Read Rushton's post on this thread .
2 responses Add your response
The thread Doug mentions is a good start. I would add that a voltmeter alone is difficult to use as the crosstalk signal is very small. If you insist on a voltmeter, I would add a 1KHz bandpass filter (assuming you use a 1KHz signal on the test record) to avoid excessive broadband noise and measure crosstalk accurately.
A much better way IMO is a computer with a decent soundcard that you connect to the output of your phonostage. You can then use any spectrum analyzer software to measure the 1KHz signal and crosstalk in each channel directly and more accurately.
Now if you can also measure the relative phase between the main signal and crosstalk signal then you can successfully implement Chris Feickert's method for optimizing azimuth.