.

It is possible to achieve some level of compensation, but I don't think it could be perfect, and more importantly it will be critically dependent on the matching of the power drawn on each secondary (plus the initial matching of the reactive components).

If you come up with actual values like voltages, capacitor values and current drain, simulation is very easy, but if we have to guess what you aim at, things will be more difficult.

Yes, like almost all tuned circuits, their working conditios are dependent on load conditions. Thanks for the time and the will to run simulations.