Eventide's classic SP2016 reverb unit has been recreated and enhanced by the original designer. But how does the sound match up to that of the high-spec multi-effects units of today?
Although most reading this will be aware of Eventide and its range of sophisticated digital effects processors, few — if any — will have come across Princeton Digital before. This new company was founded a little over two years ago by Tony Agnello — but over 25 years ago he was the Eventide engineer who developed the original algorithms for the still highly regarded SP2016 reverb processor, as well as the software in the original Harmonizer. The foundation research began in 1978, and the original SP2016 — one of the first digital reverbs — was launched in 1983. It quickly became very popular, partly because of the naturalness of its reverb programs, but also because it was so controllable and immediate — qualities sadly absent from so many modern hardware reverbs.
|Photos: Mark Ewing|
The SP2016 — a 3U rackmount box with vertical faders to adjust the parameters, chunky '80s buttons, and a small LED alphanumeric display — was fairly crude by modern standards, but three of its algorithms have stood the test of time: Stereo Room (with source-related stereo early reflections), Room Reverb (with a mono input) and High Density Plate. These reverb programs all sounded far more natural than any other reverb technology could manage at the time (plates, springs, tape echoes, and tiled rooms) and they always seemed to sit well in mixes. Furthermore, all their parameters were instantly adjustable, which was revolutionary. The SP2016 reverbs sounded quite distinctive, but they were also very believable thanks to the way the unit generated a surprisingly complex early-reflection pattern, complete with a realistic build up of reverb density over time, combined with a smooth and natural decay tail.
Agnello formed Princeton Digital to apply the advances in DSP technology to the reproduction of high-quality, modern versions of some of those 'legendary' Eventide effects. The first results of this work are now available both as a bespoke hardware unit — the topic of this review — and as a software plug-in.
The original SP2016 digital reverb ran on a purpose-built 'array processor' — the state of the art twenty years ago. This approach wasn't unique to Eventide; DSPs as we know them today were simply not available, and microprocessors were incredibly basic compared to those available now. Consequently all of the early digital reverb manufacturers — including AMS (the Ambience program on the RMX16 is still one of the best) and Klark Teknik (the DN780 remains one of my favourite reverbs) — had to adopt variations on gate array hardware techniques to achieve the sophisticated digital processing required.
Although it may seem surprising, current DSP technology has only recently reached a level where it can match the computational complexity of those early bespoke arrays. So the Reverb 2016 employs a 120MHz DSP to do the number crunching, making it possible not only to recreate the SP2016's original algorithms, but also to take them a stage further, beyond the limits imposed by the original array-processor hardware. As a result, the new Eventide Reverb 2016 contains not only the original algorithms, but also new, more sophisticated versions of each of them. The 'enhanced' facilities include the improving of the Diffusion parameter — it was rather limited and subtle on the original, and is now far more expansive and useful. There is also provision of Diffusion and EQ controls to the Room and Plate algorithms, and the Position control is active in the Plate algorithm too. The high EQ response has also been tweaked slightly to give a brighter sound.
Princeton Digital have also produced a software-only plug-in version of the Reverb 2016 for the Pro Tools HD platform. Not surprisingly, the plug-in is also called Reverb 2016 and provides a rather stylish graphical user interface which replicates the look and feel of the hardware unit. The reverb algorithms are also apparently identical, which explains why it is also pretty DSP hungry! A free fully-functional 30-day demo version is available from the Princeton Digital web site and the full version can also be purchased on-line.
The Reverb 2016 is a 1U rackmount box, with black-painted metalwork throughout, including the front panel. It measures roughly 205mm behind the rack ears, making it an easy fit (in terms of depth) into a flight case. However, the main part of the box measures about 449mm in width — the standard is about 435mm — and that means it won't actually fit in many racks or cases because there simply isn't sufficient space between the rack strips. This really is a case of try before you buy, because you'll be very frustrated if it won't slot into that rack space you had in mind!
The rear panel is straightforward but well equipped. As the 2016 is capable of true stereo-in/stereo-out processing using certain algorithms, it is provided with two channels of inputs and outputs. The analogue I/O is catered for with both XLRs and TRS sockets, switched globally for balanced +4dBu or unbalanced -10dBV operation via a rear-panel button. The converters are standard 24-bit designs, but the sampling rate is not standard. The original SP2016 ran at 40kHz — typical of the technology available in the era when it was conceived — and the new version remains faithful to that. As a result, the frequency response of the processed signal is limited to 16kHz (±1dB) at the high end. The dynamic range is quoted as greater than 98dB ('A' weighted) and distortion is below 0.006 percent. The D-A output is capable of driving a peak output level of +24dBu.
