AES British Section lectures - May 2007
Balanced Audio
Bill Whitlock, President and Chief Engineer Jensen Transformers Inc
www.jensentransformers.com
On May 15th the UK Section of the AES hosted "Design of High-Performance Balanced Audio Interfaces" by Bill Whitlock. Bill is the President of Jensen Transformers and has had a long history of designing high-performance analogue audio components.
Bill started by addressing the myth that improper AC power wiring causes noise. Bill went on to warn against the practice of disconnecting the ground pin as being exceptionally dangerous.
Bill went on to define the primary difference between balanced and unbalanced interfaces being their impedances. An unbalanced line has one line tied to ground, whereas balanced lines have the same impedance to a third line (normally ground). He noted that the only time matched input and output impedances are required is at high frequencies, where the line exhibits transmission-line effect (i.e. not at audio frequencies).
Bill discussed the problems with unbalanced interfaces involving leakage currents that flow in the grounded conductor, while the essential feature of balanced interfaces make noise pick-up identical in both conductors. He stated that only shield resistance makes a difference to the noise in unbalanced cables.
Bill showed that using a transformer significantly improves in input stage's tolerance to variability in the output stage's impedance. He noted that the common-mode rejection ratio was often measured with a perfect source, but that modern measurement standards require an imperfect source.
Bill described how using a boot-strap op-amp to significantly increase the common-mode input impedance at audio frequencies allows differential amplifiers to imitate the characteristics of transformers.
Bill then discussed various different line driver designs and how good they were for driving unbalanced inputs.
Bill covered the topic of balanced cables and the effects of unbalanced line capacitance. He also gave his reasoning for why the shield should be only connected at the driver end. He also discussed the Pin-1 problem, where designers connect the shield pin to a PCB ground trace also used for signal circuitry introducing noise.
There followed a lively question and answer session. A recording of the lecture is available from the UK section website at www.aes.org/sections/uk/meetings/a0705.html.
Report by Tim Harris