I sat down and did some testing. I hooked up a 100 watt amp into a
10" guitar speaker, and used a specific spot on the grill cloth as my
reference point. I measured both mic output (from a mixer channel direct
out) and speaker drive signal, both in dBV. Distortion by visual
observation on oscilloscope.
I tried all three setups: unmodded, half-modded (my version), and the fully
modded, impedance balanced version posted by ljudatervinning. I did two
sets of tests, one with 2.2k load resistors and one with 10k load resistors.
Panasonic WM60-AY capsule (same capsule for all tests).
2.2k load resistors:
unmodded mic output -10, speaker at 2.3, mic almost distorted
half-modded: mic output -3.5, speaker at 10.5, mic almost distorted
fully-modded: mic output -6, speaker at 10.5, mic almost distorted
10k load resistors:
unmodded mic output +5.1, speaker at 8, mic almost distorted
half-modded: mic output 0, speaker at 12, amp at limit
fully-modded: mic output -4.3, speaker at 12, amp at limit
10k load resistors, capsule almost touching speaker dust cap:
half-modded: mic output +4, no visible distortion, amp at limit
fully-modded: mic output -2.5, no visible distortion, amp at limit
Results? I think I'm right. Wish I could get it loud enough to see the two
different mods into mic distortion, though. It also looks like the classic
PZM phantom power circuit would better match the Panasonic capsules if
implemented with a higher load resistance. It's worth pointing out that the
unmodded half-modded versions have twice the load resistance of the
fully-modded version. Hmm...