TC Works' PowerCore is a PCI card incorporating a Motorola PowerPC CPU and four Motorola 56362 digital signal processors. The PowerCore hardware comes bundled with PowerCore 1.5, a suite of six digital signal processing (DSP) effects plug-ins, and a mono software synthesizer modeled after the famous Roland SH-101. The included effects are primarily aimed at pro-quality mixing and mastering, freeing your CPU to run more specialized, native effects. Several third-party providers — including Antares, Sony, and Waldorf — have announced upcoming PowerCore versions of their effects.
You need a reasonably modern computer that accommodates full-size PCI cards to use PowerCore, but you don't need a tremendously fast CPU — the whole point of PowerCore is to extend the audio-processing capacity of your system. PowerCore is not an audio interface; it simply passes audio back and forth to your plug-in host application, which can be any software that supports VST or MAS plug-ins. (MAS support is Mac only. The Macintosh version includes TC Works' Spark LE sample-editing software, which can be used as a host.) Audio quality depends on your system and audio interface, but PowerCore's processing remains entirely in the digital domain and supports full 24-bit, 96 kHz systems.
For this review, I installed the PowerCore card in a Macintosh G3/300 MHz and used Emagic's Logic Audio Platinum 4.8.1 and TC Works' Spark XL 2.02 as VST hosts. Although the minimum requirements specify OS 9.0, I had no problems using PowerCore with versions 8.6 and 9.1.
HOW IT WORKS
Once you have the PowerCore card and its software drivers installed, your only interaction with PowerCore is through the PowerCore plug-ins, which are installed like any other VST or MAS plug-ins. Also, like other plug-ins, you can copy the PowerCore plug-ins for use in different host applications.
After a cold start of your computer, launching any host application and inserting a PowerCore plug-in initializes the PowerCore card, as evidenced by a fuel gauge that indicates the 20-second lag during which the PowerCore starts up. Thereafter, until you power down your computer, the operation of the card is mostly transparent, and the plug-ins operate just like native plug-ins that run on your computer's CPU. I say ‒mostly” because there are several differences of which you need to be aware.
Even though the PowerCore hardware is doing the heavy lifting, there is some drain on your computer's CPU as it performs the processing necessary to manage the audio transfer to and from the PowerCore hardware. The PowerCore 1.5 drivers, which run on your computer's CPU, have been greatly improved to minimize that drain, but it is still apparent. One thing that affects CPU drain is the audio-buffer size. That is usually set in your host application's I/O preferences, and larger buffer sizes result in lower CPU drain.
Shifting data to and from the PowerCore hardware also takes time, as does the PowerCore DSP. That can cause audio tracks using plug-ins to be slightly delayed relative to unprocessed tracks, and unfortunately, the delay gets worse as the audio-buffer size is increased. Because latency can also be a problem with native effects, many host applications have a built-in delay-compensation option. For cases in which that is not available or satisfactory, TC Works provides a native plug-in called TC Compensator that can be used to delay all unprocessed tracks. Being native, TC Compensator uses no PowerCore resources but adds slightly to your computer's CPU load.
Finally, all PowerCore plug-ins have a No Latency mode. In No Latency mode, data passed from the audio buffer is processed immediately, forcing your CPU to communicate continuously with the PowerCore hardware. Needless to say, that adds considerably to CPU drain, but it is ideal for situations, such as live performance, in which latency is less tolerable. In short, the latency problem can be addressed in several ways, each with its own advantages and disadvantages. PowerCore does a lot of the work, but there is no free lunch.
The way your system determines how many plug-ins can be used is also different for PowerCore than for native plug-ins. In the native environment, you add plug-ins until your CPU chokes. On the PowerCore card, each plug-in takes a fixed amount of one of PowerCore's four DSP chips. For example, a single DSP chip can run two MegaReverbs, three EQsats, or four stereo or six mono Vintage CL compressor/limiters. (You can mix plug-ins on a single DSP chip.) The allocation of plug-ins to the four DSP chips is managed automatically, but it is necessary to be aware of the total plug-in count and to monitor native CPU usage (for which most host applications provide visual gauges).
To arrive at a rough benchmark, I compared the CPU usage for eight tracks of audio with no plug-ins (15 percent) with that for eight PowerCore plug-ins (25 percent). The eight PowerCore plug-ins used up the four DSP chips on the PowerCore card. No native plug-ins exactly match the quality and functionality of the PowerCore plug-ins, but eight native plug-ins of similar function required a 75 percent CPU load.
POWERCORE 1.5 PLUG-INS
PowerCore's software bundle of six effects and a mono synthesizer includes a trio of mixing and mastering effects known collectively as TC Tools. This set contains a high-end reverb (MegaReverb), a multipurpose delay (Chorus/Delay), and an EQ (EQsat) with high and low shelving and three parametric bands. The other three effects are dynamics processors with individual features designed for specific tasks: VoiceStrip for vocal processing, Vintage CL for emulating analog compression, and MasterX3, a 3-band compressor/limiter/expander designed primarily for mastering.
