This series of articles was originally published in 2021. It was very well read at the time and has continued to draw visitors, so we are re-publishing it for those who may have missed it first time around.

There are 11 articles in the series:

The first audio vectorscopes were based on oscilloscope technology and used cathode ray tubes. The scanning mechanism of the tube was arranged as shown in Fig.1 such that the left audio signal caused the beam to be deflected along an axis 45 degrees anticlockwise from the vertical whereas the right channel causes deflection along a line 45 degrees clockwise from the vertical.

Later vectorscopes replaced the CRT with a flat screen but presented the same display using suitable signal processing on the inputs. Given that audio waveforms are bipolar, the light spot would move equal distances above and below the center of the screen and the bottom half of the screen didn't contain any more information than was already present in the top half.

Some vectorscopes took advantage of that by suppressing the bottom half of the display and placing the origin near the bottom center of the screen.

As stereophony results in mapping sound sources over 360 degrees into the space between the speakers, we are more interested in how the image between the speakers will appear than what the original sound image was.

Vectorscopes are intended to work with coincident microphones. If used with binaural (dummy head) signals, the vectorscope cannot accurately point to dominant sound sources because of phase differences between the channels that are inherent to binaural signals.

Applying identical test tones to both left and right channels should result in an upright central trace. If there is an amplitude difference, the trace will lean towards the louder signal. In modern digital equipment amplitude errors are unlikely, whereas with traditional analog equipment the possibility of drift is always there.

If one of the inputs is reversed in polarity, the display will become a horizontal line. Feeding a test tone into a pan pot should allow the line on the display to be rotated from -45 degrees to +45 degrees. The same linear result should be obtained by feeding a real audio signal into a pan pot, except that the length of the line will change with level. The pan pot simulates a zero-width sound source.

These ideal line displays on a vectorscope are easily obtained with test signals or pan pots, but from actual stereo microphones they are most unlikely as a number of mechanisms combine to prevent the ideal. We would need anechoic conditions, a sound source of zero width and perfectly coincident microphones of ideal directivity followed by electronics having precisely matched frequency response.

In real life, that isn't going to happen. Practically all real sound sources are distributed, especially musical instruments. All live sound venues have reflections and reverberation. We want to reproduce both to obtain realism.

Stereophonic sound is supposed to reproduce a sonic image, and like all images the result can be good or poor. In other imaging applications, from astronomy through photography, television and so on, there are tests that can be made to assess imaging accuracy and a poor test result can be used to reject or improve things.

It is therefore something of an omission that no standardized test exists for the accuracy of a stereophonic image. The result is that the majority of stereo images one hears are not as good as they could be. Defective equipment and inappropriate techniques continue in use because there is no objective measure of the results.

The audio vectorscope can be used to identify shortcomings in stereo systems and intelligent use will help to improve image quality.

If a coincident pair of crossed-8 microphones is considered, along with a central sound source, that source will be 45 degrees off axis to both microphones, which should then produce near identical signals in anechoic conditions. The result will be a vertical display on the vectorscope with a relatively narrow trace.

However, if the sound source is then moved off center, it will be closer to the axis of one microphone and further from the axis of the other one. With ideal microphones, the result would be that the vectorscope trace simply rotates to the new direction. With real microphones, we discover something in addition. The directivity of microphones is never independent of frequency, and this affects the imaging. The level difference between low frequencies and the level difference between high frequencies is not the same.

That level difference is responsible for the position of the virtual image, so poor microphone directivity causes the virtual image to spread or smear. The result is the same as with an out-of-focus camera. There is less information in the image and small details will be missed. The results can be seen on the vectorscope.

The criteria for stereo microphones are more stringent than for single microphones used with pan pots. The questionable directivity of a single large diaphragm microphone may not matter a whole lot if the vocalist stays in front of it, whereas two such microphones used as a coincident pair will cause smear. In general, the smaller the diaphragms of a stereo microphone, the better the imaging will be.

Microphone smear will only be detected with adequate monitoring, and this is not guaranteed. Loudspeakers intended to be mounted vertically are often put on their sides for appearance reasons, with the result that the various drive units subtend different angles to the listener, causing frequency dependent image smear that will disguise any microphone deficiency and render the system useless for any critical monitoring purpose.

The linear traces on a vectorscope can only be obtained when the left and right signals are in-phase and differ only in amplitude. Phase differences between left and right signals results in the vectorscope drawing circles and ellipses. In a reverberant environment, the time of flight of reflections will be greater than that of the direct sound, so there will be some phase shifts when the direct and reverberant sound is added. The thickening of the vectorscope trace gives an indication of the degree of reverberation.

One of the fascinating features of the audio vectorscope is the way that the display graphically points to the dominant sound sources in front of the microphones, giving an immediate confirmation that the image is correct.

If that doesn't happen, there are several possibilities. The microphones may not be positioned correctly, there may be inter-channel gain differences or they may not have been set to the correct directivity. Microphones often have a hard life and in the case of a variable directivity microphone, it is not unknown for an internal failure to make the actual polar diagram different from what is set on the control.

It is a good idea to test microphones before use to see that their directivity actually is what it is supposed to be. A rough check requires no more than turning the microphone in front of a sound source to see if the output rises and falls according to expectations.

Another way to obtain value from an audio vectorscope is to use it to compare different microphone techniques. Fig.2a) shows a pair of spaced omni-directional microphones exposed to an arbitrarily chosen frequency from some direction. It will be evident that there is a significant phase difference between the signals but only a small amplitude difference.

Fig.2b) shows the result with a different frequency. The amplitude difference is the same, but the phase difference has changed. The vectorscope will depict these phase dfferences as circles or ellipses. With real sound sources the vectorscope display may resemble a ball of knitting wool. The differing phase relationships will put the virtual image between the speakers in different places. Spaced microphones do not and cannot reproduce a stereophonic image because the position in the virtual sound stage is a function of frequency.

Fig.3a) shows the effect of adding left and right signals from spaced microphones to make mono. The result is a comb filter where the delay is the path length difference of the two microphones. That should be compared with Fig.3b) in which a path length difference due to a reflection turns a single microphone into a comb filter. Fig.3b) is considered poor practice whereas spaced omni-directional microphones are not. The physics are the same.

Some spaced microphone techniques add a central forward-facing microphone that is fed equally into the left and right channels. The result is a central image from the central microphone, but the rest of the sound stage is filled with frequency dependent smear that gives the virtual sound image spatial attributes which fail to have anything to do with the placing of the original sound sources. Such tree configurations give better mono compatibility.

Perhaps the most telling comparison between coincident and spaced microphones is to record a cocktail party in which a large number of conversations are going on simultaneously. With good loudspeakers, the listener can use attentional selectivity to listen to the conversations one by one with the coincident recording. With the spaced microphone recording it is not possible.