If this transformer box is designed to be driven by a high power headphone amplifier, it's not trivial and it would be quite pricy.
Without the box the CA-1A costs $ 2k with the box = $ 2.5k
The box itself is sold for $ 650 as it can als be used for their other model but this then requires using 'baffle compensation' RCA/XLR cables in front of the amp.
One can also use this box with speaker amps b.t.w.
It can handle 6W @ 30Hz and 12W at 60Hz, 24W at 120Hz.
Above those power levels the transformer saturates (compression/higher order harmonics)
RAAL recommends amps to be used with a minimal rating of 2W@16 ohm or 2W@32 ohm
The massive low-freq distortions seem to be happening by the front/rear chambers leakage, the membrane probably has the gap. Hence, this is not really harmonics but a turbulent noise. I see no way to reduce the gap down to zero due to that motor topology, maybe a wider gap may reduce turbulence but the low-frequency response is the sacrifice in this case. IMO, ribbon headphones as an inversion of a ribbon mic topology is a completely dead idea in terms of high-fidelity sound reconstruction.
RAAL could (but did not) claim 0.04% THD @ 1kHz @ 94dB SL and would not be lying.
And indeed the gap on the left and right side of the ribbon could well be responsible.
A very flexible 'roll edge' could well prevent this but there would need to be some other intentional leakage that has to be introduced.
For tweeters (this is a spin off) these issues do no exist.
5% @ 100Hz is a bit on the high side.
This is a weak point of these drivers, for ortho's this happens to be a strong point.
I wonder how distortion would be with the open pads.
Alas we might not know unless
@amirm can still create those plots from measurements with the open pads.