Product Review -
Music Labs DP-102 MKII DAC - September, 1996
By John E. Johnson, Jr.
|Audio Labs of Melbourne, Private Box 67, Abbotsford 3067|
|Phone 61-3-9645-8455; Fax 61-3- 9645-8412.|
|E-mail [email protected]|
|Music Labs USA, Inc., P.O. Box 148|
|Denver, New York 12421|
|Phone 607-326-7689; Fax 607-326-3436.|
|E-mail [email protected]|
Music Labs DP-102 MKII DAC; Digital-to-Analog Converter; Sampling frequencies, 32 kHz, 44.1 kHz, 48 kHz; Frequency response DC - 22 kHz plus or minus 0.5 dB; D/A conversion 20 bit; Digital filter 20 bit, 8X oversampling; Digital inputs AES/EBU balanced 110 Ohms, coaxial 75 Ohms, Toslink optical; Analog outputs 2 AES/EBU balanced, 2 RCA coaxial, 2 Vrms; Output impedance 47 Ohms; Size 3"H x 17 1/4" x 10"D; Weight 10 pounds; $1,999.
Some time ago, I took a $98 portable CD player to a high end audio store and asked the salesperson to connect the earphone jack output of the player to their reference system so I could compare the sound with their $10,000 CD transport and Digital-to-Analog Converter. He was shocked to find that they were very similar. The treble was there, the mid-range, the bass, all there. There was no particular harshness with my portable. It sounded great, as a matter of fact, on their very expensive preamp, power amps, and speakers. Something was different, we both agreed, but what? And was the difference worth 100 times greater price?
A decade ago, when CD players were introduced, we just had the players. Everything was in one chassis. You plugged the player into the wall AC, and the RCA jack outputs into the hifi system. That was it. Now, we can separate the "player" into components, and you don't necessarily have to have the same brand of components throughout, in order for the CD system to work properly. Not only can we have a separate transport and DAC, but stuff in between too, such as jitter reduction devices and components that will add bits to the word length. Perhaps the DAC is the least fancy-looking of all the components, because it has just one job: to convert the digital data stream to analog audio. There are no adjustments on the DAC, and many of them are just plain black boxes with nothing more than a small LED to indicate it is on and operating.
Digital-to-analog conversion is a relatively straightforward process. There are a number of commercially available chips on the market, and DAC manufacturers use them. Some are better and more expensive than others, but the most expensive DACs don't necessarily have the most expensive chips. Why? Partially because not everyone agrees on what is the best DAC chip, and partly because the associated electronic circuits . . . those that make the signature for the particular designer . . . may necessitate one chip over another. Some use pairs or even quartets of chips together, and this changes the overall design requirements as well.
The reason my $98 CD player sounded "similar" to the $10,000 one is that digital audio is very high tech. There are basic constraints that have to be met in order for it to work at all. The $98 player has to meet those constraints just as does the $10,000 player.
Even though CD digital audio consists of 16 bit words (each word is a code of sixteen 1s and 0s that represent a particular voltage in the music signal), over the past couple of years, DAC chips with 18 and 20 bit "resolution" have entered the consumer electronics arena. A 20 bit DAC chip can decode (resolve) a digital word that is 20 bits (twenty 1s and 0s) in length. So, if the DAC chip only had a resolution of 16 bits, and a 20 bit word came along, the last four (least significant bits) would be ignored (truncated). But, why bother making a 20 bit chip, if the data stream always consists of 16 bit words? Ah well, along with these high resolution chips have come some very neat devices. They take the 16 bit word and, through proprietary algorithms (mathematical formulas), they try to reconstruct the longer word length that was present in the original studio recording (the studios use longer word lengths when recording the digital master). These devices (let's call it a Digital Transmission Interface or DTI) are connected between the CD transport and the DAC. So the 16 bit word output from the DAC is fed into the DTI where it is increased to 20 bits, and the 20 bit word is then fed to the DAC which is capable of resolving the 20 bit word. Of course, the process of increasing the word from 16 bits to 20 requires assumptions, and these constitute the algorithms mentioned above. It is not as good as having the actual original 20 bit word from the master, but there is some audible improvement over just processing the 16 bit words in a 16 bit way (there is, of course, great controversy about this).
