Recording - November 2012 - 72

attenuators, just reading the levels off the scale. This worked well for two reasons: First, the audio levels were considerably lower than the telephone levels the meters were intended for, and second, the audio people continued using 600 ohm lines. And this is where the “+4 dBm” standard comes from. +4 dBm is in fact 6 mW. If you put +4 dBm through a line and bridge a VU meter across it, it will read 0 VU. Consequently since it’s such a convenient number, +4 dBm became a standard operating level in the studio world. But (!), the reading on the VU meter (in Volume Units) is always 4 dB lower than the actual power reading in dBm through the cable. Why? Because the reference point for telephone circuits was a lot higher than one milliwatt and now we’re stuck with it. Is this crazy enough? Wait, it gets crazier! In the 1970s the audio industry made a transition from 600 ohm constant impedance lines (where the source and load were both 600 ohms and probably transformer-coupled) to

Guest editorial by Scott Dorsey

Taking A Historical VU
The VU meter is the most maligned, misunderstood, and misused measuring device in all of audio. It wasn’t designed to do most of what people use it for, and it actually doesn’t do many of the things people think it does. In the beginning, the world was without form and void and telephone circuits had levels that varied considerably. The folks at Bell Labs wanted to measure the perceived volume levels of voice signals on 600 ohm balanced telephone lines, and to do so they came up with a device to do it. Now, the original device consisted of two parts. The first part was a special attenuator that could be adjusted from 0 to 20 dB of attenuation. The second part was the VU meter. The VU meter was a meter that was intended to be bridged across a 600 ohm line (that means 600 ohm source, 600 ohm load, and the meter all in parallel with one another), marked off in decibel increments, and with very specific ballistics (more about that later). It was an average-reading meter and not designed to follow signal peaks. The meter was calibrated so that “0 VU” on the meter means 6 mW is flowing through a 600 ohm line when it’s bridged across it. This is important. 0 VU on the meter means 6 mW flowing through a 600 ohm transmission line. This was considered a very low signal level in the telephone world, about the lowest you’d want to measure. As originally used, the user would put a voice signal into the test set, then adjust the attenuator to drop the signal until the meter reads zero. Then they would read the level off on the attenuator set. The meter wasn’t intended to indicate level, the attenuator did that. The meter was used only as a reference to set the attenuator. But... fast forward a few decades and you see people using VU meters without
72
RECORDING November 2012

systems where low source impedances drove high load impedances. This was much cheaper to do in the solid-state world but it meant that all of our standards intended for 600 ohm lines were useless! The amount of power transmitted down the cable was dependent on the load impedance and so it no longer reflected the actual signal level. What to do? Why not change the standard again? So now we have a different standard, now we talk about voltages instead of powers... and, using the same old VU meter and the same old electronics, the old +4 dBm power measurement produces 1.288V across the 600 ohm load. So now, we whip up this new standard which we call +4 dBu, which has the same voltage but isn’t necessarily into a 600 ohm load— although, of course, we still need to have 600 ohm termination on an actual genuine VU meter. But wait! It gets crazier! The VU meter was originally designed to have very specific dynamics. If you apply 0 VU (+4 dBm) to the meter abruptly, the meter is supposed to take 300 ms ±10% in order to get from resting point (no signal) to 0 VU, and then when you remove power, it needs to take the same amount of time to drop. It cannot overshoot by more than 1.5% or less than 1%. This means that you can have a very high peak which is smoothed out by the meter and that you’ll never see. You could set a voice and a trumpet to read the same value on the meter, but when you digitize it and look at the peak reading meter on the digital converter, the trumpet is 20 dB higher! This is because the trumpet has a very spiky waveform with very narrow peaks that are much higher than the average level. It also means that genuine mechanical VU meters are precision instruments that are difficult to make. As a result, a lot of manufacturers of semipro gear and even some cheaper professional gear have skimped and used meters that did not have accurate ballistics and whose needles didn’t move precisely the way a VU meter needs to. You’ll see plenty of cassette decks with meters marked “VU” on them, but I have only seen a couple with actual VU metering... and ironically, with a tape recorder you probably want something that measures faster than a VU meter does! Furthering the irony, in the digital age it’s now possible to build inexpensive and accurate VU meters with digital displays... but the VU meter has become less important because everyone needs to know peak levels for digital recording. The VU meter is not peak reading and a peak-reading meter is not a VU meter. Do not confuse them or call them the same thing. Please. Perhaps it’s a pet peeve of mine, but they are two meters that don’t measure the same thing... because they were never designed to measure the same thing. Scott Dorsey (dorsey@recordingmag.com) suggests that readers who are really interested in metering should take a look at Audio Metering: Measurements, Standards and Practice by Eddy Bøgh Brixen, available from Focal Press.

Recording - November 2012

Table of Contents for the Digital Edition of Recording - November 2012