Sony HAP-Z1ES high-resolution file player Measurements

Sidebar 3: Measurements

I used my sample of the top-of-the-line Audio Precision SYS2722 system (see and the January 2008 "As We See It") to examine the Sony HAP-Z1ES's measured behavior. Initially I tried playing the WAV files of my test signals from a FAT32-formatted USB flash drive, but the Sony didn't recognize the drive's format. (It did recognize that a drive had been inserted and asked me if I wanted it formatted.) I then hooked up the Sony to my home network via Ethernet and transferred the WAV files from my Apple MacBook Pro using Sony's HAP Music Transfer app. This worked flawlessly, and for the measurements I played the files from the Sony's internal hard drive using the front-panel controls.

While the Sony will play DSD64 and DSD128 files, I don't yet have any test signals encoded in those formats; all my measurements were therefore performed with LPCM files. The Sony played files with both 16- and 24-bit word lengths and with sample rates ranging from 44.1 to 192kHz and all stops in between. Although I'd loaded a WAV file sampled at 384kHz onto the Sony's drive, the filename was grayed out on the Sony's display and couldn't be selected for playback. I didn't test specifically for the effect of Sony's Digital Sound Enhancement Engine, but the DSD Remastering Engine was engaged for the measurements.

Except where noted, all measurements were taken from the balanced output jacks. The maximum output level at 1kHz was the same from the balanced and single-ended outputs, at 2.07V left and 2.106V right, the latter 0.15dB higher than the former. Both sets of outputs inverted absolute polarity, the XLR jacks being wired with pin 3 hot, the opposite of the AES standard. The balanced output impedance was a fairly low 455 ohms at low and middle frequencies, rising inconsequentially to 513 ohms at the top of the audioband. The unbalanced output impedance was a low 239 ohms at 20Hz and 220 ohms at 1kHz, and an even lower 39 ohms at 20kHz.

The Sony's impulse response with data sampled at 44.1kHz was reproduced in inverted polarity (fig.1), and the time-symmetrical ringing to either side of the impulse indicates that it uses a linear-phase digital reconstruction filter. A wideband spectral analysis of the HAP-Z1ES's output while it decoded 44.1kHz data representing white noise at –4dBFS (fig.2, cyan and red traces) suggests that this filter has a null at half the sample rate, indicated by the green vertical line (footnote 1). The filter totally suppresses the image at 25kHz of a full-scale tone at 19.1kHz (fig.2, blue, magenta), and the second and third harmonics of this tone are very low in level, at –103 and –109dB, respectively. The Sony's internal transcoding of PCM data to DSD results in an ultrasonic noise floor that rises with frequency with both signals in this graph, but only to a moderate extent.


Fig.1 Sony HAP-Z1ES, impulse response (4ms time window).


Fig.2 Sony HAP-Z1ES, wideband spectrum of white noise at –4dBFS (left channel cyan, right red) and 19.1kHz tone at 0dBFS (left blue, right magenta), with data sampled at 44.1kHz (20dB/vertical div.).

The HAP-Z1ES's frequency responses with 44.1, 96, and 192kHz data are shown in fig.3. With the two lower sample rates, the output drops like a stone just below each Nyquist frequency (ie, half the sample rate), but the rolloff is gentler with 192kHz data, reaching –10dB at 92kHz. There is a very slight rise in response just below the rolloff with the two higher sample rates. This was with the balanced output; the rise was absent with the unbalanced output. Channel separation (not shown) was superb at 1kHz: 132dB in both directions, and still 114dB at 20kHz. The low-frequency noise floor was free from spuriae, other than a tiny amount of 120Hz in the left channel; at –139dB, this will be completely inaudible under all conditions!


Fig.3 Sony HAP-Z1ES, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), 192kHz (left blue, right red) (0.5dB/vertical div.).

For consistency with my measurements of digital components going back to 1989, my first test of a product's resolution is to play first dithered 16-bit data, then dithered 24-bit data, both representing a 1kHz tone at –90dBFS, while I sweep a 1/3-octave bandpass filter from 20kHz to 20Hz. The result with the Sony and 16-bit data is shown as the top pair of traces in fig.4; all that can be seen, other than the peak at 1kHz that peaks at precisely –90dB, is the spectrum of the dither noise used to encode the signal. With 24-bit data (middle traces), the noise floor drops to unmask the Sony's own analog noise, which is almost at the 20-bit level, and easily enough to resolve a 1kHz tone at –120dBFS (bottom traces). Repeating this analysis with a more modern FFT technique gives similarly good pictures (figs.5 & 6).


