Assessment of the Importance of Audio
Measurements in High-End Design
Jeff Rowland, President and Chief Designer, Jeff Rowland Design Group
(Excerpted from Ultimate Audio magazine, Spring 1999 pp.16-17, "Friend or Foe? High End Designers Reassess the Importance of Audio Measurements" by Myles B. Astor)
There is definitely a close relationship between test measurement (specifications) and subjective sound quality. However, current testing procedures commonly used in audio development, manufacturing and the audio press are inadequate to properly correlate and relate measurements with subjective experience.
Many well-meaning critics hold that so far as an audio component measures "perfect" in certain areas, the audio component will not have a "sound of its own," and it will be indistinguishable from a component with similar measurements. The measurement criteria usually involve flat frequency response, insignificant static and dynamic nonlinearities, high input impedances, low output impedances, low noise and crosstalk levels, etc. Yet these common notions fail to uncover the actual performance potentials of complete audio systems.
The speed and processing abilities of current computer-interfaced testing equipment enable Jeff Rowland Design Group (JRDG) to perform tests which were impossible a few short years ago. Among the most important tests for musical reproduction accuracy is the Fastest procedure developed by Audio Precision, Inc. This test will uncover wide band, dynamic nonlinearities in DUTs (devices under test) previously unexposed by twin-tone IMD (intermodulation distortion) and THD + N (total harmonic distortion plus noise) testing procedures. A composite of either 32 or 64 discrete tones, non-harmonically spaced throughout the 20 to 20 kHz bandwidth, is introduced to the DUT inputs. The DUT outputs are routed back into the computer interfaced analyzer, which sharply attenuates each of the 32 or 64 original tones. The resulting intermodulation (sum and difference) tones of the original tones are then integrated and displayed for analysis and/or continuing product development. Note that this test signal more accurately represents a musical signal due to its high content of discrete frequencies. Any nonlinearities present in the DUT directly create a multitude of spurious difference frequencies, which fall throughout the lower amplitude ranges of the entire audio spectrum, significantly limiting and compressing the usable dynamic range. A poor result on this test correlates to the common listening experience of dynamic range compression, congestion, loss of detail and obscuration of the silence and harmonic integrity in music as the material becomes increasingly complex.
Another overlooked area of testing is evaluation of the DUT in a radio frequency interference (RFI) environment. Thorough testing and analysis of RFI, immunity in audio equipment is increasingly important due to the proliferation of telecommunication devices and computers throughout the world. This testing involves injecting minute amounts of continuous or multitone RF signals into the input, output, AC mains, and chassis of the DUT. The output of the DUT is analyzed much the same as above. Poorly designed equipment which under common lab tests can measure fine, will escape this particular scrutiny. In a metropolitan environment, for example, multiple RF signals can be demodulated within the input circuitry of poorly designed audio equipment. The resulting sum and difference frequencies will manifest throughout the audio frequency range and create an obscuring effect upon the musical signal much in the same manner as mentioned above.
The act of transferring an audio signal from one system component to another without audible degradation is very difficult. The existence of significant ground voltage differences and the resulting noise currents that flow between system components must be considered in the system as a whole. Carefully designed and executed system grounding schemes can reduce these problems but not eliminate them. Audio equipment that may display excellent test results in isolation can fail miserably in a typical audio system configuration.
The degree of isolation and immunity from these undesirable effects and their resulting problems is indicated by the CMRR (common-mode-rejection-ratio). (Note that balanced interconnection schemes are referred to in this discussion. Single-ended interconnected systems, by their inherent design limitations, can never overcome the problems discussed here.) Very few manufacturers give specifications relating to CMRR; the audio press fails to test for this important test parameter altogether. To make matters worse, most audio designers who do test for CMRR do so by driving the opposite phase inputs shorted together, which is both unrealistic and misleading. Tests which account for slight source impedance mismatching between opposite signal phases, a reality in all audio systems, are rarely done. Noise rejection in a balanced system has nothing to do with signal amplitude symmetry, erroneously regarded as an asset. Noise rejection has everything to do with the balance of common-mode impedances. Unfortunately, this is rarely accommodated for in equipment design or mentioned in performance specifications.
Audio hardware and interconnect cables are sensitive to mechanical vibration (microphonics). Sound energy, transferred through structural and air mediums, can significantly impact the performance of all audio equipment in many domestic environments. Again, equipment sensitivity to this condition is rarely tested in manufacturing or considered during the design process.
In conclusion, if more thorough testing procedures would be implemented at the design, manufacturing and reviewing stages within a product life cycle, definite correlations could be made between technical specifications and subjective critiques. However, this is not generally the case today. The result is confusion and a distrustful attitude towards technical tests altogether. Our hope is that more designers and reviewers will explore many of the test and measurement possibilities that are now possible.
Bio Notes: Jeff Rowland's involvement in consumer electronics began in 1980, when he designed and manufactured a line of audio power amplifiers, preamplifiers and electronic crossovers, under the company name of Rowland Research. Recognition of his unique approach to audio amplifier design led to worldwide acceptance and distribution of a complete line of amplifiers and preamplifiers in 1986 under the current name of Jeff Rowland Design Group, Inc. Mr. Rowland has provided innovative solutions to design problems, such as his pioneering implementation of balanced circuitry (1986), transimpedance topologies (1988), preamp remote control (1988), low-resonance chassis design (1992), and battery power supplies for audio applications (1992), transformer interfacing (1995), and high power integrated circuit applications (1997). He is currently involved in the design of ultra-low distortion, high efficiency, modular amplifiers for stereo and multichannel audio applications.
2nd of August, 2010