The counter-example is as follows: We have a rectangular toroid ferrite(ring ferrite), magnetized in a closed loop around the ring. We put capacitor plates on top and bottom surfaces, with suitable direction. Now the Poynting vector points inwards or outwards. We look at a cylindrical surface passing through the middle of the thickness of the ring. There is surface integral of the Poynting vector! But there is no free current, and electromagnetic energy is constant. Which seems to violate the Poynting theorem. Any exmplanation?