Friday, November 14, 2014

Measurements and Opinions: Differences in Transport Jitter.

Measurements and Opinion: Balanced vs Unbalanced in consumer audio applications

Sometimes I wonder why we are stuck in the RCA/single ended era. I feel all amps should have transformer coupled inputs or something if we still want to have single ended output at the transducer. Its just better! No need to deal with the lack of noise rejection in wiring. Whoever said," Oh single ended is okay in short run high end applications." was either stupid or stupid.

Here we have a shielded RCA cable that I got from the local AV store, I purposely ran the cable across a DAC with R-core transformers and loopback for the results. This may be induced but the situation is true for many an audiophool running unbalanced lines with equipment racks, the induced EMI can be a nightmare. Perhaps there is a case for audiophool ultra-shielded RCA after all? The cable's length is 2m.


 Blue is for induced noise with a DAC's transformer, Red is for the same wire adjusted to induct the least noise.
 For the same loopback though with a balanced line running pass the transformer, we can see the benefits for balanced drive which is, no spurious tones.

XLR wire run pass a R-core transformer in a DAC's case, as you see, no induced noise as all of it is rejected






















Wednesday, November 12, 2014

ODAC Measurements

These measurements were made with an EMU0404, The tests are indicative of the ODAC's performance on a good day. The ODAC as suspected test really well, these are single channel test with the right channel of the ODAC testing worse than the left(0.003% vs 0.00068%) As expected the performance is resonable however due to the nature of Sabre based devices, we get noise floor with plenty of spuriae.

Fig 1.
Low level resolution of a 1khz tone @-90.31dbFS, as expected, not free of spuriae. 


Fig 2.

Fig 3 shows the Dunn Jitter Test of the ODAC on a good day/time where the ODAC sounds clean and reasonably transparent. In the worst case scenario, this jitter sidebands. rises to -115dbFS with a lot more sidebands. While arguably inaudible, to me the changes were so significant you practically have to be deaf not to hear it. A usb reset by plugging in and out does the trick but a device like the Wyrd helps the clock maintain good lock on the data stream.

Fig 3.

Adding a usb decrapfier or linear power supply helps with increasing the ODAC's performance as shown in Fig 3a.

Last of all, a test showing the reconstruction filter of the ODAC. This test is made by playing a full scale 19.1khz tone and white noise in 44.1khz sample rate to test the ODAC's reconstruction/anti-aliasing filter. This is then captured by the EMU0404's ADC at 192khz to observe the ultrasonic behavior of the device under test. An aliasing product at 25khz will be observed due to the mathematical implications of using a 19.1khz tone which nears the 22.05khz border of the bandwidth of the sampling frequency at 44.1khz. 

44.1khz-19.1khz=25khz

The odac seems to use quite a strong linear phase filter going by the sharp rolloff of 44.1khz white noise. Suppression of the aliasing product at 25khz is very strong at ~ -118dbFS However, the ultrasonic harmonics of the 19.1khz go on outside the audio band and at pretty significant levels.

Fig 4.
 Subjectively, I thought the ODAC is an okay DAC at most given its current price point. There are more consistent sounding DACs from the same price point to be had like the UD120 from Stoner Acoustics. The Geek products would be a better choice for a Sabre sound these days given what trusted ears say about them. A problem I think is that the Sabre tonality gets to me over a long period of time and while I like the clean transients, I found it too tonally lean and fatiguing over time which is probably cause by the measured spuraie. The EMU0404 I have tops out as the better sounding interface.


Schiit Wyrd, does it work?

This is meant to complement the Wyrd thread on Changstar. http://www.changstar.com/index.php/topic,1652.0.html. In my test with the Objective DAC and a few Tenor 7022 usb chip based DACs, I can acertain with measurements and subjective hearing that the Wyrd does work.


With the ODAC on a good day(there are bad days with the ODAC) the Wyrd brings to the table a better sense of speed and delineated transients meaning that a note's decay is shorter and the attack comes in quicker subjectively. 

In measurements, the biggest differences can be seen in jitter test. With an Audio GD with a Tenor 7022 isochronous Adaptive interface, the harmonics of the signal changes a lot with the Wyrd in the chain. I credit this mostly due to the change in jitter performance by adding the Wyrd into the test chain. Some sidebands disappear however there are some new spuraia, but overall the change is positive.




Somebody has asked me whether the Wyrd would work with Asychronous devices, the answer is a firm yes. Reduction in spurious artifacts.