Created on 2012-11-09 00:52:00
Crosstalk is expressed as a ratio. The measured crosstalk level in an undriven channel is divided by the measured level of the stimulus tone in the driven (source) channel. Crosstalk is typically stated in decibels (dB).
Crosstalk should normally be measured as a function of frequency, since the values are likely to vary strongly with frequency. The most common circuit mechanisms causing crosstalk are stray capacitive and inductive coupling. If circuit impedances are approximately constant with frequency, crosstalk caused by a single instance of capacitive coupling will increase with increasing frequency at a 6 dB per octave rate. That is, crosstalk due to single pole capacitive coupling will be 6 dB worse at 4 kHz than at 2 kHz, etc. Crosstalk can also be caused by inductive coupling, shared power supplies, shared ground returns, etc., so the relationship is often not the simple capacitive coupling model.
For traditional techniques, crosstalk is measured selectively. That is, a narrow bandpass filter in the analyzer is tuned to the generator frequency, in order to measure crosstalk at or below the wideband noise level. This is not merely an academic concern; the human ear is able to distinguish coherent signals such as sine waves even when the signal is 10 dB to 20 dB below the wideband noise level.
Crosstalk from “Channel X” to “Channel Y” is often not identical to crosstalk in the other direction, from “Channel Y” into “Channel X.” Circuit layout and complex stray coupling mechanisms often result in somewhat different values for crosstalk in the two directions. Good values for crosstalk in electronic devices range from 50 dB upwards, with crosstalk in excess of 100 dB achievable by sufficient isolation. When the application is stereo or surround sound, such values are far higher than really necessary for a full effect. However, if the two devices are carrying independent program material, values of crosstalk of 60–70 dB and more are desirable.