This review page is supported in part by the sponsor whose ad is displayed above
Outside is a new feature so named for a number of reasons - the unconventional nature of the main system with its parametric equalizer and pro-world bass amp; the unusual comparisons; the regular solicitations of listener feedback from a small circle of fellow audiophiles who will create multiple points of view; and the deliberate pursuit of review subjects outside the beaten path. Manufacturers submitting formal review loaners for the Outside features will ship to Mike Smith and afterwards issue the call-tag for his premises. Mike's circle of listeners will forward the items from one listening space to the next independently or congregate at Mike's place for joint listening sessions - Ed.
Reviewer: Mike Smith
Source: Red Wine Audio battery-powered Squeezebox 3, Lessloss modded Rega Planet or Modwright-modified Music Hall CDP
Preamp: Lamm LL2 Deluxe
Subwoofer amplifier: Crown K2 fed by dbx Driverack 260 parametric EQ
DAC for main array: Lessloss 2004
Amplification for main array: Yamamoto A-08s or Red Wine Audio Signature 70s
Loudspeakers: Zu Audio Definition 2s
Cabling: assorted, mostly Zu
Room - Finished and carpeted basement, 26 x 17 feet, 9-foot ceilings. Several doorways and acoustic ventilation features. Moderate room treatments.
Source: Red Wine Audio Squeezebox 3 or old Technics CDP
Amp: Red Wine Audio modified Teac AL 700P (Tripath)
Loudspeakers: VMPS 626r or Tannoy LR8 monitors
Room: 40 x 23 feet, 9-foot ceilings, suspended wood floor, no dedicated room treatments
Review component retail: starting at €2,995
Brace yourself. This article is kinky. Correction. It isn't risqué in itself but does (ahem) beat around the bush a fair amount because the gear is, shall we say, adventurous. Hmmm. Maybe that's not true either but one of Lessloss' core design philosophies is the master/slave relationship. I heard about S&M as a kid but never knew what it meant. Maybe that's why I chose this assignment. Too freaky? It might be time to turn the dial.
As a living language, English grows branches and trims others. Like a tree, more is produced than lost and misunderstandings are inevitable. Disappointingly, master and slave in the context of digital audio reproduction have nothing to do with sadomasochism as I might have earlier hoped. Rather, these terms refer to clocking schemes between different portions of the reproduction chain. In this parlance, master refers to the clock generator and slave denotes the receiver. Where a typical transport/DAC will utilize two master clocks to time data transfer, the Lessloss strategy hands all responsibility to the DAC by disabling the transport clock. The basic premise is that no two clocks will operate at precisely the same frequency. If true, it necessarily follows that having two active clocks in a digital system guarantees compromise. I cannot prove this theory and am stuck with those most imperfect testing mechanisms, my ears, to form opinions.
These gents from Lithuania are not the only manufacturer to elevate the importance of the clock - Wadia, dCS and Esoteric all have current products utilizing alternative clocking schemes. Some offerings include a master clock as a separate component between transport and DAC, a statement of perceived importance to be sure. Why?
|Lessloss believes the central problem with digital audio is jitter, a subject of much debate in the audiophile world. Some believe the pursuit of crushing jitter is a marketing lead balloon akin to Sonic Rocks (a confirmed hoax) while others hold it to be the central limitation of the Redbook format. Clocking is thought to be a central culprit of jitter and a logical target if eliminating it is a design goal. Lessloss believes their solution effectively addresses clock-induced jitter but the pursuit does not end there. They believe it hides in numerous places. Various measurements are posted on their website clearly showing differences in jitter performance between connection schemes though some would argue how audible the shown differences would be.
Power supplies are said to be another primary cause. The 2004 uses a hybrid power supply with AC on the digital section and batteries for analog. Some prototypes were entirely battery powered but experimentation proved AC superior on the digital circuit. The review sample runs about 10 hours on a charge after which it must be turned off. I have never run out of time on the batteries on either my personal unit or the review sample. A minor update, standard on all current units, automatically recharges with continuous operation possible. Batteries last around 2 years on average, replacements are user-serviceable and cost around $30. Integration of the battery charger allows one IEC power cable to run the entire machine. As with all the design ingredients, the Lessloss website has an excellent description of the power supply and why two sources are used.
Digital conversion service, per channel, is provided by a Burr-Brown (now Texas Instruments) PCM1704U-K parallel multi-bit converter chip no longer in production. More information on this chip is available at the Lessloss website of course. To quote directly from there:
In slave mode, all sampling rates up to 192kHz are supported. In master mode, the typical sampling rate supported is 44.1kHz with 33.8688 MHz or 16.9344 MHz output to the clock output for slaving a CD player to the device. However, upon special user request, we will accommodate all of the following sampling rates: 44.1kHz, 48kHz, 88.2kHz, 96kHz, 176.4kHz and 192kHz. Full 24-bit resolution is supported at all frequencies. Digital deemphasis is supported at 44.1kHz and 48kHz:
--> 8x oversampling at 44.1 and 48 kHz = 352.8 and 384 kHz;
--> 4x oversampling at 88.2 and 96 kHz = 352.8 and 384 kHz;
--> 4x oversampling at 176.4 and 192 kHz = 705.6 and 768 kHz.
