hi smps experts
i making my smps transformer for half bridge smps, i was thinking about shielding the transformer, with cooper foil, like one turn of cooper foil covering the primary and one turn covering the secondary
Wrapping metal foil around the windings will become part of the transformer circuit.
I wonder if the copper shields need to be earthed?
Easy with copper since you can solder on leads. Much more difficult with aluminium.
Does a gauss band that works at LF, also work at HF?
Would that help with shielding?
a earthed coppershield, not fully closed turn (obviously or kaboom) in between pri and sec serves as a short to ground for any capacitve coupling pri-sec and is good practice , when necessary of course.
a gauss band , is that the same as a belly band or a full , closed copper band turn outside the windings?
It is common practice to shield the whole of the SMPS to stop any radiation.
The common mode choke will also radiate and any transistors and heatsinks used that arent earthed.
internal copper shields are electrostatic screens, and have the following constraints:
- must not create a shorted turn
- needs to have its ends insulated.
- must be connected to an appropriate AC return. A primary screen should be connected to either +Vdc_in or -Vdc_in (the input supply return). And a secondary screen should be connected to either Vout or 0V (its return)
[note my careful avoidance of the term "ground" - I use "protective Earth" when I mean exactly that]
- should have its connection in the center (lowest impedance from either end)
- must be very thin (to minimise eddy current losses)
- must have high resistivity (to minimise eddy current losses)
- needs a low inductance connection - the interconnect inductance controls the HF behaviour of the shield. this is true of all shields, be they internal electrostatic shields or external belly bands. long skinny wires pretty much ruin shields.
one cute trick for a foil winding is to add one extra turn after the DC end (usually 0V) as a shield with a free interconnect. its very easy to do, but may increase shield losses because the foil is invariably a lot thicker than a separate shield could be, and the shield connection is highly asymmetric. on the plus side its really, really easy to do.
A Gauss Band, or belly band, or flux shield has an entirely different set of constraints. Unlike an internal E-screen which is fully immersed in the transformer flux, a belly band is loosely coupled - its only the leakage flux that cuts it, and that ought to be fairly small.
so no matter how thick you make your belly band, you have almost no influence on the transformer operation at all (which is what we expect) - or to look at it another way, the eddy currents that leakage flux causes in a belly band are pretty much constant. its loosely coupled, so behaves like an independant current source.
So to minimise losses in belly band, you need to:
- form a closed loop
- use thick shield material
- use conductive sheet material (these two minimise I^2R by minimising R)
all three of which are the exact opposite of an internal electrostatic screen.
it also needs:
- to be insulated (thats an electrical safety thing, it wraps around the xfmr so gets close to pri and sec and core)
- to have a low inductance connection to a suitable AC return (I have used both -Vdc and 0V in different designs, for reasons of convenience)
a belly band can also be used to connect the core itself to some convenient AC return.
I wish there was a reasonably prices source for pre-made mu-metal boxes or that more transformers came with formed shielding. The only really effective shield I know is a foot or so of space.
skin depth in Copper at 50Hz is ~ 10mm, so 2 inch thick copper boxes ought to work.....
I've never used mumetal, but IIRC its also very fragile - a good whack can drop its permeability significantly.
I did find this convenient formula for low-frequency shielding effectiveness though:
SE_LF = 20*log(1 + 0.5*u_r*t/r) dB
u_r = relative permeability of shield material
t = thickness of shield material
r = distance from field source to material
|All times are GMT. The time now is 01:16 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio