Я так понимаю, что с 85db усилением в XTC та же история, судя по публикации в SOS:
“
The OctaMic XTC's digitally controlled, low-noise mic preamps are based around the same Texas Instruments PGA2500 mic-preamp chip that RME use in their Fireface 400, Fireface UC, Fireface UFX and Micstasy, and can accommodate a range of input levels spanning an impressive-sounding 85dB, although the controllable preamp gain range is actually only 65dB. The gain-staging arrangement is unusual in that, although it starts at unity gain (0dB), the first step jumps straight to +10dB and then increments in 1dB steps all the way up to +65dB. At the maximum gain setting, a mic input signal of -63dBu will produce a practical digital recording level of -10dBFS. To put that in everyday terms, its equivalent to an 84dB SPL source in front of a Shure SM57 mic, or 66dB SPL in front of a Rode NT2. In other words, there's plenty of usable gain on hand here, and even at maximum gain, the power-supply mains hum components all measure below -105dBFS.
At minimum gain and with the relay-switched 20dB pad engaged, the mic input XLR would need +32dBu to make it clip, and that would require something like a Neumann M149 tube mic exposed to 150dB SPL — so it's not likely to happen in practice! The TRS line inputs can cope with an even more extreme +39dBu with the pad engaged, while the instrument inputs clip above +21dBu. Presumably because of the high-impedance buffer stage, the minimum gain for the instrument mode is +10dB and the overall gain range is reduced to 55dB. Nevertheless, the maximum gain setting will deliver -10dBFS from a -45dBu input signal, which should cater for most guitar pickups very comfortably.
As with so many modern preamp designs, the line inputs are padded down to increase the impedance and decrease the sensitivity, before being fed into the corresponding channel's mic preamp. In theory this approach risks unwanted distortion and noise, and would have been frowned upon by high-end electronics designers 20 years ago. However, RME's design introduces less than 1dB of extra noise and distortion, which is insignificant compared with the practical advantage of precise, repeatable and recallable gain control.
RME's flagship Fireface UFX interface employs 'Advanced Parallel Conversion' technology (combining two converters per channel) to maximise the signal-to-noise ratio, but advances in converter performance have overtaken that approach, and the AKM AK5388 chips used here (and in all of RME's products since the ADI-8 DS MkIII) delivers about 2dB less noise overall than APC. My own AES17 dynamic range measurements gave a figure of 116dB (A-weighted) and distortion below 0.0003% via the line input at +4dBu (with the mildest tendency towards odd-order harmonics). The new converters also feature lower latency than many, typically introducing less than 0.28ms at 44.1kHz sample rates and proportionately less for higher sample rates.
The mic input's EIN figure was a smidgen over 126dB at maximum gain, and 125dB at 30dB gain, and channel crosstalk was below -100dB at 10kHz, which is very impressive indeed. The input frequency response is flat to below 10Hz, reaching -1dB at 8Hz. Headphone outputs were capable of delivering +17dBu in the high setting, and about +4dBu (+2dBV) in low mode, and the D-A stage delivered an AES17 dynamic range figure of 117dB (A-weighted) with distortion of 0.0004% and a slight tendency towards second-harmonic dominance.”