It is interesting to note that, although the digital processing and sample rate (as already mentioned) are identical to those of the original, the analogue interfaces are radically different, particularly in respect of the A-D and D-A stages. The originals were crude 16-bit devices with early brick-wall anti-alias and reconstruction filters which mangled the high-end response with amplitude errors and horrendous phase shifts. The new design uses modern 24-bit oversampling filters which provide a far better dynamic range and a smoother, more open high-end sound character.
A pair of phono (RCA) connectors provides co-axial S/PDIF digital audio in and out, and a pair of MIDI sockets offer In and Out ports. The last socket is another TRS type which allows an optional momentary-action footswitch to either 'kill' the reverb input (in other words stopping the reverb send signal, but allowing the existing decay to continue) or bypass the unit completely. Just to finish off the rear panel, on the extreme left (as viewed from the back) is an IEC mains socket for the integral switched-mode power supply (SMPS) which accepts any mains voltage between 90V and 240V, at 50Hz or 60Hz. There are no user-accessible fuses, and the unit consumes 45W of power. The mains power switch is on the front panel at the extreme right-hand side.
The front panel is plain and simple, but arguably the more attractive for it. A pair of simple input-level LED bar-graph meters occupy the leftmost part of the control panel, showing signal levels above -40, -20, -6 and 0dBFS, with an additional pair of lights to indicate overload of the A-D converters. A pair of simple black rotary controls set the analogue input and output levels, and a yellow LED adjacent to the input control illuminates when mono algorithms are in use. In this case, the two input signals are internally summed to mono before being passed to the reverb processor. Running above the meters and these control knobs are three black buttons, each with an associated indicator LED. The first selects the S/PDIF digital input, while the others provide manual control of the features operated by the optional footswitch — the reverb input 'kill' and bypass functions.
Using The Rotary Encoders
Most of the other front-panel controls are rotary encoders, all but one with a ring of LEDs around the base to show the current setting. These provide real-time control of the wet/dry mix (zero to 100 percent), pre-delay time (0-999ms), decay time (maximum varies with program), notional distance from the source (which affects pre-delay, diffusion, early reflections, frequency response and more besides!), diffusion, and high and low EQ. All of the encoders have an over-press facility, and in all but the two EQ knobs, pushing the encoder causes its numerical value to be shown for about five seconds on the LED display. (The numerical value is also shown while the knob is being adjusted.)
The LED displays are different for each parameter, but selected to show the relevant setting as clearly as possible. For example, the Mix control is shown as a solid block of light from the zero position at seven o'clock around to the current Mix level. The Pre-delay is shown with a dual pointer, one part circling faster than the other. The first shows blocks of 30ms, while the other steps for single-millisecond increments. A nice operational feature here is that, by pressing and turning the knob simultaneously, the pre-delay is adjusted in 30mS steps, while simply turning the knob normally allows 1ms step precision. Both the control operation and display are unusual, and take a little getting used to, but the combination provides great precision over a huge range of values.
The Decay control value is shown in the same way as that of the Mix control (using a solid block of lights), while the Position control shows its setting with a single line. Diffusion is shown as another block of lights, but this time building outwards symmetrically to either side of the 12 o'clock position.
The EQ controls are slightly unusual. The low band provides between +4dB and -8dB of gain centred on any one of ten frequencies between 50Hz and 500Hz (in 50Hz increments). The high band only provides attenuation (up to -8dB) at frequencies between 1kHz and 8kHz (in 500Hz increments). For both controls the gain is adjusted by rotating the knob in the usual way, while the turnover frequency is adjusted by pressing in the knob while turning it. This action is easy enough with the unit mounted in a rack, but the pressure required is such that you can easily end up pushing the box all around the table if it is left free standing! It is also worth noting that applying too much low boost with a long decay time can cause runaway feedback within some programs.
To the left of the numerical display window is a column of four LEDs and a button used to select the reverb algorithm. Stereo Reverb programs are shown by a red LED, Room Reverb programs are yellow, High Density Plates are green, and the three new algorithms (stereo, room and plate again) are shown in combination with a blue LED. Pressing the button repeatedly simply cycles through the modes. Under the three-digit LED display window is another pair of buttons, each with its own indicator LEDs. The first accesses the System menu — of which more in a moment — and the second is used to save the current settings to a preset memory. The last encoder, mounted to the right of the display window, enables presets to be selected for storing or recall.
There are 99 preset memories, all initially loaded with factory programs, and any can be overwritten with user settings if required. Presets 0-59 are all configured with no dry signal at all, whereas presets 60-99 are identical programs with an appropriate wet/dry mix. The presets are organised in groups of ten effects, based on each of the six algorithms (the three originals and the three new variations). The effects in each group gradually increase in scale from small to large environments. Pressing the encoder knob loads the selected preset, at which point the indicators around all the parameter encoders change to reflect the new settings. Any program can then be further adjusted by manipulating the relevant encoder. To save a preset you simply have to press the Save button, dial up the required preset number, and press Save again.
