 |
|
|
Ask Joe! VP of Engineering Joe Bryan Answers More Questions About the 2192
We covered the UA 2192 Master Audio Interface our high-end, 2-channel AD/DA converter and master clock, in the September 2003 Webzine . We've had so much interest in the 2192, so we decided to ask the doctors (actually Joe Bryan, our VP of Engineering & Technology) some more questions about it...
Q: The specs say the 2192 has a "DC-coupled, fully dual-differential, matched-FET, all discrete Class-A, no-compromises analog signal path", what the hell does that mean?
Basically, it means the analog signal paths (from the line-ins to the A/D converter, and from the D/A converter to the line-outs) are as good as they can be.
Each element of the description means something different in terms of sound quality:
"DC-coupled" means there are no capacitors in the signal path to introduce phase distortion, which can smear your high frequencies and take the punch out of your low frequencies.
"Fully dual-differential" means there are independent, identical circuits processing both the + and - sides of your differential signals. This reduces distortion, and improves common-mode rejection and dynamic range. It also improves imaging by eliminating cross-talk between channels.
"Matched-FET" means we use high quality field-effect transistors (FETs) with matching characteristics. FETs are a special type of transistor that share the good-sounding characteristics of both tubes (low-order, "musical" 2nd and 3rd harmonics) and standard bipolar-junction transistors or BJTs (clarity, transparency, fast transient response). If FET circuits are designed correctly, they don't share any of the negative aspects of tube or BJT circuits (noise, harshness, slow transient response). However, if used incorrectly, FETs can be affected by a type of parasitic circuit capacitance (called the Miller Capacitance) which reduces transient and high-frequency response. We use a special biasing technique in our FET op-amps to eliminate this problem.
"All discrete Class-A" means there are no IC op-amps in the signal path (just individual, discrete parts), and every gain stage is pure class-A biased. Without getting too technical, class-A bias basically means the amplifiers are always drawing current from the power supplies, and they amplify both the positive and negative half-cycles of the signal. Standard IC op-amps use class AB biasing, which amplify the positive and negative half-cycles using different parts of the circuit. The class-A "constant-current" mode means the circuit doesn't suffer from "supply-droop" distortion when handling low frequencies and sharp transients, and there is no "cross-over" distortion inherent with class-AB IC op-amp designs. Many companies tout their circuits as "class-A", but neglect to tell you they're only using one or a few single class-A transistor stages, and the rest of the circuit is all IC op-amps. This is like calling a VW bus a racecar because it has a turbo-charged engine.
Q: How should I clock my setup, and why does it matter what clock I use?
This is a complicated issue, and some important aspects are not very well understood by some audio engineers.
Digital clocking is used to maintain synchronization between different digital devices. There are two primary purposes for clock synchronization:
- Digital Conversion. Analog-to-digital (A/D), digital-to-analog (D/A), and sample-rate conversion (SRC) all need extremely accurate clocking in order to properly convert the digital data. A low-quality clock can degrade the signal in many ways, including loss of transparency, clarity, imaging and transient response, and increased noise and distortion.
- Digital Transmission. All digital devices need accurate clocking in order to properly transfer digital data between devices. A low-quality clock can cause data reception errors, which add distortion and noise, and if the clock isn't synchronized correctly, samples may be dropped or repeated causing clicks or dropouts.
Clock quality is defined two ways: First, the sample rate must match the signal. This is referred to as "sample rate synchronization". Second, the clock signal must be stable over both short- and long-term clocking intervals. "Jitter" refers to short-term clock accuracy, and "stability" or "drift" refers to long-term clock accuracy. These terms are discussed in more detail below.
Sample rate synchronization is required for proper digital transmission, and is relatively easy to maintain. Basically, there must be only one "clock master" for all devices running at the same clock rate connected together. This is done by setting one device in "master-mode" (i.e. transmit clock, and synchronize to internal clock); setting every other device in "slave-mode" (i.e. receive clock, and synchronize to external clock), and then routing the appropriate clock signal between the master and slave devices. Keep in mind that any device, whether it's the clock master or a slave can send or receive data once everything is synchronized correctly.