The Music Labs DP-102 Mark II DAC uses two Burr-Brown PCM 1702 DAC chips, one per channel, each of which containing two DACs that handle the positive and negative portions of the signal separately. These chips, which contain laser trimmed components and FETs, allow resolution of 20 bits. But before the digital data stream reaches the DAC chips, the data stream must have other things done to it. A digital filter is the first chip that the data stream encounters. This digital filter (Yamaha YM3623B in the case of the DP-102) does several things. It isolates the incoming digital audio data from the digital management data and then reclocks the data so that it has minimum jitter. The reclocked, de-jittered digital audio data are then "oversampled" (8X for the DP-102) by a NPC5842 oversampling digital filter, which makes it easy to then remove all frequencies above 20 kHz. The reason that > 20 kHz frequencies must be removed is that the sampling frequency for CDs is 44.1 kHz, that is, the signal is sampled 44,100 times per second, each sample being 16 bits long, for each channel. According to the Nyquist Theorem, the sampling frequency must be at least twice the highest frequency that is being coded. A sampling frequency of 44.1 kHz is slightly more than twice 20 kHz (a typical upper limit for human hearing). If coding is attempted of a higher frequency, then "aliasing" can occur (remember the old cowboy movies where the wagon wheels appeared to be moving backwards? That is aliasing.) Also, even with frequencies within the limits of the theorem, some frequencies would be interpreted as a square wave (for example, 20 kHz has only about two data points, and would be reconstructed as a square wave). Since square waves are fundamental frequencies with all their harmonics, by removing everything above 20 kHz, the harmonics of, for example, 20 kHz, are removed, and a sine wave results instead of the square wave. Oversampling allows this to be done at the input, through the digital filter, making an analog filter at the output unnecessary, or at least, the job is easier for the analog filter. The DP-102 circuitry is so well designed, that an analog filter is not used at the output at all. This is very unusual and rather unique to the Music Labs DAC. Theoretically, since the analog output circuitry is not there, this is one less set of electronics in the signal path that might otherwise attenuate the signal quality. The DP-102 is completely balanced in the analog section. In the digital domain, the DP-102 is only partially balanced, as some of the crucial components are not easily used in a balanced configuration (they weren't intended to). As the analog and digital sections are completely isolated electrically (by optical couplers), it does not matter whether the digital section is completely balanced.
Jitter has to be reduced before the data reaches the DAC chips. Jitter consists of timing errors in the data stream such that the 1s and 0s are not being received when they are expected [Click here for diagram of jitter]. This results in signals of lower voltages (quieter) being lost, which includes not only small details like the rosin sound of the bow rubbing on a viola, but harmonic texture of the music (each musical note played by an instrument, or any sound in nature, has a fundamental frequency and harmonics which are multiples of the fundamental; most musical instruments and nature sounds have even order harmonics, that is second order, fourth order, sixth, etc., while a few have even and odd). The harmonics of a musical note are lower in intensity than the fundamental, and can be lost if the jitter is high. [Click here to see a diagrammatic representation of how jitter affects the music] If the jitter is really high, the music can sound very harsh. Reclocking the data stream as it passes through the initial circuits of the DAC is one part of jitter reduction. Many DACs, including the DP-102, use a Phase Locked Loop (PLL) to reclock the data so that they are handled in the proper timing sequence. Five elements make up a PLL: a Voltage Controlled Oscillator (VCO), a programmable divider, a phase detector, a loop filter and a reference frequency source. The purpose of a PLL is to take a signal from a VCO, divide the frequency by an integer, and compare that result to a precise and stable reference frequency in a phase detector. The phase detector produces an electrical output that indicates positive or negative phase difference. The phase detector output is fed back to the VCO through a filter. The loop works to adjust the output of the phase detector to a value of zero automatically. That means the output of the programmable divider is precisely on the same frequency as the reference. The whole idea here is to have a continuously correcting circuit to stabilize a frequency of choice. This is of utmost importance when one is converting digital data to analog outputs (such as in a DAC) at precisely the right time. Any skewing of the timing of conversion produces incoherent distortions at the output that are quite audible. The DP-102 MKII reduces these down to non audible levels. The PLL that Music Labs utilizes, stabilizes their conversion timing down to a absolute maximum jitter of 20 picoseconds. Music Labs' PLL is of their own design and is patented in the US as well as other parts of the world.
