- Written by Rick Schmidt
- Published on 15 July 2013
The Design of the Parasound Halo CD 1 Player
The Parasound website has an extensive white paper on the design of the CD 1:
The CD 1 is a slot loader, the loading mechanism feeds a CD ROM drive instead of a conventional CD drive. The CD ROM drive is connected to a passively cooled Intel ITX computer running the Linux operating system and proprietary software from Holm Acoustics meant to improve the reading of CD disc data. The CD ROM drive runs at 4x the CD Data rate to get a head start on the data, allowing the computer re-read if errors are encountered. Re-reading can occur multiple times, with repositioning of the laser if needed, until there are two reads that match. The data re-reads are limited only by the size of the data storage buffer in the CD 1. If that buffer is about to run out the CD 1 will switch to interpolation mode. Even this process is optimized in that the CD 1 takes care to perform this operation on an isolated section of bad data and then switches back to its normal re-read methods as quickly as possible. According to the white paper the process 'almost always results in error free data'. I have long wondered just how many errors we might be getting off of discs. After all, a sophisticated Reed-Solomon Error Correcting Code is part of the Red Book standard. Furthermore it would be interesting to know what sort of error rate (not jitter, errors) is typical an S/PDif interface. I am still wondering. I had some email exchanges with Richard Schram and Thomas Holm about the CD 1 and how it works but they couldn't supply this data. To be sure it would take an exhaustive study of new and old discs to say anything definitive but it would be good to know what kind of numbers we're talking about and compare players in this regard.
The paper goes on to list an Important take on the benefit of upsampling: "The upsampling process minimizes the staircase step size of the output current from the audio DAC to achieve the smoothest possible DAC raw output without any related aliasing. This is particularly important because large current source changes that exceed the slew rate of the analog op amps." In the case of the CD 1 data is upsampled by 8x.
It has become almost customary to use multiple DAC chips in high end DACs these days but again the CD one has an interesting angle on the topic and does it differently. If you have dual DAC chips, how well are they matched? I would guess that if the chips came from the same production run then the differences – in both delay from input to output and in the analog sections – would be small. But how small is small? Parasound's discussion on the subject mentions a figure of a 10 nanosecond difference input to output for a given pair of DAC chips. A small number to be sure but if your clock to those two chips is more precise than that, and in the case of the CD 1 it is (10 picoseconds, or one thousand times more precise), then there is a clear cost to go along with any gain you might get from the dual DAC chips. Some of this cost can be measured in the production as some amount of hand tuning is possible to accommodate the DAC chip differences but of course such tuning is an imperfect process so there is a cost to the sound as well. To avoid all this the CD 1 uses a single AD1853 DAC in stereo mode. The balanced output signal is then generated from the op amps (National LME49990's) in a differential output mode.
A button on the CD 1's front panel gives you the option of listening to the analog outputs as driven by LME49990 opamps or via discrete transistor output stages. The discrete output stage is a modernized version of the discrete output stage used in the vintage Parasound D/AC - 2000. The transistors are in a Darlington configuration that operates in the feedback loops of the LME49990s so that the specs for THD and noise are as low as the opamps alone. This seemed to be borne out by Chris Heinonen's measurements which could detect little or no difference between the two modes.
Of course power supplies are of upmost importance in any hifi. Besides the expected separate supplies for digital and analog, the CD 1 has 12 separate point-of-load power supply voltage regulators spread across the digital and analog realms. This use of multiple regulators is something that, to my knowledge, only Naim used to do but is now being adopted by many of high end manufactures. We are all benefiting.
There's much more in the whitepaper so I'll just sum up and say that Parasound and Holm Acoustics have put a lot of engineering into this player. You can see from my photo here that there is heavy isolation between the various circuit functions, a detail which is also covered in the paper.
The output connectors are Neutrik XLR connectors for the balanced outputs and gold-plated Vampire RCA jacks for the unbalanced outputs.
The included remote is a small, mid-weight plastic affair. Not quite what we're used to for some similarly priced equipment but certainly adequate and not unpleasant to use. The textured bottom is rounded with a protrusion to fit the hand. The top portion of the remote is dedicated to the CD 1 while the bottom is for the JC 2 preamp. There is no facility for repeating a track or entire CD. Probably only an issue when you are breaking in the player or perhaps attached cables.