When doing digital conversion, it's best to have the converter be the clock master. For example, if you're recording, clock everything off the A/D converter. Likewise, if you're mixing, clock everything off the D/A converter. If you're running multiple converters, clock them all off the same high-quality master clock.
For all-digital transfers, e.g. a digital transfer from one DAW or storage device to another, clock synchronization is maintained by simply setting up the proper master-slave relationship between devices. Digital transfers can be affected by clock jitter, but not in the same way clock jitter affects analog conversion. This is a widely misunderstood concept we'll discuss in detail below.
Clock jitter is short-term variations in the edges of a clock signal, and clock drift is long-term variations in the clock rate. A clock could be very stable over the long-term, but still have jitter, and vise versa. Timing variations are caused by noise and/or interference. If the noise/interference is a high-frequency signal, you get clock-jitter, and if the noise/interference is a low-frequency signal, you get clock-drift. A car with an out of balance wheel may drive straight, but you'll get lots of vibration (jitter), conversely, a car with a loose steering wheel might have a smooth ride, but it'll drift all over the road.
Clock drift affects long-term synchronization, like sound to picture, and can introduce slight pitch variations in the audio. Usually however, the drift is so slow that these pitch variations are only tiny fractions of a cent, and thus unnoticeable.
Clock jitter affects digital transmission and digital conversion differently.
Clock jitter in digital transmission can be caused by a bad source clock, inferior cabling or improper cable termination, and/or signal-induced noise (called "pattern-jitter" or "symbol-jitter"). Digital signal formats like AES/EBU, S/PDIF, and ADAT all embed a clock in the digital signal so the receiving device can synchronize to the transmitted data bits correctly. The clock used for data-recovery is extracted from the signal using a clock synchronization circuit called a phase-locked-loop (PLL). This data-recovery PLL must be designed to respond very quickly to attenuate high-frequency jitter and avoid bit errors during reception. This clock from the data-recovery PLL cannot be used to generate the clocks used for digital conversion without further clock conditioning! This is a very common design flaw in most low- and mid-range digital converters.
Clock jitter in digital conversion is what most people refer to when they discuss jitter. It's easily observed on a digital signal by looking at its spectrum in the frequency domain. A jittery signal will have "side-lobes" around each frequency, and/or spurious tones at random, inharmonic frequencies. Usually, the jitter will be worse with higher signal frequencies. You can test your converters by sampling a high-quality 10kHz sine wave, and viewing it the frequency domain (available with any good wave editing software package).
All modern over-sampling digital converters require a clock (called "m-clock") that is many times (typically several MHz) higher than the sample clock. M-clock is easy to generate when the converter is the clock master, but quite difficult to generate correctly when the converter needs to sync to an external clock.
External clock is either from a BNC clock input, or from the digital AES/EBU, S/PDIF or ADAT receiver. The BNC clock cannot be used by the converters until it's multiplied up to the m-clock rate. This requires a PLL or other frequency multiplier circuit which will either be cheap and jittery, or expensive and clean, depending on who makes your converter. As we said earlier, the clock recovered from the digital inputs is unsuitable for use as the converter's m-clock, but because it's conveniently at the same frequency, many designers don't bother cleaning-up this signal.
Since the clock recovery, clock multiplier, and clock conditioning circuitry define the jitter for analog conversion, no external clock source can cleanup the jitter introduced by these circuits, regardless of how perfect the external source clock is. The best they can do is avoid making it any worse, but this is hardly worth the cost: It's much better (and less expensive) to get a good converter than it is to try and fix a bad one with an expensive master clock. The only reason to spend money on a high-quality master clock is to ensure multiple devices are synchronized correctly. This is essential for working with audio for film/video, or when synchronizing multiple high-quality converters. A poor master clock can affect imaging and clarity in a multi-track environment.
The 2192 provides high-quality analog conversion for recording and/or playback, master clock generation, resynchronization and distribution, and digital transcoding (format conversion). With its pristine audio path, high-quality clocking, and simple front panel controls, it makes the perfect master audio interface for your digital studio, and is a very cost effective way to improve your overall sound quality.
- Joe Bryan