Great design with measured performance that almost knocks the AHB2 off its throne.
My guess is that Topping is using a balanced input opamp or regular multiple opamps to achieve balanced input operation (probably OPA1642) and as driving the TPA32xx with custom optimized PFFB (optimized for extra low gain / higher feedback) as the power amp. Choice of higher Class D carrier oscillation frequency allows smaller inductors and capacitors to be used on the output filter and keeps ultrasonic noise leakage to a minimum in the audible band. The TPA32xx is capable of an extremely low noise floor, currently in the tens of microvolts with PFFB. Increasing level of negative feedback will bring the noise down even lower. TI’s factory designed
PFFB white paper specifies 36v power rail, 600kHz carrier frequency, 12.75dB of gain when using TPA3251 and this gives 28.3uV noise. To drop that to 16uV requires modifying the factory PFFB settings for even higher feedback of about 1.75x more. This would make TPA3251 intrinsic PFFB gain close to only 10dB. To get 19.1dB overall gain, Topping is adding about 9dB gain in the input stage opamp, not an unreasonable request. They will probably use dual rail +/-15v supplies for the opamp to provide the high swing needed to drive a 10dB power stage to clipping at max of 68Vpp output. We will see once the insides are revealed…
The factory distortion vs frequency curve has an oddly similar shape to the measured one so this is another clue that we have a TPA32xx core with PFFB, well implemented.