Digital Processor Reviews

After being connected to a transport, the Lens will show a number on the display's right-hand side. This number indicates the transport's speed accuracy, measured in Parts Per Million (ppm). We're not talking about the transport's jitter, but its average deviation from the standard output frequency. Because of crystal oscillator tolerances, some transports run slightly faster or slower than the correct frequency. The highest deviation Genesis has seen is 250ppm speed error, with 50ppm being a typical value. The Mark Levinson No.31 read 8ppm speed error, the Sonic Frontiers SFT-1 had 45ppm error, and the Parasound C/BD-2000 showed 35ppm.

How it works
Nearly all jitter-reduction boxes that fit between your CD transport and digital processor use the same approach to reducing jitter, filtering jitter from the transport's output with a circuit called a phase-locked loop (PLL). By cascading two PLLs within the jitter box, significant jitter reduction can be achieved. The first PLL locks onto the incoming signal, the second PLL filters jitter. The Meridian 518, Audio Alchemy DTI v2, DTI•Pro and DTI•Pro 32, Sonic Frontiers UltraJitterbug, and others use this technique.

Genesis has taken a totally different approach to jitter reduction. The Lens receives the S/PDIF or AES/EBU output from your transport with a conventional input receiver (the Crystal CS8412) and PLL, but there the similarities end. Having recovered the datastream, the Lens totally breaks it down into its component parts. The raw audio data are extracted and fed into a 500 kilobyte Random Access Memory (RAM). The subcode is displayed on the Lens front panel, then not used again. Other housekeeping bits in the subcode not needed by the digital processor are thrown out. At the Lens output, the S/PDIF datastream is reconstructed from scratch.

The Lens achieves its jitter rejection by putting the audio data through the half megabyte of buffer memory. The clock recovered from the transport clocks the audio data into the memory, but the output clock that feeds the data to your digital processor is generated by a precise, carefully realized clocking circuit in the Lens. This technique totally isolates your digital processor from the transport's clock—and its jitter. As in the opening thought experiment, where we read music from a CD into a huge memory, the Lens's output clock driving the digital processor bears absolutely no relationship to the input clock from your transport. In theory, the Lens should be an impenetrable barrier to transport jitter.

The audio data may be read into the Lens with your transport's jitter superimposed, but the data are read out with the precision of the Lens's output stage. Considering that the Lens's buffer can store more than two seconds of audio, no trace of the transport's jitter signature could possible remain at the Lens's output—in theory.

Implementing this drastic approach to removing jitter is trickier than it sounds. The Lens actually measures the transport's output frequency and assigns just enough memory to accommodate this error. A transport whose clock is close to the reference frequency (the Lens's output oscillator, which has an error of 2-3ppm) would need less memory space than one that was less closely matched. Because the input and output clocks aren't synchronized, the buffer tends to fill up slowly over the course of playing a CD on fast transports and run toward empty with slow transports. Genesis claims that the Lens's half megabyte of RAM won't overflow or empty with transports having a 1000ppm error over a 70-minute CD. Most transports use about a quarter of the RAM's two-second storage.

It's easy to see how the buffer's elasticity could compensate for transports whose output clocks ran faster than the Lens's fixed-frequency clock—the buffer gets fuller and fuller as the CD plays through. But if the transport runs slower than the Lens, the microprocessor lets the buffer fill with up with as much as two seconds of music before the output stage starts clocking it out. The buffer would then get less and less full as the disc plays through.

Note that the Lens's output clock runs at a fixed frequency, and can't be "pulled" in frequency like a Voltage Controlled Crystal Oscillator (VCXO) can. Genesis found that VCXOs had inherently higher jitter than the Temperature Controlled Crystal Oscillator (TCXO) used in the Lens (footnote 2).

Genesis's RAM-based approach in the Lens allows the input and output clocks to be asynchronous, yet maintain perfect bit-for-bit data integrity. It just won't perform sample-rate conversion, a feature not needed in a consumer product.

By making the memory shorter when driven by transports with a precise output frequency, the Lens avoids the situation of having a two-second delay between the time you push Play and when you hear music. With a 2s delay, there's the danger of pressing Pause, hearing the music continue, thinking the transport didn't accept the command, and pressing Pause again—which would start filling the buffer again. This RAM assignment technique saves a lot of confusion. In addition, the buffer dumps its data when the transport is put in Pause, then starts over when the transport goes back into Play mode.

---

Footnote 2: The Analog Devices AD1860 Asynchronous Sampling Rate Converter chip, the device at the heart of the Digital Domain VSP and Z-Systems jitter-reduction boxes, also has input and output clocks that are independent (that's what the "Asynchronous" means). The AD1890, however, changes the audio data by interpolating new samples between the input samples. At the output, the chip continuously throws out lots of unneeded samples to achieve the desired output sampling rate. Only a small percentage of the final output samples are identical to the input samples. What you put in isn't what you get out. The AD1890's output is jitter-free, but the sample amplitudes can be slightly different from what is required. The part takes in the right samples at the wrong time, and outputs the wrong samples at the right time. [The difference, however, is specified by Analog Devices as being below the 16-bit noise floor.—Ed.] Ironically, the error introduced by this process is conceptually identical to the error introduced by jitter.—Robert Harley