TC Tools of the trade
TC Tools was the original PowerCore bundle, and it covers the standard bread-and-butter mixing jobs. One DSP chip can handle two reverbs, two delays, or three EQs. Using all four chips, you can have four complete TC Tool kits. The obvious omission is compression, and PowerCore 1.5 includes three dynamics processors.
MegaReverb (see Fig. 1) is a true stereo reverb with flexible room-design, time, filtering, and mixing controls. It sounds as good as any I've used, and given the amount of flexibility it provides, it's surprisingly easy to set up. By tweaking the presets (all of the usual candidates are offered), I was always able to achieve the sound I wanted very quickly.
Room characteristics are managed using graphic windows labeled Shape, Size, and Wall Diffusion. Six shapes are provided: Hall, Horseshoe, Prism, Fan, Club, and Small — each incorporating the mathematical characteristics of a specific room. (Fan, for example, is based on the structure of La Scala, the famous Italian opera house.) Each room can be scaled from 0.04 to 4 times in size. The room's reflective surfaces (diffusion) can be scaled from -50 percent (tile and brick) to +50 percent (cottage cheese) with 0 representing the original room's diffusion characteristics. The room design can be heard most clearly in the early-reflection portion of the reverb.
Prefiltering, pre- and postdelay, and the reverb tail are controlled by the three graphic windows at the right in Fig. 1. You use the Highcut Filter window to set the cutoff and slope of a stereo lowpass filter that comes first in the signal path. The Decay/Frequency window controls the decay times of the reverb tail in three frequency bands. There is independent control of the bandwidth and decay time for each frequency band.
The Predelays window sets both the time (ranging from 0 to 160 ms) before the early reflections begin and the time (from 0 to 100 ms) between that and the beginning of the reverb tail. That is also where the early-reflections and reverb-tail levels are set (over a range of 0 to -• dB).
The mixing controls include input and output level, wet/dry mix, and separate left-right balance for the early reflections and the reverb tail. You can also set the stereo width from mono to full stereo, and for each channel, you can separately control whether the early reflections are crossfed into the other channel.
Chorus/Delay is another good-sounding, easy-to-set-up plug-in. Its graphics have the same look and features as MegaReverb, and starting from the presets, you can tweak your way to victory in short order. It is somewhat limited in its off-the-wall-effects potential, but for general feedback-delay purposes (chorus, flanging, echo, and so forth), it is fully capable.
Like MegaReverb, Chorus/Delay starts with a lowpass filter, but in this case, the raw signal can be routed around it. That is what you want when you use the chorus as an insert effect, because it allows the source signal to be heard without any processing. In the context of the delay effect, you may or may not want the raw signal in the mix. The chorus section is a short delay line (40 ms maximum) with an LFO for modulating delay time. In addition to delay time, the Chorus/Delay has controls for LFO speed and depth, LFO phase between the right and left stereo channels, and feedback amount and filtering. The chorus section can produce anything from standard chorus and flanging to highly resonant feedback effects. Its one limitation is that only a sine LFO waveform is provided, and that can't be synchronized to MIDI tempo.
The delay section features longer delays (as long as 500 ms), and conveniently, the delay time can be set in beats per minute (120 bpm, for example, is the equivalent of 500 ms, or one beat). Unfortunately, like the chorus, the delay time cannot be synchronized to MIDI tempo.
The delay parameters are as simple as it gets — delay time, feedback-filter shape, and feedback amount. The feedback filter is the same for the delay and chorus sections. It is a bandpass filter for the signal being fed back to the input. You can set the filter's upper and lower band limits independently. (In other words, you can control the bandwidth.) Chorus/Delay will fulfill all the requirements of a basic feedback-delay line, but the absence of separate right- and left-channel delay times with cross feedback does limit its range of effects.
One of the more interesting uses for Chorus/Delay is as a dual resonator. Because both the chorus and delay sections are capable of very short delay times and the feedback filters can pick out different frequency bands for feedback, you can resonate two separate bands of the input signal. MP3 example_1, at www.emusician.com, illustrates that effect on a bell-like sound effect.
The third TC Tool, EQsat, is a 5-band equalizer with shelving filters on each end and three bands of parametric equalization in the middle. Actually, each section's frequency is free ranging, which lets any of the bands overlap. The parametric bands are bell shaped with a gain range of ±18 dB and selectable bandwidth from 0.10 to 4 octaves. The shelves can slope from 3 to 12 dB per octave, up or down. A soft-saturation (nonlinear distortion) stage after each filter segment adds warmth by simulating analog-tube distortion.
