AudioChainsDB
OUTPUT SCENARIO

Truthear HEXA on a Focusrite Scarlett 2i2: do you need a separate DAC?

Yes — but not because the interface is too weak. It is 7.1 dB too strong. The reason is 50 ohms of output resistance against a 20.5 ohm in-ear: a ratio of 0.4 where 8 is wanted, the worst in the whole catalogue. A 25-euro dongle does what the 200-euro interface cannot.

Yes, buy the dongle — and then notice that the reason has nothing to do with power.

Truthear publishes the HEXA at 120 dB/V and 20.5 Ω. For a 110 dB peak it needs 0.316 V, 4.88 mW, 15.43 mA — one of the easiest loads in the catalogue. The Scarlett 2i2 4th Gen has 2.45 V on tap, seven times more than required.

Then the 50 Ω output resistance meets the 20.5 Ω load. The divider hands 71 % of the voltage to the interface's own output stage: only 0.712 V reaches the earphone, which is still 117.1 dB, 7.1 dB above target. The level was never in doubt. The impedance ratio is 20.5 / 50 = 0.4, against the 8 the rule asks for — the rule is missed by a factor of twenty, and this is the worst-matched pair in the catalogue.

Unlike a 0.3 dB wobble on a DT 770, a ratio of 0.4 on a multi-driver in-ear is not a footnote. The crossover of a hybrid earphone presents a swinging impedance across the band, and every ohm of source resistance turns that swing into a frequency response change. What you hear is a different tuning, not a quieter one.

SourceOutputLevelHeadroomRatio
Scarlett 2i2 4th Gen2.45 V, 50 Ω117.1 dB+7.1 dB0.4
Audient EVO 42.48 V, 50 Ω117.2 dB+7.2 dB0.4
FiiO KA13 dongle1.22 V, 1.7 Ω121.0 dB+11.0 dB12.1
FiiO K114.24 V, 1.2 Ω132.1 dB+22.1 dB17.1

The pearl: the interface loses more of its own voltage into this earphone than the dongle produces in total. The 2i2 drops 10.7 dB in its own output resistance — 2.45 V becomes 0.712 V — while the KA13, with 1.7 Ω, keeps 1.127 V of its 1.22 V and ends up louder than the interface that started with twice the voltage. Nothing about that is visible in either specification sheet, because output resistance is printed in a different row from output voltage, and neither is printed next to the load.

One limit we must state rather than calculate: with 22 dB of surplus on a K11, hiss becomes the operative question, and our schema has no field for output noise. We can give you the level and the ratio as numbers; on hiss we can only tell you where to look.

Loading the catalogue…