20 August 2026
Mitsubishi’s Digital Recording Revolution
The limits of analogue recording
By the middle of the 1970s, professional analogue tape recording had reached an extremely high level of refinement. Better magnetic tape formulations, improved heads, more accurate transports, noise reduction systems and increasingly sophisticated studio maintenance had all extended its performance.
However, analogue tape still imposed unavoidable limitations. Every recording, overdub, bounce and copy introduced additional noise and distortion. Variations in the magnetic coating could produce modulation noise, while the non-linear behaviour of the tape caused harmonic distortion. Wow and flutter affected pitch stability, print-through could allow loud signals to become faintly audible elsewhere on the tape, and repeated generations progressively moved the recording further away from the original signal.
Mitsubishi’s engineers saw pulse-code modulation as a way of separating the recorded information from these physical imperfections. Rather than storing the continuously changing audio waveform directly as magnetisation, a PCM recorder measured the waveform at regular intervals and expressed each measurement as a binary number.
The recorded pulses could still be affected by tape wear, dirt, imperfect head contact or transport errors, but moderate changes to their physical shape did not alter the numerical information they represented. Provided the recorder could distinguish between a zero and a one, the audio data could be reconstructed without the gradual degradation associated with analogue recording. Performance was therefore determined less by the magnetic characteristics of the tape and more by the sampling system, word length, conversion circuitry, clock accuracy and error-correction design.
By the middle of the 1970s, professional analogue tape recording had reached an extremely high level of refinement.
Mitsubishi’s early PCM development
Mitsubishi Electric’s professional digital programme was already producing results by September 1976, when the company completed a PCM tape deck operating on 6.3 mm tape at 38.1 cm per second. Mitsubishi described it as the first PCM tape deck of its kind. In October 1977, the company developed a PCM recording system intended for amateur use that stored its signal through a video recorder. It also demonstrated an early digital audio disc system. Development continued towards a self-contained domestic PCM recorder, a semiconductor-laser disc player and a professional fixed-head two-channel recorder.
The professional machine was demonstrated at the Audio Engineering Society convention in New York during November 1979. It would emerge commercially as the Mitsubishi X-80, one of the earliest practical open-reel digital mastering recorders. This early start gave Mitsubishi valuable experience before professional digital recording became a major commercial market. The company was not simply adapting an existing analogue recorder to carry digital information. It was developing the tape format, error correction, servo system, editing method and studio workflow together.
Why Mitsubishi chose fixed heads
Early digital audio recorders generally followed one of two approaches. Some used rotating heads and video-style transports, while others used stationary heads arranged across conventional open-reel tape. Rotating-head recorders could achieve very high recording densities with relatively few electronic circuits. They could also be made comparatively compact. However, they were less suited to familiar studio operations such as simultaneous synchronised playback, rapid duplication and physical tape editing.
Mitsubishi chose the fixed-head approach for its principal professional systems. The tape travelled past a stationary multitrack head assembly in much the same way as it did on an analogue recorder. This allowed the machines to behave like professional audio tape recorders rather than modified video recorders. They could provide separate record and replay heads, confidence monitoring, synchronised playback, punch-in recording, fast digital copying and, in the case of the stereo machines, razor-blade editing.
The compromise was complexity. Digital information had to be divided across numerous narrow tracks, requiring extensive conversion, modulation, timing and error-correction electronics. The resulting machines were large and expensive, but they offered an operating style that professional engineers already understood.
The Mitsubishi X-80
The Mitsubishi X-80 was a two-channel digital master recorder using 6.3 mm high-density tape running at 38.1 cm per second. Its audio was sampled at approximately 50.4 kHz and encoded as 16-bit linear PCM. The sampling rate was high enough to provide an audio bandwidth of approximately 20 kHz. Contemporary specifications quoted a frequency response from 10 Hz to 20 kHz within ±0.5 dB, dynamic range better than 90 dB, total distortion of 0.02 per cent at peak level and crosstalk of approximately −85 dB.
