Last post from me on this thread, I promise. And thanks to
@Rja4000 who indirectly helped, by providing me with key information on the same, which triggered the below.
I wanted to come back on Dynamic Range, and how to compute it, with some illustrations too as this is interesting, I think. For the purpose, I used a
Topping D50III, which Amir reviewed. This device was measured to have a 126dB DR. My own ADC does not go that deep, as it's not a High End Audio Precision, but the below results are still relevant.
The Dynamic Range is measured, per the AES standard, this way:
- Dynamic range: This test measures the ratio of the full-scale level at the output of the EUT to the weighted noise and distortion level in the presence of a low-level signal. It includes all harmonic, inharmonic, and noise components. The test signal shall be a 997 Hz sine wave with a level of -60 dB relative to the maximum input level (see 6.2.1). The output of the EUT shall be filtered with the standard low-pass filter (see 5.2.5) and the standard notch filter (see 5.2.8), whose center frequency is set to 997 Hz. The output of the standard notch filter shall be filtered with the standard weighting filter (see 5.2.7). The rms level of the final filter output shall be measured. The dynamic range shall be the ratio of the maximum output level (see 6.2.6) to this measured level. It is reported as dB CCIR-RMS.
The above is how Amir measures the DR, with his Audio Precision. From the above, it means the DR is computed from the SNR in the presence of low level (-60dBr) signal. And so the DR = Measured SNR + 60dB. The below measurements I performed respect the above, 997Hz frequency @-60dB relative to max, notch and low-pass filter, only the weighting filter (or no filter) is changed for educational purpose.
So, let's measure the Topping, first unweighted, showing a DR of 129.7dB:
View attachment 478501
Next with the obsolete A-weighting, showing a DR of 132dB::
View attachment 478503
You see the correction A-weigthed curve applied to the signal. It lowers low and high frequencies.
And last, with the CCIR-2k curve, which should be used, and that gives me 122.4dB:
View attachment 478506
Note the emphasis put at 6kHz by the CCIR curve.
For the record, Amir found 126dB of Dynamic range for the Topping (balanced output), so my interface is shy of 4dB compared to the AP, which is reasonable at this very low level.
As a matter of facts, my Cosmos Grade0 is given for 129dB(A) and 132dB(A) in mono mode. So, seeing it reports 132dB(A) in this test not only means I get there indeed, but confirmation that the Topping goes below that, as Amir showed.
Conclusion
Vendors like the A-Weighting curve for the obvious reason it provides them with the best results. But the new standard is the CCIR curve, as the AES mandates.
PS: I included the two CAL files for those using REW, for the CCIR and A-weighted curves. I created them with my AI best friend and checked it did no go rogue (it did for the A-weighted curve, too many prompts were necessary to make it right, it would have been faster to do it myself
). They are limited to 20kHz, FYI.
I know quite a few non degreed individuals who have excellent knowledge in EE principles/concepts and also enough, unfortunately degreed individuals who somehow even managed to graduate with 1st class honors, but not really have good enough understanding of some important concepts/EE principles, especially in the field of power systems, control systems, and communications (telecommunication that is). So again, I thought it would be fair to make allowance for exceptional cases, but in general I do agree with you.
