This is my work in progress filter for the Power Supply of the GM70 project – some photos.
The setup is supposed to constitute two independent blocks, each consisting of symmetrical voltage rails: positive 890 VDC and negative 890 VDC with reference to ground, which is running through the “middle”.
The resistor networks are simply equal-valued resistor networks, serving as equal-voltage dividers, but at the same time, bleeder networks, which after switch – off shall discharge the lethal charge / voltage of the capacitors. Each “single” resistor consists of a series grouping of two individual 560K resistors within a pair totals up to c.a. 1,1 Megaohms. Such series pairing is supposed to alleviate / mitigate the risk of high voltage arcing over an individual resistor.
A voltage divider, consisting of a total of 1,1M + 1,1M = 2,2 megohms, when exposed to a voltage of c.a. 900 V DC, will conduct a current of 900V / 2200k = 0,4 mA. This should effectively swamp any potential variations of the leakage current within the individual elko capacitors and even out the series DC voltage division between them.
Each choke is of the dual winding, quad connector variety. Actually, from the puristic point of view, these are rather 1:1 winding ratio transformers. Each is connected in the following manner:
The “underline” denotes the fact that both windings are on the SAME magnetic core.
I route the current through the first winding, so that its DC magnetizing current works “against” (i.e. antiphase) the DC magnetizing current flowing through the second winding. One would think that this does not provide any net benefit. That these currents “null” themselves out, and provide no filtering benefit. Well, in terms of “DC current” – indeed, they do.
But here comes the CATCH: Please take note of the MIDDLE “C” between the winding.
This “Middle C” is a shunt path to ground, for the AC ripple component flowing through the first winding. And guess what. The ripple component flowing through the second winding is at a different PHASE relative to the ripple current component flowing through the first winding.
So, in terms of AC, we have a totally different “flow” of currents, when compared to the flow of DC currents. The AC current shunted to ground from wining no. 1 actually works “against” the ripple current component in winding two. Essentially, the ripple currents “fight against each other …
The concept works. If you do not believe me – simply construct a makeshift test setup and measure the ripple component on the output of such a filter.
Before you engage in any expensive choke purchases or investments, for starters, it suffice that you use a “safety” torroidal separation transformer as a choke.
A separation transformer as in: 230V AC to 230 V AC, or such as a 115V AC to 115V AC.
Or maybe you have two separate 115V AC primary windings in a transformer in your drawer somewhere ? Try it.
You will notice that as opposed to two “independent” chokes, such “antiphase connected” transfomer-choke actually renders much better ripple filtering results.
Ah, yes, indeed, a hawks eye might have noticed that there are yet some more “through-holes” drilled through the PCB. These indeed are required. I am waiting for some additional bypass foil capacitors. As soon as I receive the parcel, as purchased on the Polish auctioning service, www.allegro.pl, I will be putting them in, as bypass caps for the elko’s.