Interfacing The Line (Part 1)
What is 'line level'? What significance do the numbers -10, 0, +4 and +8 have to recording? What does MOL refer to? Plug in to page 48 and find out.
This month Ben Duncan sifts through the spaghetti at the hack of the rack, unravels the level relationships between consenting equipment and identifies some oriental arch-villains.
What is line level? The word 'line' refers to any length of cable - implying that the length itself is uncritical, so conceptually, it's a connection following on from a low source impedance. Line level is also broadly synonymous with 'zero level', and line connections are those between powered equipment - namely, tape machines, mixers, effects units and amplifiers. The category of line connections emphatically excludes the transducers at the signal's ultimate source (microphone, guitar pickup, etc) and destination (tape, disc cutterhead, speakers or headphones).
Naturally, it helps if all equipment in a studio interconnects at a standard level. Whenever this happy state of affairs can be arranged, the standard level chosen is known as the line level, and is subtly distinguished from zero level in so far as the latter is a critical calibration above which distortion or damage threatens, while line level is more of a nominal figure.
Figure 1 shows five standard line levels in common use, while Figure 2 puts these into perspective against noise and headroom. The differences between the middle three (0, +2.5, +4) are small and usually readily overcome or ignored, but clearly, equipment designed to operate at the +4dBu or +8dBu levels used in broadcasting and PA will get very hostile when confronted by budget gear set to operate at -10dBu. That represents nearly a ten-fold difference in levels which will result in distortion.
Looking at Figure 2, do you see how the headroom in the -10dBu column falls sharply compared with the other columns? By the same token, certain signal processors primarily designed for the broadcast market may be working at less than their optimum signal-to-noise ratio when driven from the -10dBu levels of typical home studio equipment. Additionally, a power amplifier with a +8dBu sensitivity cannot be driven to full power by a -10dBu line level signal without running that line feed into distortion.
At this stage there are three big questions to be asked:
1. Why so many different standard levels?
2. Which one should I use?
3. How is line level defined when music levels vary?
To a great extent, all three questions are tied together like chicken and egg. The archetypal line level, derived way back in the 1920s when Fostex was merely a glint in some marketing man's eye, is 0dBu, which happens to equal 775 millivolts. If you recall from HSR November 'Decibel Games', 775mV is a nice tidy figure when you're driving a 600 ohm load, for it corresponds to a power level of one milliwatt. The old operating level of one volt was also handy because it was comfortably above the hum and hiss of primitive valve equipment (by around 25dB!).
The +4dBu level arose because the very first (1939) VU meters suffered a 4dB level loss, courtesy of the isolation and rectification elements within the original design. With the VU meter reading 0VU on its display, there was +4dBu across the line. Convenient? Perhaps, but it really is confusing for beginners.
Meanwhile, -10dBu line level derives from the limited headroom or MOL (Maximum Operating Level) of budget electronics. MOL is the maximum possible level a signal can register before overload or clipping arises, and is tied in with the line level. How? Well, remembering that any uncompressed music signal contains short term peaks (transients) some 10, 15 or even 18dB above the average level (as measured on a VU meter), the MOL needs to be at least 18dB above the line level. That gives us a +22dBu MOL for a +4dBu line level, but only +8dBu MOL for the budget -10dBu operating level. The significance of this is economic, in that a +8dBu maximum output is obtainable from a small, cheap low voltage power source such as a 9 volt battery.
MOL is a well-defined parameter, being primarily dependent on the supply voltage inside equipment, which is apt to be fairly standard. You can check the MOL on most gear by measuring the supply rails, then reading off the specification from Figure 3, which displays typical MOL values for a wide range of supply voltages. Simple unbalanced circuits working from a single 9 volt supply (a PP3 battery) such as effects pedals, typically have a +9dBu MOL, which is a comfortable 19dBu below overload when operating at —10dBu levels.
Oriental electronics has done great things for musicians and music over the past ten years, sure - but it has also brought headaches to those who adopted the established pro-audio line levels of 0dBu or +4dBu. The -10dBu level is simply a necessary economy when the electronics budget is sliced to the bone. Column 4 of Figure 3 indicates one saving grace, however. The headroom can effectively be boosted by 6dB on budget level processors if balanced outputs are employed, making interfacing with pro level equipment a lot less hassle.
Ultimately, the discrepancies between the various line levels results in two possible outcomes:
1. Insufficient level - leading to unwanted noise, or...
2. Excess level, manifested as overload.
In any specific instance , either of these can be quickly corrected by the insertion of a line amplifier or line attenuator (pad) respectively. However, if all 12, 16 or 24 channels of your mixing desk need level compensation, this can prove a tedious, not to mention expensive, task to undertake.