The tape carried ten longitudinal tracks. Six contained the encoded stereo audio data, two carried parity information, one provided an analogue cue signal and one carried SMPTE timecode. The analogue cue track was important because the digital audio could only be reproduced within a limited range around the normal tape speed. During editing, the engineer listened to the cue track while manually moving the tape to locate the required point.
The X-80’s most remarkable feature was its ability to accommodate physical cut-and-splice editing. Editing digital data with a razor blade might appear inherently unreliable, but Mitsubishi designed the format around this requirement. After locating the edit through the analogue cue track, the engineer marked the tape with a suitable video marker and made a straight vertical cut using the supplied editing block. The two pieces were joined using thin video splicing tape designed to preserve close contact between the tape and heads.
A splice inevitably interrupted the digital data. Mitsubishi dealt with this by interleaving the samples before recording and using a powerful combination of Reed-Solomon parity information and cyclic redundancy checks. The damaged data surrounding the splice was detected and the recorder automatically crossfaded across the interruption, preventing an audible click. Thick splicing tape could lift the recording tape away from the heads and increase the period of data loss, so the type and application of the joining tape were critical. Nevertheless, a correctly made splice could be reproduced without an audible discontinuity.
The capstan servo also had to remain locked when it encountered a physical join. Mitsubishi used a phase-selection system that generated alternative synchronisation signals separated by 120 degrees. If the phase changed abruptly at the splice, the recorder selected the signal that produced the smallest phase jump, helping the transport remain synchronised. This combination of digital recording and analogue-style editing made the X-80 unusually accessible to engineers accustomed to working with conventional stereo master tape. It preserved the physical editing techniques of the analogue studio while removing most of the noise, print-through, crosstalk and generational degradation associated with analogue copying.
The professional machine was demonstrated at the Audio Engineering Society convention in New York during November 1979. It would emerge commercially as the Mitsubishi X-80, one of the earliest practical open-reel digital mastering recorders.
Tom Jung became an enthusiastic advocate for the Mitsubishi X-80 during the formative years of his Digital Music Products label. Its most celebrated outing was Flim & The BB’s’ 1983 album Tricycle, recorded live to stereo and subsequently established as an international audiophile reference recording.
From stereo mastering to digital multitrack
Mitsubishi’s larger ambition was a complete professional digital audio system. Its proposed chain included a multichannel recorder, a two-channel master recorder, digital editing equipment, a digital mixing console and a digital delay for disc mastering. The centrepiece of the multitrack system was the Mitsubishi X-800. It recorded 32 audio channels onto 25.4 mm tape running at 76.2 cm per second.
Rather than combine several audio channels into a single tape track, Mitsubishi assigned each audio channel its own digital track. This was essential for professional multitrack work because individual channels needed to be placed into and taken out of record independently. The 32 audio channels were divided into four groups of eight. Each group used ten physical tracks, consisting of eight audio data tracks and two Reed-Solomon parity tracks. The complete recording therefore required 40 digital tracks for 32 audio channels.
Each audio channel also carried its own cyclic redundancy check information. The Reed-Solomon coding operated across the tape, while the cyclic redundancy check operated along the direction of travel. Mitsubishi described this as a two-dimensional error-correction system. The arrangement addressed a specific problem found in high-density fixed-head recorders. A slightly misaligned or worn head assembly did not necessarily cause all tracks to deteriorate evenly. Instead, one or two tracks could become substantially weaker than the others.
With two parity tracks protecting each group of eight audio tracks, the recorder could reconstruct missing information even if an individual track failed. It could continue recording, replaying and performing punch-ins on the affected audio channel while using information recovered from the remaining tracks and parity data.
The centrepiece of the multitrack system was the Mitsubishi X-800. It recorded 32 audio channels onto 25.4 mm tape running at 76.2 cm per second.
The Mitsubishi PD tape format
The recording system developed for the Mitsubishi machines became known as the Professional Digital, or PD, format. It was subsequently marketed as ProDigi and was supported by Mitsubishi, Otari and AEG. The format was designed around four practical requirements. It needed to use tape efficiently, withstand the relatively high error rates associated with dense digital recording, support professional editing and punch-in operations, and remain reliable when tapes were exchanged between machines.
