Misleading Measurements

[Miska]

2 hours ago, pkane2001 said:

As discussed, it is not THD but interchannel differences that determine soundstage “quality”. These are caused by differing levels and amounts/types of distortion between channels.

 

Overall phase behavior, and also to some extent modulator behavior. Not just inter-channel differences, but differences that apply to both channels. And for inter-channel also things over channel cross-talk.

 

[Miska]

On 4/21/2021 at 3:30 PM, pkane2001 said:

Most DACs I've measured, for example, measure better at 2v output than at 0.316, the same with head-amps simply due to better SNR.

 

Most optimal level vs THD+N figures for DACs seem to be around -10 dBFS. THD drops while noise is not yet starting to dominate.

 

In addition, many clip inter-sample overs.

 

[Miska]

On 5/10/2021 at 4:19 AM, March Audio said:

The audio data is usually normalised to come close to 0dB to maximise signal to noise ratio.

 

Which is one big source of problems (inter-sample overs), since this is typically done at the sampling rate. When such data is oversampled and/or converted to analog correctly (not case with all oversampling DAC chips) such will commonly exceed 0 dBFS of the source sampling. Because peak sample values at low sample rate (signal is close to Nyquist) rarely coincide actual peaks of the signal...

 

This difference can be up to +3 dBFS without source signal being clipped.

 

 

DSD is not bound to 0 dB scale, but instead the specification allows short term peaks of +3.15 dB. How this is handles varies by DAC chip. For example ESS Sabre scales 0 dBFS PCM = 0 dB DSD meaning that with DSD sources output level could be higher than with PCM sources. AKM chips is DSD Direct mode have output level of 0 dB DSD = -3.5 dBFS PCM. With TI chips the exact DSD output level depends on the selected analog filter.