The answer is to sketch out a 'gain structure map' to arrive at the most satisfactory operating level for your particular recording setup. Draw a series of boxes or triangles pointing left-to-right to represent all of the units in the signal chain. Indicate the level loss or gain of each unit or stage, and also mark the various peak and average operating levels to check what happens. It's a particularly handy method for those who believe that a picture is worth a thousand words.
Figure 4 features a series of three abstract gain structure maps of outboard effects chains. Each brings to light an overload error, which might easily cause you to tear out your hair if it arose inadvertently.
As a rule-of-thumb, "gain is a pain" in all processing or effects chains. Okay, a few dB of make-up gain is handy to compensate for small losses, but in general, equipment with a fixed positive gain ie. anything more than 0dB, really is a nuisance. Figure 4a shows why this is so. The figure above the line represents the signal's peak level, the one below, its average level as read on a VU meter.
In this instance, the input of the delay unit is fed in at a very reasonable +4dBu. However, this particular American device has an overall gain of 3dB between its input and output. Next in line, there's an equaliser (EQ) boosting the midrange by 5dB. The resultant level is then brought back down, but only by means of the mixing desk's auxiliary return fader.
Looking at the figures, notice how the average level at the output is sailing close to the wind with an MOL of +24dBu, whilst the signal peaks are trying to swing the output to +26dBu - some 2dB higher than the MOL! The result is a clipped midrange sound, due entirely to the gain boost of the delay unit. This would not have been the case, however, if the delay's input were connected for unbalanced operation. Moral: differences between Balanced and Unbalanced connections can progressively upset the gain structure in an effects chain, causing distortion.
Second in the rogues' gallery is Figure 4b. Here, a compressor (with 0dB in/out gain) feeds a well known budget noise gate. This gate is excellent value for money but prefers to be driven at the -10dBu line level because it runs from an 18 volt supply and overloads at +12dBu. As before, the input signal to the first unit (the compressor) is +4dBu. The peaks, of course, are larger at +19dBu, but the compressor takes care of these, knocking them down to +10dBu, which is still some 2dB below the MOL of our wonderful noise gate (+12dBu). If the compressor is accidentally switched out of circuit though, the +19dBu peaks of the input signal would be presented to the noise gate whose maximum operating level (MOL) remember, is only +12dBu. Result: heavy distortion.
Lastly, Figure 4c displays how badly placed gain controls can conceal potential distortion. The first two units have excess gain of +3dB and +6dB respectively, plus the output of the second unit is also 6dB into overload on peaks (24dB output, 18dB MOL = +6dB excess), but this doesn't show up via the meters on the mixer's auxiliary return because the second effects device (reverb) incorporates an ill-placed attenuator. Dropping the aux return fader -10dB sets the line return level at 3dBu (within 1dB of the send level), so there's no obvious defect save for the odd distorted signal peak.
So? We'll now try observing the peak-to-average level ratio. The average input level is +4dBu, while the peak is +15, giving us a ratio of 15:4 which equals +11dBu. In this situation any extension in the signal dynamics will dramatically increase the distortion. Obvious? Yes, but ironically the problem could well come from lowering the actual sound level at the instrument as there'd be less compression at the microphone and at any limiters/compressors further down the line.
No examples of 'underload' are offered as signal levels which are too low merely produce more system noise or hiss than need be. Gain structure allows us to determine the highest and safest line level, knowing the headroom available. Reducing a +4dBu line level to 0dBu in one fell swoop merely to overcome one or two small areas of excess gain is no real solution. A much better route is to check level variations around the system with a portable peak-reading LED meter or PPM.
Also called 'pads', attenuators are widely used to reduce excessive levels, especially in line level circuits. Passive attenuators comprising two or three resistors work well, provided they don't upset the loading on the equipment. Active attenuators require battery or mains power but do away with matching worries by isolating the impedance of the offending pad resistors from the surrounding equipment.
Figure. 5 gives a circuit for an unbalanced, passive pad. The values provided in the table have been chosen to present a healthy 10K 'bridging' load which suits the majority of UK and US professional and semi-pro gear. Budget Japanese gear prefers to see a 50K load impedance or higher. Amending the table to cope, we can simply multiply all resistor values by ten. This is fine for the source equipment, which now sees around 50K, but it has also raised the source impedance defined by resistor LAR. For the low value attenuation, LAR now approaches the value of the load impedance of the succeeding equipment in the chain, producing more attenuation than bargained for - around 9dB instead of 5dB. This inaccuracy is no reason to abandon passive pads completely, but it helps if you can contrive your own set of values tailored to give a known degree of attenuation.
We'll explore line level impedances and balanced pads in more depth next time.
Feature by Ben Duncan