- Written by John E. Johnson, Jr.
- Published on 20 July 2009
On the Bench
I have so many graphs to present, what I will do here is show the CD test followed by the same test using SACD. I utilized a 22 kHz bandwidth for distortion measurements, except where the X axis goes out to 50 kHz. I used Filter 1 for all tests, except for the frequency response curves where I show the results for all three filters. I also present the graphs for a 100 kHz load and 600 ohms load, both channels driven. The yellow graph lines are the left channel, and the red lines are the right channel.
At 1 kHz, recorded at 0 dB (the highest level of digital recording), the harmonic peaks were almost non-existent, both for CD and SACD, and for 100 kOhm vs. 600 ohms. In fact, if anything, there was a bit more distortion with SACD and its noise level was a little higher. For the SACD, there was a second and third order harmonic peak, while for the CD, it was just a third harmonic. This is significant only because even-ordered harmonics are euphonic - that is, they are pleasing to the ear - while odd-ordered harmonics are very irritating, especially higher order harmonics.
This is superb performance, with the noise being 130 dB below the signal which was recorded at 0 dB. There was only a 0.6 volt drop when going from the 100 kOhm load to the 600 ohm load, which indicates a top notch power supply. The Y axis reads in dBV, where 0 dBV = 1 volt. Voltages greater than 1 volt read as positive numbers on the Y axis, such as the 3.6 volts in the above graph reading as 11.1 dBV. Voltages less than 1 volt read as negative numbers, such as the -99 dBV harmonic peak at 2 kHz being 0.000001 volt, which represents 0.001% distortion. The other harmonic peaks in that graph, such as the one at 3 kHz, add to the value at 2 kHz, plus calculated noise values, giving a total THD+N of 0.006%.
Using a combination of 19 kHz and 20 kHz sine waves as the test signal yielded the following results:
In CD mode, there were more IM peaks surrounding the 19 kHz and 20 kHz input peaks than in SACD mode, although they were quite low in any case. There was no B-A peak at 1 kHz for either CD or SACD. However, note that in CD mode, there are some significant peaks at 24 kHz and 25 kHz, as well as some smaller peaks above 38 kHz. This is the first time I have seen such peaks with CD tests (although I have seen them reported in reviews in other publications). They are not multiples of the input signals, and are probably an aliasing issue. It indicates that there is no filtration above 22 kHz (the analog frequency limit of 44.1 kHz sampling). Although these peaks are all above the standard audible band of 20 Hz to 20 kHz, the human ear does not have a brick wall filter at 20 kHz. It is just an "average" limit based on thousands of tests of human hearing. A very young person could likely hear those peaks at 24 and 25 kHz, and it can affect the perception of sound in the 8 kHz area. In any case, with reference to SACD performance, there is essentially a complete lack of IM peaks surrounding the input signal peaks, and this is extremely impressive. Again, just a slight drop in voltage between the 100 kOhm load and the 600 ohm load.
Shown below is a graph of a different SACD player, in CD mode, using the 19 kHz and 20 kHz sine wave test. This is the sort of result I am used to seeing. Note the fall in the noise floor above 22 kHz, illustrating the active filtration. The results with the Marantz almost look like aliasing in reverse, where frequencies below the Nyquist Frequency (1/2 of the sampling frequency, which in the case of CDs is 1/2 of 44.1 kHz, or 22.05 kHz) are mirrored in frequencies above the NF.
Based on the anomalies I found with 20 kHz and 21 kHz signals in CD mode, I also ran the test using a combination of 5 kHz and 6 kHz sine waves (sine waves that are close together, like the 19 kHz and 20 kHz test). The results are seen below. Note that the out of band peaks are at a higher frequency (38 kHz, 39 kHz, and 49 kHz) than they were when I used 20 kHz and 21 kHz. So, it would seem, the lower the frequency of the input pairs, the farther out into the inaudible area the extra peaks appear.
Shown below are the results for the IMD tests. For both CD and SACD, the performance was excellent. There were no visible side peaks near the 7 kHz input frequency, so the resulting IMD value was very, very low in all cases. I often see harmonic peaks at 14 kHz and 21 kHz with this test, and there are none here, either with CD mode or SACD mode.
Here are the frequency response graphs. In the first graph, in CD mode, I only tested one channel, so that it is easier for you to see the differences that the three filters make on the response. For Filter 1, the response drops off at 18 kHz, with filter 2, it begins a sharp rolloff at 10 kHz, and with Filter 3, the rolloff begins to be steep at around 7 kHz. I asked Marantz about this, and they said that Ken Ishiwata designed these filters to emulate the response in some of his earlier players. When I was listening, and turned on Filter 3, I expected it to sound like an AM radio, but in fact, it sounded very nice, with enough high frequency material to be pleasing. Kind of laid back, and not in your face. It should be noted that most of the energy in music is between about 60 Hz and 6 kHz, so perhaps that is why the steep filter was not that noticeable.
For SACD, the response was down 0.5 dB at 20 kHz, and down 1 dB by 30 kHz. The filters made only a very slight difference in the response. With the 600 ohm load, there was about 0.25 dB rolloff at the low end, for both CD and SACD, beginning at 80 Hz