Note the frequency axis in Mr. Atkinson's graph. The square wave is simply cut off.I am puzzled by the jitter test:
If this is J-test, why is the spectrum of the square wave missing? Here is a comparison to a $300 DAC (with 48 kHz sampling):
You can see the square wave being resolved on the left. But I don't see that in your measurement.
We are used to much better performance around here.I am puzzled by the jitter test:
If this is J-test, why is the spectrum of the square wave missing? Here is a comparison to a $300 DAC (with 48 kHz sampling):
You can see the square wave being resolved on the left. But I don't see that in your measurement. Ignoring that bit, pun intended, the rest of the response is just as clean as the Weiss (note that I go down to -160 dB vs your -150 dB).
John,It is not appropriate to take just one measurement, the harmonic distortion spectrum, to contradict my conclusion that the Weiss DAC204's "The measured performance of the Weiss 204 is state of the digital art." If you look at the complete set of measurements at the link below, you will note that as well as both harmonic and intermodulation distortion still being extremely low in level, the DAC204 offers vanishingly low linearity error down to –120dBFS with 24-bit data, an extremely low noisefloor with a measured resolution of 21 bits, and complete rejection of data-related jitter. Taken together, this led to my conclusion.
Weiss DAC204 D/A processor Measurements | Stereophile.com
Sidebar 3: Measurements I performed a full set of measurements on the Weiss DAC204 using my Audio Precision SYS2722, repeating some of the testing with the magazine's higher-resolution APx555 analyzer. I used optical and coaxial S/PDIF data—the TosLink input accepted data sampled at rates up to...www.stereophile.com
John Atkinson
Technical Editor, Stereophile
This is the waveform at the D/A processor's analog output, ie, after the reconstruction filter. (You can see the ringing of the filter on the waveform's leading edges.) The 16-bit data aren't dithered and at this specific level, -90.31dBFS, the data points consists of -1 least significant bit (LSB), 0, and +1LSB. In the twos-complement encoding used by 16-bit digital audio, –1 LSB is represented by 1111 1111 1111 1111, digital zero by 0000 0000 0000 0000, and +1 LSB by 0000 0000 0000 0001. If the waveform is symmetrical, this indicates that changing all 16 bits in the digital word gives exactly the same change in the analog output level as changing just the LSB.
When you look at the time axis, the test signal has a period of 1 ms, and it was not supposed to be a sine wave type signal. Below are a rough estimation of the samples (assuming fs = 44.1 kHz), and the reconstruction using a linear phase reconstruction filter. The measured output was actually reasonably close to the textbook case.
