Tech Tip:Plugging In Plug-Ins
Where you plug in matters as much as what you plug in
Reprinted from harmonycentral.com with the permission of the author and publisher Craig Anderton
If your mixes don't make it, maybe you're not plugged in. Effects can further enhance good recordings: You can tweak your sounds so that they blend in better, stand out more, acquire more character, or whatever is needed to create the ultimate mix.
As with physical mixers, most host programs offer three places where you can insert plug-ins. Choosing the right location maximizes the effect's impact, conserves processor power, and allows the whole system to work at optimum efficiency.
First, though, let's investigate the two main types of plug-ins.
WHEN SOFTWARE MEETS HARDWARE
There are two main types of plug-in technology: host-based (also called native) and hardware-based. Host-based plug-ins use the computer's microprocessor to do digital signal processing, and therefore require no specialized hardware. Hardware-based plug-ins run only with certain pieces of hardware designed for digital signal processing, such as Digidesign's TDM-based DSP farm, the UAD-1 PCI card from Universal Audio, the PowerCore PCI card and PowerCore Firewire external box from TC Works, and CreamWare's various Pulsar family PCI cards.
Native plug-ins drain a certain amount of CPU power, so the more plug-ins you run (especially software synthesizers), the harder the CPU has to work. To stretch the limits as to how many plug-ins you can use with a software program, use a faster CPU, or increase the system latency (the time required for the system to process signals). Increasing latency gives the CPU some slack, but increases the response time when tweaking controls, responding to external control signals, etc.
Plug-ins based on cards have performance limitations as well, but they leave the CPU free to perform other tasks. The more powerful the card, the more effects you can run on it.
In either case, most programs have a "CPU meter" that shows how much power is being used, as do most cards).
The VST Performance meter shows what's happening with Cubase SX; Sonar puts readouts for CPU and hard disk performance below the main track window. CreamWare's PowerPulsar and Universal Audio's UAD-1 have their own separate resource meters.
Compare the readings before and after loading a plug-in to check its requirements. However, note that host-based loads are "subject to change without notice" — leave some power in reserve should the computer need to perform additional, unanticipated tasks. With DSP boards, though, you can pretty much red-line the performance. Once a plug-in is loaded into the board's DSP, there will be few, if any, significant additional demands placed on the DSP.
In hardware mixers, channel inserts interrupt the signal between a channel's input preamp and fader/panpot circuitry. Hard disk recording programs follow the same concept: Inserted effects appear within a specific channel, and affect only the channel into which they are inserted. Popular insert effects include dynamics control, distortion, delay, chorusing, and flanging, although almost any effect is a candidate.
Multiple effects are usually added in series (e.g., one effect's output feeds the next effect's input). There may be a fixed number of slots for effects, or you may be able to add as many as your computer can handle.
Technically, EQ is also an insert effect. In some hard disk recorders, EQ needs to be inserted into a channel; others "hard-wire" it into each channel for convenience, just like hardware mixers.
Master effects patch in series like insert effects, but they insert in the mixer's master output section, after the separate channels have been mixed to a stereo signal. Therefore, master effects process the entire mix, not individual tracks. Master effects should have stereo ins and outs; mono effects may not work, even if you're mixing in mono.
One issue with master effects is whether they appear before or after any master gain control. Post-master fader effects slots allow an effect to continue even after the main level has been turned down (for example, a long, repeating echo or long reverb "tail" that fades into silence). Otherwise, effects before the fader are turned down with the master.
Typically, the very last slot is where dithering gets added to the signal chain. Dithering converts a program's high-resolution audio (e.g., 32-bit floating point, 24-bit) to a lower resolution, typically 16 bits for Red Book CD compatibility. If dithering goes before the fader, then the fader — which operates at the higher resolution — would "undo" the dithering's conversion process. The same is true if it goes before other effects.
The most commonly used master effects involve equalization and/or dynamics (particularly multi-band compressors, loudness maximizers, limiters, and sometimes stereo compressors). Equalizers applied to program material tend to have fairly broad, gentle curves, , as gentler EQ curves tend to result in a more natural sound with program material.
If you use any master effects, be careful not to overload the master channel output. With 32- or 64-bit floating point programs channel headroom is pretty generous, but this may not be the case at the final stage.
