This post for is the data junkie who likes objective comparisons—in this case using quantitative measurements to indicate which DAC source input offers the promise of best sound quality. As it's been mentioned by myself and others, in-band audio analysis (as from an APx555) does not measure any differences, so deciding which input sounds best is a mostly subjective combination of listening tests. To help provide some objectivity, I measure the raw out-of-band (OOB) energy present at the DAC output and use this as a corollary for sound quality. This OOB energy has transited the DAC and possibly had an impact on the D/A process. So the input which measures the lowest energy has the least likelihood of having affected the DAC's sound.
The tests below used an Intel NUC for the source and Hugo2 for the DAC. The measurement device was a Signal Hound BB60c. I measured OOB energy from 10Mhz to 4Ghz with a 250 sample average. For measurement consistency, the Signal Hound was RF isolated. You can click on the links to bring up the measured plot for the input.
|
Input |
OOB Energy (dBm) |
| USB Direct (from NUC) | 3,654.3 |
| Toslink Direct (from NUC) | 3,313.7 (lowest) |
| OPTO-USB | 3,969.4 |
| Single Coax (from SRC-DX) | 23,222.0 |
| Dual Coax (from SRC-DX) | 47,114.6 |
| Single Coax (from XMOS Converter) | 23,313.0 |
| Toslink (from XMOS Converter) | 4,129.5 |
| MScaler (Single Coax) | 40,204.1 |
| MScaler (Dual Coax) | 67.086.6 (highest) |
| MScaler (Toslink) | 6,115.5 |
As expected, optical inputs, with no galvanic link, are RF isolated and inject the lowest energy. The highest energy comes from processor intensive components - and the reason why OPTO-DX improves things as much as it does. In general I am pleasantly surprised at how good USB direct is compared to Coax.
In Part II i will redo the test with inputs that can be optically isolated and hence combined with RF-STOP isolation.