HDCD is a type of encoding and decoding found on some CDs and some CD players, DVD players and DACs. HDCD provides 20 bits of resolution most of the time. This results in greater clarity, particularly during quiet passages and for the tonal nuances of medium volume passages. The quality of HDCD recorders and compatible playback filters is generally among the best, so HDCD has found acceptance in both popular and audiophile recordings. Many mainstream recordings are now done with HDCD. HDCD will play on regular CD equipment, but it will be cleanly compressed to 16 bits. For best results and uncompressed dynamics, use HDCD comatible playback equipment with them.
DAC Update: Sonic Frontiers / The Parts Connection announced two new DACs in December 1998. The Assemblage DAC-2.5 and DAC-3 both feature surface mounted Burr-Brown PCM1704 96 kHz x 24-bit DACs and Crystal CS8414 96 x 24 input receivers. Like the DAC-1.5 and 2 they also use the excellent Pacific Microsonics PMD-100 HDCD decoder/filter, socketed so it can be swapped with an available Burr-Brown DF1704 96 x 24 filter. The 2.5 is $699 and 3 is $1499 both in kit form. The DAC-3 differs from the 2.5 in having twice the inputs, including both kinds of fiber, ST Glass and Toslink, a much more elaborate external power supply, and fully balanced operation from the DACs out. The 3 has two DACs per channel, a passive analog filter, balanced outputs, etc. One of the DAC-3 inputs uses the inherently low jitter I2S enhanced interface as described in the next section.
Tip: DAC stands for Digital-to-Analog Converter, the things that turn a usually PCM-encoded digital signal (ones and zeros) into an analog signal that can be amplified, sent to loudspeakers, etc. Depending on context, 'DAC' can refer to just the chip that actually performs the conversion or to the whole box that includes a power supply, chassis, pre- and post-conversion filters (digital and analog respectively), interfaces, etc.
Most CD/DVD/LD players contain these functions internally. Performing the conversion externally can result in better performance, especially if the issue of jitter is well addressed.
Like the DAC-2, the DAC-2.5 uses an Analog Devices AD844 for I-V conversion, but it uses a Burr-Brown (now part of Texas Instruments) OPA134 for as an output buffer versus the AD817 in the DAC-2. The DAC-3 uses a Linear Technology LT1357 for I-V and OPA134 and BUF634 for output and feedback. These opamps are as used in Mark Levinson DACs, which the 3.0's fully balanced topology generally resembles. Upgrade kits for both new models are available with more chassis damping material, resistor and capacitor upgrades, etc. The DAC-3 upgrade uses gonzo, expensive Burr-Brown OPA627s in the I-V and main op amp positions.
Interestingly the excellent PCM1702 48 kHz x 20-bit DACs used in the DAC-2 and new PCM1704 used in the DAC 2.5 and 3 have nearly identical noise, distortion, dynamic range, settling time and linearity specifications, so other than different input compatibilities, the performance of either should be about the same. In both cases, dynamic range of the standard U package is rated at 110 dBA at -60 dBFS or 18 1/3 bits. Signal-to-noise ratio of either model with all zeros input (the noise floor) is 120 dBA or 20 bits. Both are outstanding specifications and their performance would usually be swamped out by other system noise such as from power supplies in anything other than a very carefully executed design. Both the PCM1702 and 1704 are among the best commercial DACs available. But they're not 24-bit or 144 dB, which is probably impossible outside a laboratory (and extremely difficult there).
PCM1702 and 1704 are full word length, laser-trimmed ladder DACs that do not use noise-shaped, limited-bit conversion like Bitstream chips from Sony, Philips, Crystal and others. Each 1702 and 1704 internally employs dual complementary DACs which are balanced around the zero crossing. This novel arrangement means the smallest steps in either direction away from zero have minimal glitching and maximal monotonicity and linearity.
Because of the 96 x 24 capability the new Assemblage models promise to be somewhat future-proof, since compatible 96 x 24 filters like the pin-compatible Burr-Brown DF1704 already exist, with DSP-based 96 x 24 filters in the works. Sonic Frontiers offers a $50-70 option to substitute the DF1704 for the PMD-100 as a way to get a 96 x 24 filter in the Assemblage DAC-3. Current 96 x 24 interfaces are basically S/PDIF and AES/EBU with a doubled clock rate. Previously they would have maxed out at 48kHz sampling and 24-bits. One way to do the 192kHz sampling rate in the draft DVD-Audio specification is to use two AES feeds. But the more likely solution would be Firewire as mentioned in the next section.
Update: the Assemblage DAC-2.7 and DAC-3.1 and other Sonic Frontiers digital products now use the Pacific Microsonics PMD-200 filter which supports HDCD and 96 x 24. (Like the 48kHz PMD-100, the new PMD-200 also features Pacific Microsonics superior filtering strategies and algorithms.) Thew new Assemblage DACs also feature Peter Schut's dual PLL jitter reduction circuitry at their inputs. His dual PLL similarly tames jitter at the input of the Assemblage D2D-1 jitter reducer/upsampler/interpolator. TPC will make PMD-200/Dual PLL digital input boards available for upgrading the DAC-3 once production capacity for the 3.1 is met. Here is a comparison review of the sound quality of the different digital filters: PMD-100, PMD-200, and DF1704.
Here are my coments on the Assemblage D2D-1 upsampler/interpolator and DAC-3.0 combination, with I2Se as the interface between them.
Here's a comparison review of the Assemblage D2D-1 versus Perpetual Technologies P-1A upsampler/interpolators. DSP algorithms definitely do not all sound the same!
Here are my coments on the excellent Benchmark DAC1 PRE DAC and headphone amplifier.
A few things to be aware of when shopping for digital audio equipment. First, jitter reduction or reclocking boxes that use S/PDIF or AES/EBU to connect to the DAC are pretty pointless. Recalling that most of the jitter we're concerned about is introduced because of the fact that S/PDIF and AES combine clock and data on the same signal, it would seem to gain little to clean up the signal then send it back out over an inherently jittery interface. No matter how well the AES or S/PDIF interfaces are implemented, the data will always pollute the clock since they're interleaved in the same signal in these interface formats. That's how we got much of the interface jitter in the first place. As I've said, the interfaces with separate clock lines (like I2S, I2Se, SDIF, TDIF, etc.) are a much better design. Don't buy a jitter, clock, upsampling or digital processing box unless it (and your DAC) has a separately clocked interface such as I2Se.
Second, not all digital audio interface formats are created equal. Of the consumer ones, Toslink, is the worst for jitter and should be avoided. AES features an XLR connector which is totally inapropriate for high frequency digital signals. RCA connectors are also inferior in this application. Here's some more information about how to choose a consumer digital audio interface.
If you're interested in trying a 96/24 DAC at minimal cost, the MSB Link DAC is a good-sounding, low cost converter that works with 96x24 Audio DVD-Video discs, Laserdisc, DVD movies, CD, etc. At $349 the future-compatibility risk is minimized. Worst case, if new, incompatible standards come out the Link could be handed-down to a video or CD system, or given to a friend interested in getting into audio.