PowerCore 1.5 comes with three dynamics-processor plug-ins. CL, the most basic, emulates an analog compressor/limiter and features soft- or hard-knee compression and full-range limiting with optional soft saturation. That is PowerCore's workhorse compressor, and you can get six mono or four stereo CLs on one DSP chip.
VoiceStrip is, as its name implies, an all-in-one voice-processing plug-in (see Fig. 2). It starts with a low-cut filter followed by a noise gate, which is in turn followed by equalization and compression stages. The switch that is provided for reversing the order of EQ and compression is a nice touch. VoiceStrip's last stage is a de-esser. VoiceStrip is truly the Swiss Army knife of voice processing, and there are lots of knobs and switches to deal with. But the controls for each section are straightforward, and the provided presets cover all the bases.
Having multiple VoiceStrip modules in your kit — you can fit two of them on each DSP chip — gives you a lot of flexibility in multitrack voice mixes. For example, slightly different EQ and compression settings are often all you need to bring several mismatched background tracks into line. MP3 example_2 illustrates another approach to using multiple VoiceStrips, exaggerating different characteristics in a multipart laugh track.
The last plug-in, MasterX3, is a 3band expander/compressor/limiter based on TC Works' renowned Finalizer hardware device (see Fig. 3). It was a temporary part of the PowerCore 1.5 bundle and, as of March 2002, is sold separately for $249. (A 5-band version is also available for $499.) So by the time you read this, MasterX3 may not be in the packages you find on your music dealer's shelf.
MasterX3 is intended primarily as a mastering tool and includes optional output dithering at anywhere from 8 to 22 bits. The limiter circuit features look-ahead delay and the option to lower the output ceiling by as much as 0.1 dB. Using the Frequency/Levels graphic at the top center of Fig. 3, you can make adjustments to the crossover frequency and gain for the three bands extremely quickly.
Although MasterX3 operates on three frequency bands, it uses a clever system of Target Curves to simplify the compressor, limiter, and expander settings. Each of those modules has a single set of parameters that are then scaled to fit the Target Curve for the three frequency bands. Four Target Curves are provided: Linear, Pink (less high-band processing), Hyped (more high-band processing), and Smiley (more low- and high-band processing). The amount of scaling for each process (compression, limiting, and expansion) is controlled by a Target Factor setting for that process. It sounds complicated, but is a lot simpler than setting three complete sets of compression, limiting, and expansion parameters. MP3 example_3 applies three mastering settings during the course of a short dance mix.
Taking the lead
For good measure and, presumably, to amuse you when you get tired of tweaking effects settings, TC Works has added a lead synth to the PowerCore 1.5 bundle. PowerCore 01 is modeled after the Roland SH-101 and is TC Works' first VST Instrument. Fig. 4 shows PowerCore 01 in one of its four skins.
PowerCore 01 is a monophonic synth with a single mixed-waveform oscillator, a resonant lowpass filter, an LFO with square or triangle waveshape, and a single ADSR envelope generator for amplitude control. The filter cutoff frequency can be modulated by a mix of the LFO, envelope generator, and MIDI note pitch. For leads and basses, PowerCore 01 offers a lot of bite, a classic analog-synth sound, and very low latency, especially when run in No Latency mode. (It was playable without No Latency mode enabled.) You can get 3 synths on a DSP chip, so if you want to dedicate your PowerCore card to it, you can have 12 synths running at once.
GET PLUGGED IN
The PowerCore hardware will substantially expand the audio-processing capacity of your computer, and the PowerCore 1.5 software bundle contains great-sounding, top-quality effects in all of the standard categories. With the TC Works offerings growing and third-party PowerCore plug-ins on the way, it is likely that more esoteric processing applications will appear in the future. But even without that, getting the load of everyday processing off your CPU frees up processing power for your favorite native plug-ins.
At $1,299, the PowerCore will put a dent in your budget, but considering that several professional-quality native plug-in bundles cost as much, the price seems well justified. If you do a lot of multitrack processing with plug-ins in the categories included in PowerCore 1.5, it is definitely worthy of serious consideration.
Minimum System Requirements
MAC: G3/233 (with PCI slots); 128 MB RAM; OS 9.0; compatible VST or MAS host application
PC: Pentium III/300; 128 MB RAM; Windows 98/2000/ME/XP; compatible VST host application
PowerCore 1.5 (Mac/Win)
plug-in accelerator card and software
FEATURES4.0EASE OF USE4.5QUALITY OF SOUNDS4.5VALUE4.0
RATING PRODUCTS FROM 1 TO 5
PROS: Top-quality effects plug-ins in all the basic categories. Greatly reduces CPU load.
CONS: Limited range of compatible plug-ins currently available.