Connections include 1 x IEC power, 1 x RCA SPDIF input, 1 x RCA clock output, 1 x RCA analog output, 1 x XLR analog output. Overall construction is very solid and the unit feels quite heavy. It is hand-built, hand soldered and though simple in appearance, both the silver and black versions look and feel first rate. The unit gives a thorough impression of a mature product.
The 2004 is a highly flexible platform. Because it has been designed with conventional and master/slave setups in mind, it has to work with many different transports and their various architectures. Options and pricing: Black or silver case 2995 euros includes superclock output and shipping worldwide. Wordclock output adds 50 euros; additional superclock frequencies 50 euros; oversampling bypass to create a non-oversampling machine 50 euros; asynchronous upsampling circuit 195 euros; distortionless volume control 495 euros.
For specs, we once again pilfer the Lessloss website:
There are no US dealers at present and all orders are processed factory-direct. To demystify some design choices, I exchanged e-mails with Liudas Motekaitis, one of the principals:
What is jitter and why is it bad?
Digital audio, as everyone knows, is 0s and 1s. But they got there because they were quantized. An analogue signal is quantized (chopped up into samples) by an A/D conversion process. The A/D conversion process is run by a clock. The clock tells the converter when to chop. In CD audio, there are 44,100 chops a second. The more continuous and evenly spread out these chops, the more accurate the conversion process. There would be no jitter if the chops were all exactly evenly spread out. The more uneven the time-chopping clock runs, the more it contains jitter. So jitter is a shake-up of the otherwise perfectly continuous and even quantization of time. Jitter is not a slow or fast clock. Jitter is a slow or fast sample. Jitter is a real-time fluctuation of the evenness of the samples. Jitter is what nobody has ever liked in the sound of digital audio.
How is jitter created? Is it on the recording originally? Is it on the disc? Is it created solely by the electronics?
Here's an analogy: take a ruler and hang pieces of string from your ceiling exactly one centimeter from one another in a line. At the end of all of the pieces of string, attach sewing needles. You have quantized a meter into 100 cm. The needles mark the spots. Now take a fan and add turbulence to the ambient air. You have added jitter to your quantization model.
Jitter is created due to real-world interference. It can come from power supply noise, physical vibration, crosstalk among components, inherent noise in the components, intermodulation with the digital data itself, antennae effects from the cables (digital, audio and mains), resonant reflections from the device's casing material and just about everywhere else you can think of. Jitter gets into the recording at the A/D process (during recording). The better the recording equipment, the better the digital data. Jitter (again) gets into the audio you hear at the D/A process. Nowhere else is jitter of any significance to our ears. When you listen to a digital recording, you hear the jitter which was recorded plus the jitter which is getting into the playback now. If you annihilate all of the jitter which is getting into the playback now, it is then time to write the recording engineer and tell him to do his part, too.
You might consider, in the analogy above, that the exactness of the ruler and your hanging skills is the jitter present during the recording (A/D) process, and the fan speed the jitter present during playback (D/A). When the needles are just hanging, they represent the CD. Sometimes you can re-burn a CD and successfully lower the jitter on it. That would be like nudging the strings around on the ceiling to align their spacing more perfectly. Then, when you turn on the fan, the average quantization will be by a small but clearly discernable amount better.
Why one clock?
The use of one clock avoids the use of asynchronous resampling. Asynchronous resampling is less effective in lowering jitter than synchronous resampling. Synchronous resampling can be taken to the extreme. Asynchronous can only get so good. It adds its own types of distortion which are not exactly jitter but a type of rounding errors.
Why the clock as close to the DAC chip as possible?
Because it is only the jitter which is present at the very DAC chip (or ADC chip) during conversion which is of interest. And the further you are from the original clock signal, the more jitter there is.
Why not a one-box player that can achieve one clock close to the DAC chip?
This is thoroughly possible. But it is far more difficult to do well. In the CD player, there are far more sources of interference since there are so many more functions which must take place there (high frequency feedback schematics, laser control, turning mechanisms, display etc.). A stand-alone DAC gives you the luxury of distance and a separate power supply and power cable. In effect, a transport and DAC is but an expensive one-box CD player, and a one-box CD player is but an expensive iPod, and an iPod is but an expensive and specialized laptop, and a laptop is but an expensive pad of paper and pen, and pen is but an expensive feather and ink, and a feather and ink is but an expensive finger on the sand.
Why battery and AC?
We've tried all the configurations we could come up with. This is the solution which yields the top possible performance thus far.
What's different about your volume control versus other solutions?
In our solution, the volume level is actually a discreet customization of the output level of the device. So, if you would normally have a 2V RMS output from your DAC and you wanted to change the volume level, you'd need an external preamp to lower the level. Our solution lets you bypass a preamp altogether and adjust the output level itself in 21 x 3dB increments. A portion of the signal is grounded according to which resistor you choose with the switch. The output buffers run at full power regardless of what output level you choose. This allows direct connection to power amps or active loudspeakers. If at a later date one would choose to use the DAC 2004 with a preamp, one needs only to put the volume control at the top level and it would be the same as having the DAC 2004 at the industry standard 2VRMS output. The signal path at all of the volume levels is the same. We use a Japanese Seiden switch, which has better silky action and is more exact than the ELMA switch from Switzerland.