Wow, have we hit a sensitive area. In trying to answer the previous readers questions, I have tried to avoid the use of technobable, since this forum is intended for a general readership. However, since I wasn't the first to bring in jargon, I'll throw in a bit of my own. Stability: Any amps stability is governed by the Nyquist stability criteria: all of the circuit poles must lie in the left half of the complex impedance plane for stability; or, an amp is unstable if the closed-loop gain plot encloses the point -1 + j0. Since these calculations may be difficult, a graphical method can be used; ie, measure and plot the phase and gain of the amp. This gives the well known Bode plot. The gain (hopefully negative) where the phase is -180 deg is the aforementioned gain margin, and the phase shift where the amps gain is 1 is the phase margin. (Reference "Integrated Electronics", by Millman and Halkias, or "Network Analysis and Feedback Amplifier Design", by Bode, a seminal work on the subject of stability. The amp designer has complete control of the phase/gain response. If he wishes to sacrifice stability to achieve a slight improvement in high frequency response, I consider that poor design practice. Cables: Audio cables are almost never matched to the source or the load, therefore, they can always be considered resonant line structures. But why bother? The wavelength in my speaker cable at 500 kHz, which is a good unity gain point, is still 1800 feet. (I measured my cables this morning, the impedance is 4.3 ohms, and the eff dielectric const is 1.2.) For such a long wavelength, one can do better by using a lumped model (discrete resistors, inductors, caps) for the cable/speaker, as opposed to a distributed transmission line model. (Reference "Transmission Lines, Antennas, And Waveguides", by King, Mimno, and Wing; also "High Speed Digital Design", by Johnson and Graham; they recommend using distributed models at frequencies above 1/4 wavelength, here, that's about 10 megahertz.) We can connect a couple of parts to the amp output, and get the same stability result as the cable and speaker. There is only a fraction of a degree of the signal spread out over the cable between the amp and the speaker. Reflections just don't exist in the speaker/cable system. This stuff is well known in the industry; computers, telephone equipment, radios, TVs are designed by standard principles every day, and no one questions whether they work. I stand by my previous statement: If an amp can be made to oscillate by a certain brand of cable, either the amp, the cable, or both are poorly designed. No apologies today for the very long post. Rich