Mitsubishi’s research showed that errors most commonly occurred as bursts running in the direction of tape travel. Simultaneous errors across several adjacent tracks were much less common. Two-track failures were far rarer than single-track errors, while simultaneous failures affecting three or more tracks were considered extremely unusual. This error pattern led Mitsubishi to arrange its correction codes as rectangular blocks. Reed-Solomon coding protected the information across the width of the tape, while cyclic redundancy checks detected errors along each individual track.
The PD system was related to a Generalised Product Code. It combined some of the efficiency of concatenated coding with a more practical level of hardware complexity. Short Reed-Solomon code sections could be processed without the very large amount of circuitry that a longer code might require. Digital audio samples were also extensively interleaved. Consecutive samples were distributed across different physical positions on the tape so that a scratch, crease, patch of dust or damaged splice would not normally destroy a continuous section of the programme. After reproduction, the samples were returned to their original order.
When damaged information remained within the correction capability of the code, the original sample was reconstructed exactly. When there was too much damage for correction, the recorder used concealment. The principal concealment method was first-order interpolation. The machine examined the valid sample immediately before the error and the valid sample immediately after it, then substituted an average value. A small number of concealed samples was generally inaudible.
Mitsubishi also avoided dependence on a dedicated control or clock track. A separate clock track could have become a single point of failure. If that one track became clogged, worn or damaged, every audio channel could be affected. Instead, the PD machines used self-clocking modulation. Timing and servo information could be extracted from the digital audio tracks themselves. The transport would lose its clock only under conditions severe enough to prevent the reproduction of every available digital track.
Punch-in recording without clicks
Independent track recording created another difficulty. When one or more audio channels were punched into record, the parity information protecting the corresponding eight-channel group also had to be updated. The recorder therefore reconstructed its coding blocks around the new audio. Mitsubishi used a hierarchy of main blocks and sub-blocks so that the machine could identify the correct data positions even when the record and replay heads were not aligned with absolute mathematical perfection.
The incoming audio could also be crossfaded with the existing playback data before the revised code was written to tape. This avoided the abrupt transition that would otherwise occur when a digital channel entered or left record. On later machines, these crossfades made punch-ins remarkably unobtrusive. Testing of the Mitsubishi X-850 found that entering or leaving record was free from unwanted clicks. Even replacing a 100 Hz signal with another signal 180 degrees out of phase was difficult to identify in the reproduced waveform and was reported as inaudible.
The Otari DTR-900 32-Track Digital Recorder was compatible with the PD (Professional Digital) format developed with Mitsubishi.
The arrival of Mitsubishi ProDigi
The production PD format established a family of compatible professional machines. Two-channel recorders used 6.3 mm tape, 16-channel machines used 12.7 mm tape and 32-channel machines used 25.4 mm tape. The standard production format used 16-bit linear PCM at sampling rates of 44.1 kHz or 48 kHz. The multitrack machines normally operated at 76.2 cm per second, while the stereo machines generally ran at 38.1 cm per second.
ProDigi competed directly with the DASH format supported by Sony, Studer and other manufacturers. Both used stationary heads and open-reel tape, but their tape formats and coding systems were incompatible. This incompatibility became one of the defining realities of early professional digital recording. Studios, record companies and equipment rental businesses had to decide which format to support, or maintain access to both. Digital interfaces eventually made transfers between different systems easier, but physical tape compatibility remained limited to machines using the same format.
The Mitsubishi X-850
The Mitsubishi X-850 represented a major refinement of the company’s 32-channel system. It retained the basic ProDigi arrangement of 32 audio channels on 25.4 mm tape but improved the transport, control, synchronisation and editing facilities. The machine used 40 tracks for its digital audio and parity information, together with two auxiliary analogue tracks, two auxiliary digital tracks and an SMPTE timecode track. Four 45-track ferrite heads provided sync replay, erase, record and read-after-write monitoring.
Its transport was based on the established Otari MTR-90 design, modified for 25.4 mm digital tape and Mitsubishi’s head assembly. The pinch-rollerless transport could accommodate 356 mm NAB reels. With 27.5 µm tape running at 76.2 cm per second, a reel provided approximately one hour of recording. The X-850 was physically substantial. Its electronics occupied large card frames behind front access doors, while several cooling fans were required to remove heat from the conversion, coding and transport circuitry. The machine was mounted on wheels and fitted with screw-down stabilising feet.
