The fundamental difference between the Roland R-880 and many reverberators of its period was the freedom it gave the engineer to get beneath the factory presets. The system consisted of the rack-mounted R-880 processor and the separate GC-8 Graphic Controller. The processor itself was relatively restrained, with an analogue input level control, six bargraph meters, MIDI channel selection and little else requiring attention during normal operation. Most of the work took place on the GC-8, which was intended to sit beside the engineer at the console.
The GC-8 used a 135mm x 35mm backlit LCD accompanied by five continuously rotating soft controls, cursor keys, function buttons and a numerical keypad. What now seems an obvious approach to editing complex digital processors was far more significant in the late 1980s. Instead of trying to represent dozens of parameters with a small two-line display and endless menu scrolling, the R-880 could show signal paths, EQ curves, delay patterns, reverberation shapes and other processing information graphically.
One GC-8 could also control multiple R-880 processors, with the system capable of addressing as many as 16 units. In a large studio installation an engineer could therefore have several processors in the machine room while operating them from a single controller beside the console.
Serious processing hardware
Internally, the Roland R-880 used 16-bit A/D conversion, 18-bit D/A conversion and 28-bit internal signal processing. Dynamic range was greater than 90dB and total harmonic distortion was specified at less than 0.015 per cent. Analogue connections were provided on balanced XLR and unbalanced jack sockets, while digital connections included coaxial S/PDIF and optical interfaces. The R-880 could operate digitally at 44.1kHz or 48kHz, making it particularly relevant during the period when DAT was becoming established in professional studios.
There were two inputs and four outputs, but describing it merely as a two-input/four-output reverb understates what was happening internally. The processor could be configured as 1-in/2-out, 1-in/4-out, 2-in/2-out or 2-in/4-out, and the two processing paths could be treated independently. Its effects resources included two main reverberation units together with EQ, dynamics, delay, chorus and early reflection processing. Internal mixers allowed these components to be interconnected in different ways. This was where the R-880 became something considerably more interesting than a conventional digital reverb.
Building the effect
Calling up a factory program was straightforward enough, but the real power of the R-880 appeared when the Algorithm section was opened. The processor presented the component parts of an effect as elements that could be patched together. An engineer might route one input through chorus before feeding the reverb while sending another input directly to the reverberator through compression. Dry signals could be routed around processing sections, split, recombined and mixed with processed paths.
The two principal reverberation sections could be configured for Reverb, Plate or Non-Linear operation. Reverb and Plate could also use Stack or Tap configurations, providing further variation in the construction of the effect. There was sufficient flexibility that programming could be approached at several levels. A user wanting a quick result could modify the major parameters of an existing program. Someone prepared to spend considerably more time could rebuild the algorithm itself. That distinction was important. The R-880 did not force every engineer to become a programmer, but neither did it limit the more adventurous user to Roland’s idea of what a reverb should be.
Deep reverb control
The amount of control available over conventional reverberation was remarkable for the period. Reverb programs provided Hall, Room and Garage types, with room size variable from 0.2 to 80 metres and reverberation time stretching from 0.1 to 99.9 seconds. Early reflection level, brightness, density, high and low frequency damping and predelay could all be adjusted, with predelay extending to 800ms.
Early reflections could be manipulated in considerable detail. Twenty individual early reflection taps were available, with their level and delay adjustable either individually or in groups. This made it possible to do far more than simply select a generic room and change its decay time. The engineer could substantially reshape the way the simulated acoustic space developed around the original signal.
The Plate section offered its own selection of types and parameters, while the Non-Linear processor was capable of much more obviously artificial treatments. Gate times extended to 1200ms, while reverb time ran from -9.9 to +9.9 seconds. Negative values reversed the shape of the decay, creating backwards effects, and additional controls allowed the envelope of the reverb to be manipulated in sections. The GC-8 displayed the resulting level-versus-time shape graphically, making these more complex processes substantially easier to understand.
The sub-reverb
One particularly useful feature was the Sub Reverb. This allowed a second reverberation component to operate from the same input, with its own level, predelay and decay characteristics. It meant that layered ambience could be constructed within a single R-880 program rather than requiring another effects send and another external reverb.
Roger Nichols specifically found this useful for applications such as vocals, where two different reverberation characteristics could be combined while consuming only one console send. It is a good example of what made the R-880 unusual. Its complexity was not there merely to generate unusual effects. Much of it addressed very practical studio problems.
More than reverberation
The R-880’s additional processing sections were equally comprehensive. Its three-band parametric equalisers covered 20Hz to 20kHz, with each band offering 48 selectable frequencies and up to 12dB of boost or cut. The high and low bands could operate as peaking or shelving filters, while variable Q was available for peaking equalisation. Two delay units provided delays of up to 400ms in 1ms increments, together with positive or negative feedback and controllable output polarity.
