- Written by John E. Johnson, Jr.
- Published on 19 April 2010
On the Bench
THD+N measurements for CD mode were within an 80 kHz bandwidth, and for SACD mode, within 22 kHz (to eliminate the inclusion of the out-of-audio-band noise that is characteristic of SACD). The XLR outputs of the MCD500 were used for the measurements. Except where noted, yellow graph lines represent the left channel, and red is the right channel.
At 1 kHz, THD+N for CD was 0.004%, while for SACD, it was 0.014%. You can see that the noise floor in SACD mode is at - 130 dB, far lower than with the McIntosh MCD201 or the Marantz SA-7SI. The greatly improved noise floor is at the cost of small distortion peaks becoming visible.
At 10 kHz, distortion for CD was 0.003%, and 0.015% for SACD.
The results for 19 kHz, 20 kHz combined test frequencies are shown below. Although there is no visible B-A peak at 1 kHz for CD, it is visible for SACD at - 109 dBV. There are also visible side bands in the SACD spectrum (on either side of the 19 kHz and 20 kHz input peaks).
The IMD measurements were 0.005% for CD, and 0.08% for SACD.
The measured frequency response for CD was 20 Hz - 20 kHz, - 0.2 dB, and for SACD, it was 20 Hz - 30 kHz, - 0.45 dB.
The jitter spectrum (graph shown below) for CD mode indicated an average of about 60 ps (picoseconds or trillionths of a second). The yellow and blue lines are when a test signal or music was playing. This is quite excellent performance. The spectrum was gathered by connecting the digital coax output from the MCD500 to the digital coax input on the Audio Precision. You can see very slight differences when a test sine wave was playing vs. no music playing (yellow and magenta arrows). The no music playing spectrum is simply the clock which sends its signal out of the coax jack whether music is playing or not, and jitter is in the clock by itself (magenta graph), with 16 peaks along the 1.4 MHz spectrum. The baseline jitter (excluding the 16 high peaks) is about 5-10 ps.
You can also get an idea of what jitter is doing by looking at the analog spectrum (the analog output of the DAC) of a single input sine wave peak and its surrounding symmetrical side bands that are the same height and the same distance from the fundamental. However, this is not actually a jitter spectrum, but rather, the effects of jitter on the analog output (the music), and includes whatever correction the DAC has performed on the jittered signal as well as any distortion produced by the analog output stage. The best way to measure jitter is by looking at the digital signal before it is processed by the DAC.
Because the MCD500 has digital inputs and outputs, I ran a little experiment with my iPod, which has many of my favorite albums stored on it in uncompressed *.wav format, and a Wadia iTransport, which is an iPod dock that outputs the digital bitstream from the iPod connected to it. This forms a rather small, and very manageable, music server. I connected the coaxial digital output of the Wadia (which is an RCA jack) to the coaxial digital input on the McIntosh MCD500 (which is an RCA jack), and selected "Coax" as the input which is indicated on the MCD500 front panel.
So, here is the digital jitter spectrum from the Wadia coax digital output with the iPod. You can see that the jitter is all over the place, with at least 40 tall peaks. With music playing (I used La Peri), jitter ranges from 5 to 20 ps, which is surprisingly good at first glance. However, the presence of so many of the higher peaks in the iPod-Wadia transport jitter spectrum compared to the jitter spectrum of the McIntosh MCD500 may actually represent worse performance for the iPod-Wadia combination. I just don't really know yet. This remains to be determined as we accumulate more such graphs in future jitter analyses of digital bitstreams.
When I then connected the digital coax output from the Wadia to the digital coax input on the MCD500, and gathered analog spectra from the MCD500 analog outputs, there was more distortion than when the MCD500 was taking its digital bitstream from its own transport. But it was mainly in the IMD area where the increased distortion was seen. So, even though the jitter spectrum was lower with the iPod/Wadia combination, analog output IMD was higher. This probably is due to additional jitter produced at the coax connection jacks, and the fact that the jitter was not confined to a narrow band as it was with the McIntosh transport.
The bitstream is read into a memory buffer in the MCD500 for error correction, but it obviously cannot correct everything. The results illustrate that having a separate transport and an outboard high end DAC does not necessarily give you better results than a transport with DAC in one chassis. In fact, the results could be slightly worse, as shown here.
Jitter is a controversial topic. No one can agree on how much jitter produces an audible effect. But, it seems that it is dependent on the frequency in the audible band where it is occurring. Jitter represents the bits arriving at the DAC's input sooner or later than the DAC is expecting them, and this can make the DAC interpret a 1 as a 0 or a 0 as a 1. There is a digital clock reference, which generates a very steady signal by applying a voltage to a quartz crystal that oscillates at a defined frequency, in the MHz range, such as 24 MHz for a high end clock generator (in the world of Pro Audio, outboard clock generators are used to synchronize all the digital components used in the recording process). This is down-converted to a "word clock" which is at the same sampling rate as the music, such as 16 bit, 44.1 kHz, which is then synchronized to the flow of the audio digital bits in a precise fashion. The DAC, using the clock as a reference, expects to see the audio bits arriving at a certain time. If they arrive early or late, jitter occurs.
Jitter was pretty bad in early digital audio systems, but these days, the DACs are really very good, and jitter is low. The DAC can deal with jitter and other problems because in fact, for each 16 bit sample (in Redbook CD), there are 32 bits. The first three are the "Preambles", which tell the DAC what is coming. Bits 4-27 are the audio bits, with the first 16 being the audio and the last 8 being 0's. Bits 28 - 31 are administrative, which includes parity. So, if you really look at the 16 bit word, there is a lot of associated information to make sure that the DAC interprets that word correctly, and sends it out as the right voltage value. Of course, error correction is not perfect, so we still get the effects of jitter in the analog output.
As an aside, the Wadia iTransport has its own DAC, if you wish to use it. Here is a 10 kHz spectrum from its RCA analog output. THD+N is an order of magnitude higher than with the MCD500's DAC. So, if you plan on using an iPod and Wadia iTransport, connect it to a good outboard DAC rather than simply feeding your hifi system with the analog output from the Wadia's built-in DAC.