The 32 channels were divided into four groups of eight. Each group had its own analogue input circuitry, analogue-to-digital converters, digital-to-analogue converters, filtering, coding, interleaving, modulation and replay electronics. A front-panel digital routing section, informally known as the ping-pong panel, allowed channels to be copied internally in the digital domain. A signal could be transferred from one channel to another without returning it to analogue form. Connections were also provided for transfers between the X-850 and Mitsubishi stereo recorders.
The remote control and autolocator were essential parts of the recording system. Each channel could be placed in ready, record, input or tape status, while four complete audio-channel setups could be stored and recalled. The locator provided 100 time memories, automatic marking of start, stop, record-in and record-out events, repeated playback between stored positions, rollback and automatic punch-in and punch-out. It also included a timecode generator and could display either tape time or SMPTE timecode.
Varispeed was available over a nominal range of ±10 per cent, although laboratory testing found that the actual increments did not correspond precisely with the displayed 0.1 per cent steps. The X-850 could synchronise with additional audio machines and video systems operating at 25 or 30 frames per second, including drop-frame and non-drop-frame standards. The format could therefore be used for large music projects, television production, film work and synchronised 64-channel recording.
The Mitsubishi X-850 used different control software from the earlier X-800, but it could reproduce unspliced X-800 tapes. Physical splices made in the earlier format presented compatibility problems.
Measured performance of the Mitsubishi X-850
Contemporary laboratory testing confirmed that the X-850 delivered performance well beyond that normally possible from an analogue multitrack recorder. Input and output levels were matched to within 0.03 dB at 1 kHz. Frequency response extended from approximately 10 to 12 Hz at the −1 dB point to just over 20.4 kHz, depending on the channel tested.
At higher recording levels, second and third harmonic distortion remained below 0.02 per cent. Intermodulation performance was similarly low, while measured crosstalk was approximately −85 dB under deliberately demanding test conditions. The unweighted signal-to-noise ratio from 22 Hz to 22 kHz measured 91.8 dB, increasing to approximately 93.5 dB when A-weighted. Differences between channels were generally less than 1 dB.
Time alignment between channels was better than 500 nanoseconds. This was important because the X-850 shared some conversion circuitry between pairs of channels. The measured timing difference was sufficiently small to avoid meaningful phase or mono-compatibility problems. The error-correction system also performed effectively with worn tape. Individual track-error indicators could flash repeatedly without causing interpolation, muting or an audible fault.
The ProDigi coding could tolerate the loss of two tracks within a ten-track channel group without corrupting the audio. Through its interleaving and concealment system, it could also survive larger areas of tape damage before the output needed to be muted. If the error rate became too severe, the X-850 faded the audio smoothly rather than cutting it off abruptly. The measured mute and recovery transitions were approximately 10 milliseconds, reducing the chance of an audible click.
The practical difficulties of physical editing
Although ProDigi was designed to support razor-blade editing, handling the thin 25.4 mm tape was not easy. The analogue auxiliary tracks allowed edit points to be found by listening while the tape was moved manually. Large rollers near the head assembly provided fine mechanical control, making precise cueing possible without rotating the reels themselves.
The actual splice required considerable care. The two tape sections had to remain accurately aligned, the joining tape needed to be trimmed within the recording tape’s edges, and the machine worked best when a gap of approximately 1 mm remained at the join. A correct splice could be inaudible. A poor splice could cause a mute lasting around 60 milliseconds or more, depending on the extent of the lost data.
Unlike an analogue edit, a badly made digital splice could be extremely difficult to repair because the damaged data pattern was not visible. Backup copies were therefore essential. The X-850 also provided no clear advance warning that a tape was approaching the end of its reliable life. Analogue tape generally deteriorated progressively and audibly, while digital tape could continue sounding normal until its error-correction margin was exhausted and failure became sudden.
These concerns did not undermine the machine’s recorded performance, but they demonstrated that physical digital editing demanded greater precision than conventional analogue splicing.
The X-850 delivered performance well beyond that normally possible from an analogue multitrack recorder.
The Mitsubishi X-86 Series
The Mitsubishi X-86 Series replaced the original X-80 concept with a two-channel mastering system designed around the production ProDigi format. The standard X-86 operated at 44.1 kHz or 48 kHz. Its tape carried six PCM audio tracks, two parity tracks, two cue tracks, one timecode track and one auxiliary digital track.
Unlike the analogue cue track used by the X-80, the X-86 cue tracks used pulse-width modulation. The system provided audio for locating edit points while retaining the narrow track width required by the format. The X-86 provided cut-and-splice editing, optional electronic editing, insert recording, ±10 per cent varispeed, autolocation, timecode display and interfaces for RS-232C, RS-422, SMPTE, EBU and ESbus control.
It also incorporated an AES/EBU digital interface, allowing audio to be transferred between compatible equipment without conversion back to analogue. External clock and composite video synchronisation inputs helped integrate the recorder into music, broadcast and post-production systems. The use of 356 mm reels extended the standard recording time to approximately two hours. Mitsubishi also produced the X-86LT, which operated at half the normal tape speed and extended the recording time to approximately four hours.
The X-86HS and 96 kHz recording
The Mitsubishi X-86HS took the fixed-head stereo format into sample rates that were highly unusual for a commercial recorder of its period. It could record at 88.2 kHz or 96 kHz, extending its frequency response beyond 30 kHz. It could also operate at the conventional X-86 rates of 44.1 kHz and 48 kHz.
An internal sampling-rate converter allowed high-rate recordings to be transferred to 44.1 kHz or 48 kHz machines. The AES/EBU interface meant that the destination recorder did not have to be another Mitsubishi machine. This gave engineers the option of recording and processing at the higher rate before preparing a master compatible with Compact Disc, broadcast or other established digital systems.
Mitsubishi openly acknowledged that the X-86HS was expensive even by professional digital standards. Its advertising positioned the machine for studios, engineers and producers who considered the extended sampling rate and bandwidth sufficient justification for the cost. The X-86HS was being promoted as part of the X-86 range by late 1987 and continued to appear in Mitsubishi advertising around the introduction of the X-880.
The compatibility-focused X-86C
The Mitsubishi X-86C addressed a growing archival and interchange problem. By the end of the 1980s, mastering houses and rental companies could encounter recordings made on the original X-80 as well as later X-86 machines. The X-86C was designed to reproduce the established X-86 formats while also providing compatibility with tapes recorded on the X-80.
This made it particularly useful for facilities that needed to handle material originating from several generations of Mitsubishi fixed-head stereo recorders. Mitsubishi promoted the X-86C as the two-channel machine capable of providing compatibility with most Mitsubishi and related fixed-head digital mastering tapes then in professional use.
The Mitsubishi X-86 Series replaced the original X-80 concept with a two-channel mastering system designed around the production ProDigi format.
The third-generation Mitsubishi X-880
The Mitsubishi X-880 was the company’s third major generation of 32-channel digital multitrack recorder.
Large-scale integrated circuitry allowed Mitsubishi to reduce the amount of electronics required. The X-880 was smaller, lighter, quieter and more energy efficient than the X-850 while retaining tape compatibility with its predecessor.
The machine included an improved autolocator and could be fitted with a sample-accurate chase synchroniser for locking two recorders into a 64-channel system.
Compatibility with the X-850 was commercially important. Studios could add a newer machine without abandoning their existing library of ProDigi multitrack tapes, while rental companies could supply X-850 and X-880 machines for the same projects.
Mitsubishi was advertising the X-880 as its new third-generation multitrack by the middle of 1989.
Commercial success in major studios
Mitsubishi’s professional digital machines achieved particularly strong acceptance in major American and British recording centres. A contemporary 1986 report stated that Music Mill Studios had purchased two Mitsubishi X-850 recorders and that 12 digital multitrack machines had been installed in the Nashville area during the previous two years. This made Nashville one of the most concentrated markets for professional digital multitrack recording in the United States.