Fig.4 Sony HAP-Z1ES, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit data (top) and 24-bit data (middle), and at –120dBFS with 24-bit data (bottom) (right channel dashed).


Fig.5 Sony HAP-Z1ES, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red) (20dB/vertical div.).


Fig.6 Sony HAP-Z1ES, spectrum with noise and spuriae of dithered 1kHz tone at –120dBFS with 24-bit data (left channel blue, right red) (20dB/vertical div.).

With its high resolution, excellent linearity, and low noise, the HAP-Z1ES's reproduction of an undithered tone at exactly –90.31dBFS (fig.7) was essentially perfect: the waveform is symmetrical, the Gibbs Phenomenon "ringing" on the tops and bottoms of the waveform is clearly evident, and the three DC voltage levels described by the data are perfectly resolved. Though a touch (25µV) of DC offset is visible in the right channel (red trace), this is negligible. Increasing the bit depth to 24 gave a superbly defined sinewave (fig.8).


Fig.7 Sony HAP-Z1ES, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).


Fig.8 Sony HAP-Z1ES, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red).

As implied by fig.2, the Sony offered very low distortion. With a full-scale 50Hz tone (fig.9), the subjectively innocuous third harmonic was the highest in level, at just –117dB (0.00017%), with traces of second and fifth harmonic visible at even lower levels. This graph was taken into 100k ohms; with the demanding 600 ohm load, the third harmonic rose to –110dB (0.0003%), but there was otherwise no change in the spectrum. Tested for intermodulation with an equal mix of 19 and 20kHz tones, and with the combined waveform peaking at 0dBFS, there was very little distortion (fig.10), though the noise floor took on a rather granular appearance.


Fig.9 Sony HAP-Z1ES, balanced outputs, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).


Fig.10 Sony HAP-Z1ES, balanced outputs, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

The Miller-Dunn J-Test signal is not formally diagnostic for testing for jitter with digital systems like the HAP-Z1ES, where the clock is not embedded in the data. Nevertheless, I still perform tests with it on such systems because it sometimes reveals other problems. The result with 16-bit J-Test data is shown in fig.11. No jitter-related components are visible, and the low-level, odd-order harmonics of the low-frequency, LSB-level squarewave all lie very close to their correct levels (indicated by a green line). These harmonics disappear as expected with 24-bit J-Test data, leaving an almost perfectly clean spectrum (fig.12).


Fig.11 Sony HAP-Z1ES, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.


Fig.12 Sony HAP-Z1ES, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

This is impressive measured performance from a well-engineered, truly high-resolution digital component. I share KR's enthusiasm for Sony's HAP-Z1ES.—John Atkinson

Footnote 1: My thanks to Jürgen Reis of MBL for suggesting this way of plotting the behavior of a reconstruction filter.—John Atkinson
Sony Electronics Inc.
16530 Via Esprillo
San Diego, CA 92127
(858) 942-2400

skris88's picture

Okay. It's time you gave us a detailed article on how up-conversion works. In fact my brain tells me it CANNOT work. I'm a digital fan. I don't have vinyl. But I don't believe you can make something out of nothing, that a compressed MP3 file or similarly compressed Internet Radio stream could be up-converted to Hi Res digital audio. The ball's in YOUR court now, Stereophile!

John Atkinson's picture
skris88 wrote:
I don't believe you can make something out of nothing, that a compressed MP3 file or similarly compressed Internet Radio stream could be up-converted to Hi Res digital audio.

No-one has said that up-converting an MP3 file will produce the equivalent of a hi-res file. Nothing can put back the audio data that was discarded by a lossy codec. However, up-conversion to a faster sample rate is a legitimate tool in digital signal processing, and it is possible for an upsampled version of low-resolution data to sound better.

I first came across this technique in 1982, when looking at the chip set for the first Philips CD players. Each 44.1kHz sample was interspersed with three samples of zero value. The effect was to reduce the signal level by 12dB but also to quadruple the effective sample rate to 176.4kHz. No information was added above the original Nyquist Frequency (22.05kHz) but now a better-behaved digital reconstruction filter could be used. In addition, noiseshaping of the higher-sample rate digital data allowed Philips to wrest higher-resolution from the 14-bit DAC chips they had developed. (Before Sony got involved in development of the CD specification, the original CD proposal was for data to be stored as 14-bit words.)

John Atkinson
Editor, Stereophile

Dr.Kamiya's picture

Just think of anti-aliasing when you think of up-conversion. We do it all the time for fonts and graphics, so there's absolutely no reason it can't work for sound.

