Easy peasy
Advanced Member level 6
Sorry to say, but text books and any real world experience show that for a common two winding transformer there is flux on the pri that does not link with the sec & flux from the sec that does not link with the pri - therefore - there are in fact separate and distinct leakages for each winding - if you operate a transformer in the forward mode i.e. pri current = reflected sec current, then the leakages can often be lumped to one side or another with little practical or noticeable difference,
however for a flyback - this is not the case, the pri winding is charged with current and there is no current in the sec, then the energy has to be transferred to the sec winding by the current building up quickly through the sec wdg and load - the o/p cap in this case - if there were no sec leakage this would happen instantaneously and the flyback voltage would not spike up as badly as it does as the transformer tries to force the EMF high enough to make this happen - in other words we would not see such a large turn off spike on a flyback if there were no leakage on the sec - and where we clamp this turn off volt spike to a fixed value - the clamp time is exactly equal to the time taken for the current to rise in the sec. [ this is in fact one way to get an accurate measure of the sec leakage L ]
This can be confirmed in other ways too, for a multi wdg transformer, if you short one wdg and measure the leakage L on the "pri" you get a certain value - if you now short another "sec" you get a lower value, and so on... this is because all the sec leakage L's are "adding" in parallel to reduce the total leakage of the sec's as seen from the pri ( 1/Llk tot = 1/Llk1 + 1/Llk2 + ... ) - this could not possibly happen if all the leakage was physically on the pri side -> QED.
p.s. in spice the leakages are allocated according to the value worked out from the coupling factor L, and then divided and allocated to each wdg according to the turns ratio.
however for a flyback - this is not the case, the pri winding is charged with current and there is no current in the sec, then the energy has to be transferred to the sec winding by the current building up quickly through the sec wdg and load - the o/p cap in this case - if there were no sec leakage this would happen instantaneously and the flyback voltage would not spike up as badly as it does as the transformer tries to force the EMF high enough to make this happen - in other words we would not see such a large turn off spike on a flyback if there were no leakage on the sec - and where we clamp this turn off volt spike to a fixed value - the clamp time is exactly equal to the time taken for the current to rise in the sec. [ this is in fact one way to get an accurate measure of the sec leakage L ]
This can be confirmed in other ways too, for a multi wdg transformer, if you short one wdg and measure the leakage L on the "pri" you get a certain value - if you now short another "sec" you get a lower value, and so on... this is because all the sec leakage L's are "adding" in parallel to reduce the total leakage of the sec's as seen from the pri ( 1/Llk tot = 1/Llk1 + 1/Llk2 + ... ) - this could not possibly happen if all the leakage was physically on the pri side -> QED.
--- Updated ---
p.s. in spice the leakages are allocated according to the value worked out from the coupling factor L, and then divided and allocated to each wdg according to the turns ratio.
Last edited: