Parasound Halo CD 1 CD player Measurements

Sidebar 2: Measurements

At a time when CD players have morphed into all-purpose digital source/preamplifiers, it seemed anachronistic for San Francisco–based Parasound to introduce a CD player in 2013 that was nothing but a CD player (footnote 3) However, the Halo CD 1 ($4500) is more than it seems. Its CD-ROM drive spins CDs at four times the usual speed, controlled by software from Holm Acoustics, so that every bit is read at least twice before being sent to a large RAM buffer. Playback is thus from memory, not directly from the spinning disc.

John Marks reviewed the Halo CD 1 in June 2013: "It didn't take too much listening for me to conclude that, for playing regular CDs, the Parasound CD 1 was quite similar in essential sound to Bricasti's justly fàted M1 DAC [$8595]: detailed yet smooth, fast yet not fatiguing, and with wonderful frequency extension at the bottom and top," he enthused. "If you're prepared to keep playing your CDs and don't care about futureproofing, this is a fantastic player for reasonable money, given the high qualities of its engineering and parts."

John Marks's praise was echoed by Sam Tellig in July: "Slowly, over several evenings, I came to realize that the CD 1 was uncommonly able to distinguish good and great recordings from the poor and the mediocre." He concluded that "The CD 1 is for those who love their Compact Discs as much as Artie and Mikey love their LPs and turntables and Kal craves his surround sound."

If two of Stereophile's senior editors had been so impressed by the Halo CD 1, I wanted to find out for myself what it had to offer. I asked Parasound's Richard Schram for another review sample (serial no. 01049), which arrived at the end of August. Before doing any listening, I got the Halo CD 1 up on my test bench, using Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see and the January 2008 "As We See It") to assess its performance. For some tests, I also used my vintage Audio Precision System One Dual Domain. I did a complete set of tests using the CD 1's "op-amp" output stage, then repeated some of the tests using its "discrete" output stage.

The maximum output level was the same from both the "op-amp" and "discrete" outputs, at 3.94V balanced and 1.97V unbalanced. The latter is just 0.14dB lower than the CD standard's recommendation of 2V RMS. The output impedance was the same from both sets of outputs, at a low 102 ohms balanced and 51 ohms unbalanced. Both figures were the same at all audio frequencies, and both are half the specified impedances. Peculiarly, the CD 1 inverted signal polarity from both its balanced and unbalanced outputs when the Normal button was pressed on the remote control. Pressing Invert actually resulted in a non-inverted output signal. (The digital output's data were in the correct polarity.)

As it reads the CD data more than once before sending it into memory, I shouldn't have been surprised that the CD 1 was the best-performing player I have encountered when tested for error correction/concealment with the Pierre Verany Test CD. It played all the tracks on this CD without muting or audible glitches, even the ones that have combinations of gaps in the data spiral with minimum track pitch. RME's DIGICheck program confirmed that there weren't even any errors flagged in the CD 1's digital output. Color me impressed! This is the only CD player I've tested ever to have done this.

The CD 1's impulse response (not shown) indicated that it uses a conventional linear-phase reconstruction filter. The red trace in fig.1, a wideband spectral analysis of the player's output while it played data representing white noise at –4dBFS, reveals that this filter has the usual very fast rolloff above 21kHz. It suppresses the ultrasonic image at 25kHz of a full-scale tone at 19.1kHz (blue trace) by almost 110dB. The harmonic distortion associated with this tone is also very low in level, with the second harmonic at 38.2kHz lying at –94dB (0.002%).


Fig.1 Parasound Halo CD 1, 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 (10dB/vertical div.).

Channel separation (not shown) was superb, at >120dB between 400Hz and 3kHz, and still 115dB at 20kHz. Fig.2 shows the CD 1's frequency response with normal data (green and gray traces) and pre-emphasized data (blue and red). The former response is perfectly flat, but the latter rises above 50Hz, reaching almost 10dB at the top of the audioband. This indicates that the CD 1 does not recognize the pre-emphasis flag in the CD data. Those very rare CDs that are pre-emphasized will sound thin and bright.


Fig.2 Parasound Halo CD 1, frequency response at –12dBFS into 100k ohms with normal data (left channel green, right gray) and with pre-emphasized data (left blue, right red) (1dB/vertical div.).

When Parasound's Richard Schram received the preprint of the review, so that he could prepare a "Manufacturers' Comments" letter, he let me know that something must have gone wrong: in their testing of the Halo CD 1, it did handle pre-emphasized CDs correctly. I couldn't redo my measurement in order to check this matter, as the review sample, had already been returned to Parasound. However, it was sent back to me in time for me to remeasure it.

Using the top-of-the-line Audio Precision SYS2722, I first measured Halo CD 1's frequency response with pre-emphasized data burned on a CD-R, exactly as I'd done before. Again, the output climbed at high frequencies (fig.3, cyan and green traces). I then played the CD-R on a Pioneer SACD player and my Ayre C-5xeMP universal player, both of which recognized the emphasis flag on the CD-R and applied the correct de-emphasis.


Fig.3 Parasound Halo CD 1, frequency response at –12dBFS into 100k ohms with pre-emphasized data on CD-R (left channel cyan, right green) and with pre-emphasized data on CD (left blue, right red) (1dB/vertical div.).

I then remeasured the Halo CD 1, this time with the pre-emphasized track on the original CBS Test CD 1. The result is shown as the blue and red traces in fig.3. Now the correct de-emphasis was being applied. Then I played the de-emphasis test tracks on the Pierre Verany Digital Test CD. Again the Halo CD 1 applied the correct de-emphasis.

My original measurement was therefore incorrect, but I am at a loss as to why the Parasound didn't recognize the pre-emphasis flag on the CD-R, whereas the Pioneer and Ayre players did. (It might be relevant that the Ayre and the Pioneer are based on the same transport mechanism.) When I looked at the Parasound and Pioneer players' digital outputs with RME's DIGICheck software as each played the CD-R, neither had the emphasis flag set. However, the Pioneer's digital output had the de-emphasis applied to the audio data, whereas the Parasound's did not. All I can think is that there is some incompatibility between the Halo CD 1 and pre-emphasized CD-Rs. But as the player will almost never encounter such a disc anywhere other than in my test lab, this is not anything to be concerned about.

A problem in measuring a CD player that does not have a digital input that can handle data with a bit depth greater than 16 is that it becomes impossible to test the player's intrinsic performance. This is shown by figs. 4 and 5, which are, respectively, a 1/3-octave spectral analysis and a narrowband FFT spectral analysis of the player's output as it decoded data representing a dithered 1kHz tone at –90dBFS. In both graphs, the noise floor above 500Hz or so is actually the spectrum of the dither used to encode the signal. The CD's own noise is below that floor, except at low frequencies. The higher resolution of the FFT technique uncovers a low level of second-harmonic distortion, and a couple of power-supply–related spuriae at, again, a very low –120dB.


Fig.4 Parasound Halo CD 1, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit CD data (right channel dashed).


Fig.5 Parasound Halo CD 1, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit CD data (left channel blue, right red) (20dB/vertical div.).

These spuriae and the rise in low-frequency noise can also be seen in the spectrum of the CD 1's output while it played a full-scale tone at 1kHz (fig.6), which was taken with the "discrete" output stage. Figs. 4 and 6 were taken with the "op-amp" output; there were no measurable differences to be found between the two output types. One peculiarity is the appearance of low-level spuriae with a 100Hz spacing in fig.6. I have no idea where these might arise from.


Fig.6 Parasound Halo CD 1, spectrum of 1kHz sinewave, DC–1kHz, at 0dBFS into 100k ohms, "discrete" output stage (left channel blue, right red; linear frequency scale).

Linearity error (not shown) was less than the amplitude of the dither noise used to encode the signal. The CD 1's reproduction of an undithered tone at exactly –90.31dBFS was almost perfect (fig.7), with excellent waveform symmetry, and the three DC voltage levels described by the data well differentiated.


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

The spectra in fig.1 indicated that the Halo CD 1 offered very low levels of distortion. This was confirmed by fig.8. Even into the punishing 600 ohm load, the distortion spuriae all lie below –94dB (0.002%), with the third harmonic the highest in level. This graph was taken using the "op-amp" output stage; again, the "discrete" output gave the same result. This is not surprising, considering that the discrete devices are wrapped within the output op-amp's negative-feedback loop, and the performance was the same when I tested the CD 1's high-frequency intermodulation (fig.9). Actual intermodulation distortion was low in level, but the audioband noise floor looked rather hashy, presumably due to the 16-bit encoding.


Fig.8 Parasound Halo CD 1, "op-amp" output stage, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 600 ohms (left channel blue, right red; linear frequency scale).


Fig.9 Parasound Halo CD 1, "op-amp" output stage, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

When I tested for the CD 1's rejection of word-clock jitter (fig.10), the central spike that represents the 11.025kHz tone was well defined, with no spectral spreading of its base, and there were no supply-related sidebands. However, while most of the odd-order harmonics of the low-frequency, LSB-level squarewave were at the residual levels, those at ±229.5Hz were slightly accentuated, and those at ±457.6Hz were slightly suppressed.


Fig.10 Parasound Halo CD 1, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit CD data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

I auditioned the Halo CD 1 using the system mentioned in my review of the Electrocompaniet ECD 2 D/A processor elsewhere in this issue. For my primary reference, I used the NAD M51 D/A processor ($2000) that Jon Iverson reviewed in July 2012, driving the NAD with the Halo CD 1's coaxial S/PDIF digital output. (Only 16 bits were active in the player's digital output, confirming that it doesn't add dither or pad the data with zeros to give a 24 bit-depth.) Levels for the comparisons, as always, were matched to within 0.1dB at 1kHz, using the test tone on my Editor's Choice CD (Stereophile STPH016-2), and I made sure that both the CD 1's and M51's outputs were in positive polarity.

The niceties taken care of, I put on Kate Ceberano and Mark Isham's 2009 album, Bittersweet (M|M 10270 01670-6) and selected their treatment of "Skylark," a track I have grown very fond of. I listened first through the NAD; Ceberano's smoky tones emerged from the close-miked piano intro, her image hanging palpably between the speakers. I switched to the Parasound. Nothing changed. I confirmed that I had switched. I had. The two products sounded virtually identical. Many, many switches later, I felt that when Isham solos, there was the slightest smidgen of additional reverb surrounding his flugelhorn through the NAD. Maybe. I put on Percy Grainger's Handel in the Strand, from a 1985 collection of the Australian composer's orchestral works (CD, Chandos CHAN8377). The delightfully "clattery" piano that chugs along behind the strings in this reverberant 1979 recording by the Bournemouth Sinfonietta, conducted by Kenneth Montgomery, was presented perhaps slightly more forward by the Parasound, the soundstage being a little more delicately layered via the NAD. Perhaps.

A thought struck me. These comparisons were all being performed driving the NAD from the Parasound's digital output. John Marks had enthused about the Halo CD 1's performance as a transport, but could it be that if the Parasound had a sonic signature, it was affecting the digital output, forcing the NAD to clone that signature?

I hooked up the NAD via USB to my Mac mini running Pure Music and synchronized playback of an Apple Lossless file of Kate Ceberano's recording of "Skylark" with the same track playing on CD on the Parasound. Again, perhaps there was a touch more ambience with the M51, and now perhaps a touch more midbass with the Halo CD 1, but the sounds were otherwise virtually identical. I repeated the comparison using another reading of "Skylark," from the Tierney Sutton Band's superbly recorded Desire (CD, Telarc CD-83685). There was no change in my conclusion: If the NAD M51 is a Class A digital processor in this magazine's "Recommended Components" when it decodes CD-derived files, then the Parasound Halo CD 1 is a Class A CD player. John and Sam had called it correctly.

As Sam concluded, the Parasound is "for those who want to give their CDs proper respect, and who don't care what the 'industry' or editors want us to think." In this case, Sam, the editor agrees with you.—John Atkinson

Parasound Products Inc.
2250 McKinnon Avenue
San Francisco, CA 94124
(415) 397-7100
Share | |

Enter your username.
Enter the password that accompanies your username.