Level disparity in headphone amp reviews

[Zmix]

Here are THD and SINAD measurements from three recent reviews, each of these three headphone amplifiers was tested using a different signal level; the Topping L30 at 2V, the Topping A90 at 4V, and the THX AAA 789 at 6V.

This represents as much as 12dB difference in input signal level between the L30 test and the AAA 789 test.

In the Topping L30 review it states that the gain was set to Unity (2V in / 2V out), but it is unclear if the A90 was adjusted to produce 4V out and the AAA789 to produce 6V out.

Are the output levels of each DUT normalized to "0dBA" as shown in the FFT?

How would these measurements compare if fed from the same signal level?


TOPPING L30:
Headphone Amplifier Audio Measurements
As usual, I feed the amp a superbly high quality analog 1 kHz tone at 2 volts, set the volume/gain to output 2 volts ("unity gain") and see how much distortion and noise we get:

TOPPING A90:
Headphone Amplifier Audio Measurements
With a million headphone amplifiers out there, measurements is the only way to find the good from bad. So let's start with our usual dashboard of unity gain (same voltage output as input). Connection is XLR input and unbalanced 1/4 headphone out unless said otherwise:

DROP THX AAA 789:
Measurements
For consistency with my last couple of headphone amplifier reviews, I set the output at about 5.7 volts:

 

[amirm]

Here are THD and SINAD measurements from three recent reviews, each of these three headphone amplifiers was tested using a different signal level; the Topping L30 at 2V, the Topping A90 at 4V, and the THX AAA 789 at 6V.

Put aside the 6/5.6 volt ones as that was the initial set of tests before I standardized.

The testing now has two voltages:

1. For unbalanced input, 2 volts is fed, and 2 volts is output.

2. For balanced input, 4 volts is fed, and 4 volts is output.

In both cases, the device is in "unity gain" meaning it is not asked to amplify the signal. So is fair treatment of both classes of amplifiers.

Note that 2 volts in unbalanced is equal to 4 volts in balanced input in audio gear. The latter is differential so doubles the voltage. Internally the gear is designed to treat them the same (or should be).

 

[amirm]

Are the output levels of each DUT normalized to "0dBA" as shown in the FFT?

Yes. dBr stands for dB relative. I can set that value to anything and it becomes 0 dB on that graph. So for each type of device I test (unless I forget which is rare), I set the dBr to the current output so you can easily read the distortion levels relative to the peak.

The data is correct even if I don't set it but then you have to compensate for the main tone not being at 0 dB. Many reviewers do this making it hard to read and interpret their measurements.

 

[amirm]

In the Topping L30 review it states that the gain was set to Unity (2V in / 2V out), but it is unclear if the A90 was adjusted to produce 4V out and the AAA789 to produce 6V out.

Best way to get around this historical issue is to look at the power sweeps. There, it is apples to apples as you are looking and comparing the same output power.

 

[Zmix]

Thank you so much for clarifying, Amir. I'm a big fan of your tests.

Note that 2 volts in unbalanced is equal to 4 volts in balanced input in audio gear. The latter is differential so doubles the voltage. Internally the gear is designed to treat them the same (or should be).

The level required to produce 2V at a single ended source will produce 4V from a differential source (when measured across pins 2 and 3), so do you assume that the differential inputs attenuate the signal 6dB (resulting in an internal signal level that is equivalent regardless of input selected)?

Also, do you think the SINAD in these measurements would be different if the signal level was standardized?