British trade coverage from the same period associated the X-850 with work involving Peter Gabriel, Pete Townshend, Tears for Fears and Bryan Ferry. Hilton Sound, which had been selected as Mitsubishi’s principal British rental agent, acquired a second X-850 as well as an X-400 16-channel recorder and two X-86 stereo machines. Rental availability was an important part of the format’s success. A studio did not necessarily need to purchase several extremely expensive digital machines permanently. Additional X-850 recorders could be brought in for large projects, live recording or synchronised 64-channel sessions.
Mitsubishi’s advertising argued that the machines attracted producers who specifically requested digital recording, allowed studios to charge higher rates and provided greater productivity through reliable operation, digital copying and automated punch-in facilities. These were marketing claims, but the continuing installation of X-850 and X-880 systems in leading recording centres demonstrates that the format had become far more than an engineering experiment.
A bridge between two recording eras
Mitsubishi’s greatest achievement was not simply that its machines recorded audio digitally. Their importance came from the way they combined digital data with the working practices of the analogue tape studio. An engineer could thread an open reel, monitor from a replay head, rehearse a punch-in, mark an edit point, cut the tape and move between channels in a way that remained familiar. At the same time, the recorder offered almost no print-through, extremely low crosstalk, stable timing and digital copying without generational loss.
The physical tape remained a linear medium. Finding material still required winding reels, recording time was limited by tape length, and editing involved either cutting the original or copying the required sections to another machine. However, many of the principles behind modern digital production were already present. Audio could be copied between tracks without degradation. Errors were detected and corrected mathematically. Samples could be crossfaded during record transitions. Multiple machines could chase timecode. Audio could be transferred between compatible devices as digital data rather than as an analogue waveform.
The machines also forced the professional recording industry to confront questions that remain familiar today. Studios had to manage incompatible formats, sample-rate conversion, digital clocking, interface standards, data integrity, media life and long-term access to recordings.
The end of the large digital tape machine
ProDigi remained professionally relevant into the early 1990s, but the economics of recording technology were changing rapidly. Large fixed-head machines required precision transports, extensive electronics, specialist maintenance and expensive metal-particle tape. They occupied considerable space and were practical mainly for well-funded studios, broadcasters, record companies and rental operations.
Hard-disk recording systems gradually offered non-linear access, visual editing and easier project management, while modular digital tape systems provided multiple digital tracks at a much lower cost. As these technologies improved, the operational advantages of the large reel-to-reel digital recorder became harder to justify. The decline of ProDigi did not indicate that Mitsubishi’s engineering had failed. It showed that professional recording had moved beyond the idea of using magnetic tape as the primary container for digital data.
Mitsubishi’s digital recording legacy
Mitsubishi Electric entered professional digital audio before the industry had agreed on its formats, interfaces or working methods. The company developed not only converters and tape machines but an integrated philosophy for reliable digital recording. The X-80 demonstrated that a stereo digital master could be recorded on open-reel tape and physically edited. The X-800 extended the concept to 32 independent audio channels. The PD and ProDigi formats introduced powerful two-dimensional error correction, self-clocking operation, interleaving and professional punch-in capability. The X-850 transformed these principles into a robust production system with advanced location, synchronisation and digital routing facilities. The X-86 Series provided a more standardised stereo mastering platform, while the X-86HS introduced 88.2 kHz and 96 kHz recording with internal sampling-rate conversion. The X-880 then reduced the size, noise and power consumption of the 32-channel system while preserving compatibility with existing tapes. For a significant period, Mitsubishi stood alongside Sony as one of the principal architects of professional digital multitrack recording. Its machines were installed in major studios, requested by leading producers and used on projects that helped establish digital recording as a normal part of professional production. The enormous open-reel digital recorder was ultimately a transitional technology, but it was a crucial transition. Mitsubishi preserved the discipline, physical control and reliability expected from a professional tape machine while introducing the error correction, lossless copying, synchronisation and data processing that would define the digital studio. The company’s successful period in professional audio therefore represents much more than a collection of obsolete recorders. It marks the point at which digital multitrack recording became a credible, practical and commercially successful alternative to analogue tape.