This is a “Request For Comment” type of entry, pertaining to a specific idea, concerning ripple filtration.
Specifically, filtration based on a C1-L1-C2-L2-C3 filter, albeit a somewhat … funny one.
Imagine that from a sheer lack of space, you come up with the idea to create two, or even three choke windings, but on a single core. A multiple windings choke.
The origin of the idea came from a rework project of an existing amplifier, where the openings on the top side of the chassis were already drilled, the component locations “defined” and there was no extra room for an “additional” choke. The “choke” was defined. So was the size of it’s E-I core and the size of it’s bobbin.
So the starting point is as follows: You have a specific E-I core, and a specific bobbin, which can accommodate a specific count of turns, prior to be filled up to the very top. But these turns may be associated with a single winding, but also, just as well, they may be associated with two or more windings.
Based on this single bobbin, you are seeking, or exploring ways so as to find a means of a “better filtration” strategy. Let us assume that in all the possible variants, C1 = C2 = C3.
Now, my question to you is as follows:
Will such a setup, as presented on the graph below, provide a “better filtering” result, as compared to a simple, single winding choke, a variation of “fill-it-up-with-turns” type of single choke, and a topology of say C1-L1-C2-C3 ?
Look at this and tell me what you think:
The idea is as follows: The main thrust of ripple current flows directly from the rectifier, via C1, to ground. BUT …
A significant amount of ripple current also manages to get through, via L1, and is shunted via C2 to ground.
L1 conducts ripple current, and hence, produces a magnetic field, one that corresponds to the ripple current flowing via L1 and C1 to ground.
As this ripple current is still fairly significant, we shall make the number of turns within the L1 winding rather small.
Now, please consider L2 combined with L3. These two are actually a single winding, the second one, but I wanted to stress the fact and represent it via a simple graphical means, that they jointly have a much greater amount of turns than L1.
OK. So now, we have a yet smaller ripple current, flowing via L2 + L3, and this ripple current is also shunted to ground, this time via C3.
BUT … now comes what I believe to be the “catch”…
The ripple current, i.e. the “AC” component, flowing via L2 + L3, is actually flowing in ANTI-PHASE, with regards to the AC ripple current component flowing through L1.
As all three of these windings are on THE SAME magnetic core, we have a potentially “self-cancelling” situation for those ripple components. The magnetic fields, as generated by L1 and (L2+L3) oppose each other.
The strongest ripple current (“AC”), flowing via L1, is actually flowing through a fairly small number of turns, and generates a magnetic field, that is tightly coupled with L2 + L3.
But there is also a ripple current, albeit much, much weaker, flowing through the L2 + L3 winding, flowing in …. the “opposite” direction, and it’s AC component is much weaker, in comparison to the AC current component flowing through L1.
The turns ratio of (L2+L3) / (L1) should be arranged in such a manner, so that the ratio resembles the ratio of the ripple currents: ( I_L1) / (I_L2 + I_L3).
What I am getting at here is that I am trying to “understand” what will happen, in terms of efficiency of filtration, if we maintain the following equation:
(L2+L3) x ( I_L1 ) == (L1) x (I_L2 + i_L3) == constans == AmpereTurns.
My request for comment is of the following nature:
Is there a chance, that by filling up a single bobbin with two separate windings, the turns ratio of which is maintained in accordance to the specific ripple currents ratio as presented above, is it possible that we could, using such a setup, achieve “EXCELLENT” filtration properties ?
The idea is simple: If we are building a tube amplifier, we can assess, more or less, the estimated DC current draw of the device.
If we know the estimated DC current draw, then, we can “assess” the individual values of the ripple currents flowing through L1 and (L2+L3), simply by using the simulation package called “PSU Designer”.
Since we can get a fair understanding and assessment of the values of these currents, therefore we can also obtain the required “optimal” turns ratio between the section L1, and the section (L2+L3), which all can be placed on one and the same bobbin, and this – on one and the same core ….
Will it work ?
Will it work .. spectacularly ?
Will the “self cancelling” actually WORK, or will it be compromised by some form of “phase shift” occurring in “L1 -> C2″ current, in comparison to the “(L2+L3) -> C3″ Current ?
Will it work spectacularly, o maybe I should just forget the whole idea and stick to the C1-L1-C2-C3 concept, as it currently is implemented ?
Or maybe, instead of totally “nulling” out the AC ripple components, I should go for “equal turn” numbers for both of the windings, so as to “null out” the fairly strong DC current components flowing through this setup, and being satisfied with only but a “partial nulling” out of the AC ripple components, that go along with it ? This way, the net DC current flow would be essentially close to null, and hence the requirements for the size of the core could be miniscule …
Many thanks for your kind inputs.