[Amps] Pi-L Network Question

[Jim Garland]

I've been noodling around various tank circuit possibilities for a 160m
monoband amplifier. A pi-L network isn't very practical for an 8877 because
the relatively high plate impedance mandates very large values of C1 and C2.
For an 8877, a simple pi-network is preferred. On the other hand, a Pi-L
makes a lot of sense for, e.g., three 3CX800a7s or GU-74Bs,  because the
combined plate impedance is under 1000 ohms and the required capacitance and
inductance values are quite reasonable. 

I know that good design practice is to orient L1 and L2 in a Pi-L at right
angles to each other (or to use a self-shielding toroid for L2), to minimize
mutual inductance between the two coils. However,  I don't really understand
why this is so important. In principle, it would be very convenient to use a
single coil whose inductance is L1+L2, with a tap for connecting C2 at the
junction point between L1 and L2. I'm sure that the mutual inductance would
invalidate this approach, but I don't understand why. Can somebody who's
thought about this issue clarify it for me. Thanks!
73,
Jim W8ZR