CAUTION: STEREO/MONO MISMATCHES
Programs have definite rules about stereo and mono effects. For example, stereo channels prefer effects with stereo inputs; a mono effect may process only the left track, the right track, or sum the two tracks together — or it may be incompatible with stereo tracks.
Mono channels can often use stereo effects, but usually only one of the channel outputs will be present. Consult the host program's documentation to discover any particular constraints.
Mixers, whether hardware or software, have virtual buses. Each acts as a "terminal" for signals flowing through the various mixer channels. In most applications, the most important of these buses is the stereo master output bus. All channel outputs dump into this bus for a stereo mixdown.
In addition to output buses, you will also find auxiliary (sometimes called send) buses. Any channel can send some signal to one or more of these aux buses; this signal can come from either before the channel fader (pre-fader, therefore, the level going to the bus doesn't change when the channel volume changes), or post-fader, where bringing down the fader also brings down the amount going to the aux bus.
This diagram shows the relationship of Cubase's various buses. The lower arrow represents the master bus, which mixes together the signal from each channel and sends the combined output to the master channel. The other arrows represent 8 aux buses, whose outputs feed into the master channel (which can also have its own set of master effects) The send controls for each channel determine how much of the channel signal gets sent to a particular bus.
One potential "gotcha" with the pre/post dichotomy involves muting. If you select pre-fader send and mute the channel, the pre-fader signal will likely not be muted (verify this with your program's documentation, or by trying it out to see what happens). As muting is the same as pulling a fader all the way down, muting a track mutes the send effect signal only if the send effect is post-fader.
There's a good reason for using aux effects rather than insert effects under some circumstances. Every plug-in requires some amount of computing power, so the fewer plug-ins you use, the more processing "headroom" is available. Therefore, if you want to process several channels with the same effect, you're better off loading a single aux effect instead of inserting the same effect multiple times as a channel insert effect. For instance, when adding reverb, it makes more sense to use it as a send effect, synthesize the desired acoustic space, and mix in varying amounts of signal from the send bus, as determined by each channel's send effect level control.
Aux buses generally include a master send level. This is very handy when setting send levels. Suppose you start turning up more and more sends on the mixer, and the send effect input overloads; simply turn down the master send to obtain the correct level rather than turning down a bunch of individual send controls. Or, suppose you decide that the overall reverb amount needs to be louder or softer, but you don't want to re-adjust each send control because the balance is already perfect. Use the aux effects master instead.
There also may be a bus select option, which determines where the effects bus output terminates. For a standard stereo mix, this would be the master bus. But you might want to send the output to a separate bus, such as one feeding a headphone monitoring system, or a conventional hardware mixer if you want to do hands-on mixing.
Note that send effects are generally set for processed (wet) sound only. This is because the channel fader provides the unprocessed signal to the master bus, while the send effects contributes only the effect sound, adding it to the unprocessed signal. In fact, it's best not to include any straight sound in a send effect because of possible latency issues within the effect. Although most modern programs include Path Delay Compensation (PDC) to "tune out" any of these differences, combining the straight signal from the channel along with a straight signal contributed by the send effect with some older programs could lead to phase cancellation issues.
When working with aux effects, there are three places to alter levels:
The channel's aux send control.
The main channel fader, if the send effect control is set to post-fader.
The master send control, which regulates the overall level going to the send effect.
Furthermore, any signal processors the send effects bus feeds may have input and/or output level controls, and their sound may depend on the incoming level. If these controls aren't set correctly, overly hot levels may cause distortion, while too low a level throws away dynamic range. Here's the general procedure for proper level-setting:
If the aux effect has an input level control, set it to unity gain (i.e., the signal is neither amplified nor attenuated).
Set the master aux send level control to unity gain.
Adjust the individual send controls for the desired amount of effect. The higher you turn the individual send controls, the more that channel will contribute to the processed sound.
As the sends from the individual channels start to add up, they may overload the processor's input. Leave the processor input at unity gain, and use the master send control to reduce the level going to the effect.
If the signal going to the processor is too low, then use the master send control to bring it up. If turning it up leads to distortion and there still isn't enough level, increase the processor's input level as needed.
Okay, now you're plugged in. Start processing!
© 2001-2006 Craig Anderton and licensed to Harmony Central, LLC. All rights reserved. Harmony Central encourages linking from other sites to Harmony Central content.