Multiturn loop matching

I couldn't resist to see the LW loop hanging in the shack, unused. The quickest change I could apply was to add a transformer to match the loop high Z (circa 1800 ohm, simulated) to 50 ohm.

The transformer requires approximately a 6:1 turns ratio, for a 36:1 impedance step down.

From the bag of unknown toroidal cores I picked a small one, black, the size of most familiar T37-x. It was used as an RF choke in some computer equipment. With a thin wire I wound 24 turns on the primary (loop side), so that I would need 4 turns on the secondary (RX side). The DVM LC meter measured 900uH from those 24 turns: high, as expected! The material should be appropriate for the intended frequency range!

The result on the receiver is astonishing. The previously observed lower Q on the high end of the tuning range is fixed: the tuning is razor sharp on the whole range. The new range is 150 to 570 kHz. The transformer does not seem to be lossy in LW/MW as observed signals are equal if better than before.

I still have not understood where the figure-8 radiation pattern lies: orthogonal or on the loop plane? Sources say othogonal, but I observed it on the loop plane with aeronautical NDBs. I mean, if on the loop plane was built a yagi, I orient it for the best signal. Anyone?