There were also two chorus sections, each with independently variable rate and depth. Predelay extended to 40ms in 0.1ms increments, while the phase relationship between the modulation waveforms could be shifted through a complete 360 degrees. Dynamics processing could also be incorporated into an algorithm, making the R-880 a genuine multi-effects architecture rather than a reverberator with a few secondary effects bolted onto it.
Roger Nichols and the R-880
Perhaps the strongest endorsement of the R-880’s usefulness came from Roger Nichols. Nichols first encountered the machine during a demonstration at a Nashville studio. He was impressed enough to ask Roland whether he could try the R-880 on the album he was mixing at the time. Roland loaned him a unit for four days. He subsequently bought one.
By 1991 Nichols had owned his R-880 for approximately two years and was using it on every album he mixed. For an engineer working at that level, that is far more significant than simply being impressed during a demonstration. Nichols placed particular importance on transparency. His requirement from a reverberator was the ability to create convincing natural ambience around acoustic instruments. Once a processor could achieve that convincingly, more obviously processed effects were comparatively straightforward.
The factory programs provided some useful starting points, but Nichols soon concluded that the real potential of the R-880 would only be realised by creating his own programs. After becoming familiar with the programming system, he spent approximately a week entering his preferred reverberation sounds. Some could be created by modifying Roland’s existing programs, while others were built from the algorithm level upwards.
The level of detail was central to his enthusiasm for the machine. Room dimensions could be adjusted, early reflections constructed in detail, different chorus sections operated simultaneously and the digital parametric equalisation used to shape the resulting ambience. This also explains why the R-880 became such a useful working processor rather than simply an interesting technological experiment. Once an engineer had invested the time creating a personal library, the machine effectively contained a collection of reverbs constructed specifically for that engineer’s way of working.
Presets and memory
Roland supplied 46 factory presets, divided between internal memory and the system ROM card. Additional programs could be stored on 16K or 32K RAM cards of the same type used by Roland’s D-50 synthesiser. Capacity depended upon program complexity, with approximately 40 programs on a 16K card or 80 on a 32K card.
The system could therefore develop into a personalised effects library, but its late-1980s origins were obvious in the way the memory system operated. The system program had to be loaded from a card when the unit was started, and the internal clock and calendar needed resetting each time the system was powered up if date and time information was required for stored programs.
Program loading was not instantaneous either. Recalling a preset typically took around three seconds. None of this prevented professional use, but it illustrates the difference between the R-880 and the essentially instantaneous software environments that would eventually replace such hardware.
Programming and usability
Despite the depth of the R-880, its interface was remarkably logical once its structure was understood. Roland divided editing into Algorithm, Parameter, Mixer, Function and Memory areas. Graphical displays gave useful visual feedback, and some parameter pages were arranged in different levels of complexity. Large-scale changes could therefore be made first, with more detailed values recalculated automatically, before the engineer moved deeper into the program.
The system also warned before potentially destructive editing operations and provided information showing which parameter groups had been changed. It still demanded concentration. Building a sophisticated algorithm was not something an engineer would casually undertake during the final few minutes of a mix session. Nichols’ decision to spend around a week constructing his personal programs is revealing. Once that work was completed, however, the R-880 changed from a complicated programming environment into a personalised studio tool.
Sound quality
For all of its processing flexibility, the most important point is that the Roland R-880 could produce convincing reverberation. Its low noise floor was particularly notable, and its quoted frequency response extended from 20Hz to 20kHz. The analogue conversion and internal processing were technically strong for the period, but specifications alone do not explain the R-880’s reputation.
Its greatest strength was transparency. The machine could generate natural ambience without forcing the source to sound obviously processed, an attribute that was particularly important on acoustic instruments. At the same time, the enormous range of editable parameters meant that the same processor could produce highly artificial gated, reversed, chorused and multi-stage effects when required. That combination of natural ambience and deep manipulation gave the R-880 considerable versatility.
The GC-8’s little problem
The GC-8 was not completely silent. A faint tone around 400Hz could be produced by the circuitry powering the display illumination. Nichols had encountered it only once during two years of ownership, although the same basic problem was also noticed on other GC-8 units. It was a small flaw in an otherwise impressive controller, but one worth remembering given that the GC-8 was specifically intended to sit immediately beside the engineer.
There were other irritations. Some of the dynamics editing was less intuitive than the reverb programming, documentation could have explained certain areas more clearly, preset loading was relatively slow and the system demanded considerably more learning than a conventional preset-based reverb. None of those shortcomings fundamentally undermined what Roland had achieved.
