Do you know what the threshold is for detecting that some fraction of a signal is emanating from a two-channel system's lower-level channel, rather than 100% coming from the louder one?
Absent time-delay effects, in the region of highest perceptual acuity the answer is about minus 13dBr---that is, at less crosstalk than that the listener hears the sound as coming entirely from the louder channel.
Bear this in mind when you struggle to achieve truly impressive crosstalk numbers. The extremes to which some go in this pursuit border on the ludicrous, including the audiophile's abiding respect for "dual-mono" designs, even extending to separate mains inputs!
Bill, I like your comment about kicking the artsy-fartsy engineer out of the room but it will take a very large boot indeed.
Of course, if headphones had four wires instead of three and the ground currents only mixed at the amplifier output ground, then the whole ground current mixing would be moot.
I think this is a very dubious (and certainly not a new) technique. It begs all kinds of grief for the customer when ground loops occur (which is nearly always, except with an actual headphone). Imagine the effect of several mA of 60 Hz "harmonic soup" flowing into the jack's ground when it's connected to external audio gear. It's bad enough that the connection is unbalanced (making for inevitable noise due to common-impedance coupling in the cable itself). Computers in general, and laptops in particular, almost always have internal grounding issues severe enough to make an audio professional recoil in horror. Adoption of "compensation" techniques like this one only makes matters worse. PCB designers of these products should learn from AES48-2005 about recommended grounding practices for audio I/O ports. Kick the "artsy-fartsy" industrial designer out of the room and put the audio driver where it belongs!
Bill Whitlock, Fellow of the Audio Engineering Society and IEEE Senior Member
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...