Anti-aliasing allows us to take rasterized fonts and blow them up to extreme sizes, while still retaining smooth edges and outlines.

This tech has been in development for decades in the world of visuals. Font-smoothing, subpixel-rendering, plus any number of different anti-aliasing methods. At the end of the day it's all about creating a smooth line from a jagged source, and it's just as applicable to audio as it is to graphics.

Just_Me's picture

I know this is an OLD post :-) But its a fascinating topic, as much because the terminology is so vague and misleading. Up-sampling? Oversampling? Difference? No and yes. Ignore.

Anyway, there are several techniques that can be applied if one increases the number of samples. Most fundamentally interpolation. Imagine that you had two dots, and good reason to believe they were connect by a straight line. You could plot dots along that line. Doing so in audio mostly make sit easier to build phase-correct and good sounding analog reconstruction filters. very common.

A higher frequency also suggests tighter jitter tolerances, which is important aurally. I wont go into the science, but it is.

Finally, there are algorithms that can essentially reverse-engineer (very imperfectly) even lossy compression algorithms, although their effectiveness is limited and they will many times guess wrong.

No one creates something out of nothing. It is more a method to improve filtering.


zounder1's picture

Sigh... when are manufacturers going to realize a lot of folks don't want a player with a built in hard drive? For those technically competent they long ago purchased a NAS to share files on their home network.
And why in the world do I have to go through the hassle of transferring my music collection and purchases from my computer to the Sony player? I can tolerate music players that are targeted at folks that just want to feed the machine disc after disc to digitize their collection without effort. But the Sony does not even do that?
So Sony has released a product that won't appeal to Luddites that want simplicity. (Asking my dad to transfer files to this machine would painful.) And it doesn't appeal to folks like me that already have digitized and organized my music collection on a home network.
This Sony player would be much, much nicer if it could simply play music from a NAS or network share. So sell me this player for less with no hard drive thanks. Besides, 1TB of built in storage is completely inadequate if you want to collect DSD files. Heck my very small music collection (FLAC mainly) tips the scales at 400GB! Any serious collection of lossless music (DSDm FLAC, WAV, ALAC, etc) is likely already way, way past 1TB in music.

SJNIETO's picture

zounder1, is this correct? It can not stream from a NAS or Net share?

zounder1's picture

Yes, according to the review you cannot stream from a NAS to this device. Miyou must copy the files.

To quote the article
"As when using a mobile player, one browses, buys, and downloads the music to a regular computer. Then, with Sony's HAP Music Transfer application, you can select the computer directories/folders that contain the files you want to send to the HAP-Z1ES. "

This means you use a utility to copy files to the sony. I find that quite pathetic.

SJNIETO's picture

This means one can copy the files over wireless but not play the files over wireless?
It's has WiFi but it can not play streams? Strange...

yuppi's picture

I just buy a HAP-Z1ES and found that the XLR jacks being wired with pin 2 hot...

mrvco's picture

This would be an interesting product if the USB port could be used to connect an external DAC. I'm sure the built-in DAC is great, but why limit its lifespan and versatility?

tmsorosk's picture

I do want a built in stand alone hard drive, who wouldn't ? Take it anywhere.
Link it to your computer for a few hours and all your music is on it, a one time deal.
Sweet .

ALTY2718's picture

I purchased the HAP-Z1ES last November (2014). My impressions to date are as follows. 1. The unit is very well constructed. 2. As I had ripped my CD's to my computers HD it was an easy, though slow process (using wifi) to transfer the files to the HAP-Z1ES. I ripped some 200 cd's as FLAC's, took up approx. 70GB which is no big deal. I also had some 10GB of higher res. downloads which I added. 3. The Sony sounded better with my B&W 805 Diamond speakers (on my 'B' system) than with my Focal 1027 Be speakers (on my 'A' system). On my main (A) system I am using an Oppo 105B blue ray player and a 125GB usb stick (as storage), which sounds better to my ears than the Sony in that combination of components. 4. The Sony unit is easy to use, I make use of both the Ipad app. and the remote/front of unit controls depending on whether my wifi system is on. The Ipad app. is fantastic, I have set up the 'favourites' option and make regular use of it. 5. One frustration I have had is that the copyright protection built into the Sony unit prevents one 'Sony music' cd I burnt from playing on the unit. This is frustrating as I have already paid for that music! If you have a lot of Sony cd's this my be something you need to consider. The Oppo played the same cd's without any problem. Despite this glitch I am really enjoying using the HAP-Z1ES.

scott.w's picture

I own this and love it. Question: should I swap the speaker cables (or ICs) to address the inversion?