AES Section Meeting Reports

Los Angeles - February 28, 2012

Meeting Topic:

Moderator Name:

Speaker Name:

Meeting Location:


Bruce Hofer, co-founder and owner of Audio Precision discussed the issues designers face when selecting and placing components to achieve superior analog performance. He was careful to note that his definition of superior performance is strictly based on objective measurements of parameters such as noise and distortion.
Ideal components only exist in text books. Real world components have parasitic properties, operational limits, temperature and voltage coefficients, long-term drift, non-linear characteristics, frequency dependence and environmental sensitivities. Hofer showed that seemingly minor voltage dependencies can result in significant and very predictable distortions.
Nearly all passive components are subject to non-linear distortions. Some mechanisms are clearly understood and are easy to characterize. Others are more nefarious. For instance, two resistive networks may look identical when comparing DC parameters, however one built on a passivated silicon substrate will exhibit excessive capacitance and generate harmonic distortion between resistors. This causes unpredictable AC performance. The latter distortion was theorized to be the result of PIN diode formation between the resistive layer and the silicon substrate.
Hofer pointed out that C0G (NP0) capacitors exhibit great performance at a reasonable cost but noted that costly metal film capacitors are often hailed by audiophiles. His experience has shown that poor construction within metal foil caps can lead to diodic action at the lead attach points. He also noted that these caps should be used at no more than 10% of their rated voltages to minimize voltage dependencies. Sometimes better performance is achieved through use of extra parts rather than "better" parts. One can minimize the voltage dependence of resistors by using series combinations of smaller resistors instead of a single larger resistor.
Op-amp selection is also a complex process of negotiating trade-offs. He prefers to simply divide the market into op-amps with JFET or bipolar inputs. The bipolar designs feature lower noise voltage while the JFET designs have much lower current noise.
Hofer showed how one can predict noise performance with excellent correlation to measurements using a spreadsheet and how critical board layout is to performance. Short connection paths may seem like a good idea to an auto-router but they can lead to larger loop areas. Sometimes a longer path can reduce the loop area and lower the circuit's magnetic susceptibility. He also recommended the use of power busses rather than power planes where the actual current path may be unpredictable.

Written By:

More About Los Angeles Section

AES - Audio Engineering Society