[SOLVED] 1-Turn inductor (transformer's primary)

Status
Not open for further replies.
I think I'll go with a single (and bigger) ferrite core transformer (E71/3C90) running at 50 kHz, to minimize the switching losses (175A). Is it a good option to insert a capacitor in the transformer's primary circuit to avoid any DC component (I've seen it in various schematic diagrams)?

The primary inductance will be 10uH (one turn). I guess I'll have to use 6 layers of copper foil (0.2 x 35 mm) to build the primary winding. The secondary winding will have 18 turns.

So it's OK to wind them like that:

P (1 two-layer turn) - S (6 turns) - P (1 two-layer turn) - S (6 turns) - P (1 two-layer turn) - S (6 turns)

.. then connect all the primary windings (3 x 1 turn) in parallel and all the secondary turns (3 x 6 turns) in series?


PS: If the leakage inductance will be a problem, I'll try a resonant design later.
 

Heres my 2 pennies worth. The coupling between primary and secondary is mainly through the core, so any magnetic field not linked into the core will result in leakage inductance. Now when you pass AC current through a turn of wire on a core, there is skin effect AND the current distribution will be that the current will try to flow through the wire with the minimum current closest to the core and the maximum away from the core. So the maximum current also has the greatest air gap between it and the core.This is because of the repulsion between the wires magnetic field and the field in the core. So as the wire gauge increases, so does the leakage inductance, this is highlighted by the use of (thin) copper tape windings which are harder to deal with then (thick) wire windings.
I can not see why it is linked to the size of the core, but the max/min distance of the wire to it.
I can now see an additional benefit of my suggested solution as the 15 primary winding could be wound with wire with a thickness one fifteenth of a single winding.
I would suggest an experiment to prove the limits of leakage inductance. Get your core and put a huge short circuit turn on it ( bit of 25mm copper tape soldered end to end?). Wind a practice primary litz/tape/wire. . . on it and measure the leakage inductance. Hopefully this will be extremely low, so I would resonate it with a series 100 PF capacitor, and using a RF generator and voltmeter find its resonant frequency, calculate the leakage inductance. Then try the other possibilities so as to find its practical limits. Let us know
Frank
 
Last edited:

    V

    Points: 2
    Helpful Answer Positive Rating
Cool! Give a try. At 50kHz even the effect of leakage will be reduced.
And by the way adding a capacitor automatically nullifies the leakage reactance, if chosen correctly.
But I didn't quite get the idea of a series capacitor in the primary. Sounds cool theoretically, but do you really want to place a 30V 200A capacitor. Have you already found out which type of capacitors you wanna use?
At 50kHz the capacitive reactance would be 3.3 ohm per uF. When referred to the secondary, this translates to 3.3*36 = 120 ohm per uF. since you have three in series, you get a total of 360 ohms per uF
Your leakage will be in the range of 20 ohm to 50 ohm. So you need to get the capacitive reactance down to approx 20 to 50 ohm. That means a capacitance of approx 15 uF. Do you really want to use 15uF 200A capacitors!!? (Expect nothing cheap)

Electrolytic type is out of question due to high dissipation factor (typically about 5%)
X7R ceramic is also not impressive. (DF = 2% @ 50kHz)
C0G ceramic may work (DF = 0.07% @ 50kHZ). But what about the current rating?
Polypropylene film may work (DF = 0.1% @ 50kHz) Costs about Rs 25/- (0.4$ on Mouser) for a 3.3uF capacitor with current rating of approx 5A RMS.

I am also designing a similar X-mer with current rating approx 100A. Please keep sharing your experiences.
 

I was taking into consideration the "two-layer-only" limitation (posted above) for a copper foil winding, that's why I thought about three separate two-layer turns parallel connected for primary winding. I also took into consideration the primary/secondary windings interleaving recomandations.

Maybe your 15-windings suggestion was supposed to be a better option but it's already hard to wind 6 separate copper foil windings (especially the coil terminals).

The coupling between primary and secondary is mainly through the core, so any magnetic field not linked into the core will result in leakage inductance.

If you're talking about the 1-turn winding, I have to mention that it's actually one turn and a half (to put the coil terminals on opposite sides) so the magnetic field through the 1-turn winding will be closed enough.

So as the wire gauge increases, so does the leakage inductance, this is highlighted by the use of (thin) copper tape windings which are harder to deal with then (thick) wire windings.

Are you suggesting that 0.2 x 35 mm copper foil it's a "thick" wire? Should I go with 0.1 thickness?

Thank you very much for your suggestions.. I'll probably make that experiment (resonant frequency & leakage inductance calculation).
 

And as far as for experiment, I already have conducted quiet a few. Here are the results.

Resonating series capacitor: 1.2uF polypropylene.
Toroid core: OD = 45mm ID = 28mm Height = 11mm
N primary = 4
N Secondary = 120
N auxiliary (bifilar with primary) = 4

Xm = 52uH
X1 = 0.8uH (primary leakage)
X2 = 172nH (leakage between the bifilar primary and auxiliary)
 

Quoted from **broken link removed** (pag. 556):



(I must admit I didn't think about the capacitor current rating/technology)
 

Thats a half bridge circuit, not intended for high power rating. Although the concept can work for full bridge, think about the rating of capacitors.
 

I posted the wrong picture - actually, it was the same thing for the full bridge connection (you might check the linked pdf file).
 

I'm skeptical that one turn is actually an optimal design. With f=100khz, Vin=24V, and D=0.5 you'll get 326mT of ripple with one primary turn. At 100kHz that's huge, and your losses will be tremendously dominated by core losses. I would speculate that the optimal flux swing is less than 100mT, so three or more turns would be preferable. Make the primary out of copper sheet (20mil thick is probably enough) and sandwich it between a split secondary.
 

Hello there, using a toroid with the pri wound on the core and a single turn of copper organised by a large diameter tube thru the middle of the toroid forming the 1T sec can and does work.

However the mech side of it is not easy to prototype.

Also it would be better to have a centre-tapped secondary so that you get only one diode drop in the output.

The easiest route would be to use say 4 transformers with 1+1 turn secondaries side by side foil on an ETD59 say (for ease of size) with a pri winding above and below the sec, i.e. a sandwhich construction to reduce leakage, the 4 primaries can go in series and the rectified outputs in parallel, 44 amp each approx.

This is do-able and gives room to put snubbers on the diodes which you will likely need...

:0) Anna.
 

Actually, I've changed the inverter's design by using a LF (50Hz) output transformer. This way, I didn't have to use a HF transformer anymore.
I don't have enough time (and tools) for now to design a proper HF transformer. But your solutions seems pretty doable, thanks again for your efforts.
 

Have you already designed the LF transformer? If yes, what is the approx weight of copper required?
 

I'm using two LF output transformers from two identical APC 1500VA Smart-UPS (spare parts) which fit my needs perfectly (24V battery, 230V/3kW output).
I'm using a separate MOSFET full bridge for each transformer (driven by the same sPWM signals) so I could shutdown one transformer (bridge) when the load is small to minimize transformer (and MOSFET) losses.
 

Status
Not open for further replies.

Similar threads

Cookies are required to use this site. You must accept them to continue using the site. Learn more…