If the original recording had emphasis (frequencies from 1 kHz - 20 kHz boosted), then "de-emphasis" also occurs, after the jitter has been reduced and the signal reclocked. In the DP-102, this is done in the digital domain.
Finally, Digital-to-Analog Conversion occurs, and the analog audio signal is delivered to the two outputs (right and left stereo).
The DP-102 is a simple black chassis with LEDs and switches on the front. The LEDs indicate power on, coax or optical input, sampling frequency (32 kHz, 44.1 kHz, 48 kHz), whether de-emphasis is occurring, phase (0 degrees - 180 degrees), and mute. Power-on, input, phase, and mute are controlled by toggles. The rear panel has a coax digital input, Toslink digital input, balanced input, RCA analog outputs, and balanced analog outputs. Seven separate power supplies (one for the digital circuits and six for the analog circuits) maintain data stream handling accuracy. A single toroidal transformer powers these supplies, with each supply having its own secondary winding. The electronics are neatly laid out on a superbly crafted glass-epoxy circuit board. The digital and analog stages are optically coupled (rather than electrically), isolating them from one another and reducing digital noise. This is not something you would find in an inexpensive CD player.
We tested the DP-102 with several transports. The most revealing was the test with a low cost CD player ($299) that has a digital coax output. This is the first kind of test you should perform if you are planning to upgrade your CD player to have an outboard DAC. It will convince you that a good DAC can, indeed, improve the sound quality. The differences are subtle, even with the inexpensive player, but you need to hear these differences in as great a magnitude as possible. When you do, you will know what to listen for when upgrading a better CD player to have the outboard DAC. Make sure that a cable made for digital transmission (75 Ohm) is used. We experimented with regular audio cables (about 50 Ohms), and the sound was very gritty.
What we found with the addition of the DP-102, was that the music had more body to it. That is, we could sense the harmonics of the instruments. The sound was richer, had more texture. We could hear not only the notes that the string bass was playing, but the sound of the bow moving across the strings. We could hear finger pluck and lips vibrating on brass mouthpieces. All of these things were absent or deeply recessed when the player's own DAC was used (A/B testing). Interestingly, we could not hear an improvement when we used the Toslink optical connections and a laserdisc player. Some audiophiles like the optical path for digital, and others are not impressed. I guess we fall into the latter category. But, using the electrical connection (coax), the differences were apparent. Recent evidence indicates that a major source of jitter . . . perhaps the largest source . . . is at the point the digital data stream enters the DAC chips. This may be one reason why so much improvement can be found using a good DAC with a basic transport. I am not going to give you the typical "this is the best DAC we have ever heard" kind of hype that you see elsewhere. There is no "best" because listening is the final tool for evaluation, and listening is subjective. What I can say is that a good DAC can make a significant improvement in the listening enjoyment of digital audio, and Music Labs has done an excellent job with the DP-102. With a 20 year warranty, the quality control has to be outrageously good, so this is one component meant to last a lifetime.
John E. Johnson, Jr.
© Copyright 1995, 1996, 1997 Secrets of Home Theater & High Fidelity
Return to Table of Contents for this Issue.