The System button mentioned earlier provides access to the machine's configuration parameters. Pressing the button repeatedly causes the display to cycle through the various options, starting with the input source (both analogue and digital). Options here are input one, input two, or both — enabling the use of a single-source signal even for the stereo programs. Next is the MIDI channel (Off, 1-16, or All). The next two modes are labelled dC (Dump Current) and dA (Dump All), which provide a MIDI data dump of the currently selected program or of all programs. The Save button also flashes while in this mode, and if it's pressed the data is output via MIDI to an external data filer or sequencer, for example.
The sound of the Reverb 2016 is pretty magical. There is something about the way this neat box generates reverb — and more importantly, the way that its reverbs sit in a mix — that makes this a great machine. To be honest, the SP2016 was not a processor I spent much time with in my early audio career — I'm more familiar with the AMS and Klark Teknik reverbs from the same era. Nevertheless, I can certainly appreciate the quality of these algorithms, and the benefits afforded by Eventide having to build the hardware necessary to implement them. Pretty much all subsequent digital reverberators worked the other way around — algorithms had to be created (or compromised) to run within the available hardware structure, at least until affordable DSP power reached a certain level, after which point there was an explosion in the number and sophistication of really good-sounding digital reverbs.
The Reverb 2016 is certainly very intuitive to use — when I first opened the box I thought the manual had been omitted, so just started playing with the unit and found it very easy to master. Only later did I discover that the two sheets of A4 notes tucked in with the machine really are all that is needed to fully understand it. Compare that to the 100-page booklets that usually accompany Yamaha, TC Electronic or Lexicon reverbs!
The controls feel perfect, the displays are simple but clear and informative, and the presets all provide useful starting points. Having said that, this box is so easy to fine-tune by ear that, after an initial survey of the presets, I didn't use them again! I preferred instead to quickly recall the appropriate algorithm and dial in the necessary parameters. Unless you can remember the specific presets appropriate to a particular task, it's just as quick to configure the unit manually, and a lot more fun!
It seemed a little odd that some of the parameter controls were blank and inoperative for certain algorithms — until, that is, I read the notes, which explained that the original machine didn't include EQ or Diffusion controls for the Room and Plate algorithms. I guess that's the whole point about faithful recreations of existing products. However, I was equally impressed that these facilities are provided when using the 'New' versions of the algorithms — the turbo-charged updates — which also have other subtle tweaks and refinements to ensure the reverb quality and versatility matches the best of the current products in the market.
The High Density Plate algorithm is probably my favourite, although the Stereo and (mono) Room programs are also very good and extremely usable in the appropriate contexts. I found that recalling some presets occasionally resulted in nothing useful at the output — just an unstable kind of whistling noise. Often changing the preset and then going back to the required one again cured the problem, which seemed completely unpredictable. Whether this is part of the faithful recreation (there are several warnings about inherent algorithmic instabilities) when changing presets in the presence of audio signals, or a bug in the current operating software, I don't know. However, although a little frustrating at first, it became largely irrelevant when dialling in parameters from scratch.
Of all the parameter controls, the Position knob was probably the most heavily used parameter for all the programs apart from the original Plate algorithm (where it is not available). This control really does simulate moving closer to, or farther from, the source, changing diffusion, early reflections, wet/dry balance, EQ and so on. It's just the job when everything else seems well balanced but its all just a bit too close or distant!
The New variants of the original algorithms retain most of their inherent character, but add slightly more controllability and a subtle extra richness or reduced coarseness, depending on your point of view. In other words, it could be said they sound a little more modern than their classic siblings. However, both versions of all three algorithms were eminently usable across a wide variety of music genres, always seeming to sit perfectly in the mix rather than on it, bringing space and perspective without acting like acoustic Polyfilla — a common failing of lesser reverbs. The original algorithms could, at times, reveal their '80s heritage with a characteristic coarseness (especially if auditioned in isolation during a solo-in-place check, for example), but this rarely became noticeable within a mix. Maybe there's something here for other reverb manufacturers to take heed off. The real-world precision of modern high-spec reverbs and convolution systems may not always be the most musically appropriate choice, especially for pop mixes.
The Reverb 2016 is expensive when compared to the glut of very high-quality and versatile reverbs and multi-effects processors now available in the UK. However, it does offer a unique sound quality that works wonderfully in the context of a mix, and is fantastically ergonomic to use. If the budget can stand it, this is definitely worthy of close personal attention — but measure your rack space first to avoid disappointment!