The Alesis ADAT RMB
A 32-channel Remote Meter Bridge for the original Alesis ADAT, documented far enough to establish its connector topology and controls—but not yet its meter-data encoding. This page separates what the manuals prove, what the evidence suggests, and what must be measured next.
🎚 What This Project Is Building
32 Meter Channels
Four identical DB9 links, each responsible for eight 15-segment displays.
Dual Decay
Fast and slow behavior selected on the RMB and visible to the source adapter.
Three Peak Modes
Momentary, continuous hold, and off—with the hardware bridge authoritative.
Peak Clear
An event that must be captured reliably even though it has no dedicated pin.
Track Status
Separate record and input indicators for every track, in addition to level.
Measured Protocol
No UART, RS-232, or RS-422 assumption. Waveform first, decoder second.
The useful breakthrough
The ADAT service manual identifies every DB9 signal and its direction. That narrows the problem from “unknown serial port” to one outbound meter-data signal, three inbound control signals, a 5 V reference, and four grounds.
The present blocker
No source establishes pin 2 timing or framing. A working original blackface ADAT or Fostex RD-8 must be observed before a safe transmitter can be designed.
📐 What the Hardware Tells Us
The RMB is an independently powered 2U display for as many as four original ADAT recorders. One straight-through male-to-male DB9 cable represents eight tracks. Each track has a 15-segment level bar plus separate record and input indicators.
Audio / recorder state
Original ADAT or RD-8 produces level and track status.
RMB local logic
Renders bars, decay, peak state, record and input indicators.
Recorder display
Mirrors the RMB’s selected decay and peak behavior.
Compatibility boundary
- Original “blackface” ADAT: documented Meter Bridge Out source.
- Fostex RD-8: reported compatible and suitable as a possible reference transmitter.
- ADAT XT: omitted the meter output and cannot serve as a direct source.
- BRC: does not relay meter data; the RMB still connects directly to each recorder.
Power and cabling
The bridge has its own universal 90–250 VAC, 50/60 Hz supply and is specified at 50 W maximum. Pin 1 is therefore an interface reference used by opto-isolators, not evidence that an ADAT powers the entire bridge. Alesis specifies shielded straight-through DB9 cables and describes long runs of 100 feet or more as possible.
🔌 The Documented DB9 Interface
Directions are relative to the ADAT. The RMB sends its local control selections back to every attached recorder.
| Pin | Direction at ADAT | Label | Function | Confidence |
|---|---|---|---|---|
| 1 | Output | +5 V | Reference voltage for RMB opto-isolators | Confirmed |
| 2 | Output | Meter data | Buffered display information to RMB | Confirmed |
| 3 | Input | F/S | Fast / Slow selection from RMB | Confirmed |
| 4 | Input | PM | Peak Momentary selection from RMB | Confirmed |
| 5 | Input | PH | Peak Hold selection from RMB | Confirmed |
| 6–9 | Common | Ground | PCB ground | Confirmed |
| Shell | — | Unknown | Shield/chassis relationship requires continuity testing | Open |
The service manual contains an apparent connector typo
The paragraph first names J2 as the RMB connector, then places control pins 3–5 on J3. Its next paragraph identifies J3/J4 as Sync Out/In. Context strongly indicates that the control sentence means J2, but continuity or a live capture must still verify it.
Do not connect a PC serial adapter
This topology is neither RS-232 nor differential RS-422. Bipolar RS-232 levels can damage 5 V circuitry, and the DB9 shell does not make the protocol standard.
🎛 Decay, Peak Modes, and Peak Clear
The RMB owns these display behaviors locally and tells connected ADATs to mirror them. The eventual source adapter must therefore send meter/status data and treat the RMB’s return lines as semantic state—not expose raw pin polarity to application code.
Behavior model
Illustration only. This model demonstrates the required behavior; it is not a pin-2 protocol emulator. Momentary peaks expire after roughly two seconds. Continuous peaks remain until cleared. Peak Clear is enabled only in Continuous mode.
| Control | Visible behavior | Return path | Wire semantics |
|---|---|---|---|
| Fast decay | Bars fall quickly | F/S, pin 3 | Polarity unknown |
| Slow decay | Bars fall more gradually | F/S, pin 3 | Polarity unknown |
| Momentary peak | Peak remains about two seconds | PM/PH, pins 4–5 | Encoding inferred |
| Continuous peak | Peak remains until clear or mode change | PM/PH, pins 4–5 | Encoding inferred |
| Peak off | No peak indication | PM/PH, pins 4–5 | Encoding inferred |
| Peak Clear | Clears held peaks in Continuous mode | No dedicated pin documented | Event unknown |
The likely mode encoding
The simplest model is PM asserted for Momentary, PH asserted for Continuous/Hold, and neither asserted for Off. Both asserted may be unused. That is a testable inference, not a fact.
Peak Clear is the control trap
There is no fourth return wire, yet the manual says Peak Clear also clears the attached ADAT’s display. It must therefore appear as a transition on PM/PH, a temporary mode change, or another behavior on pins 3–5. All three lines must be captured simultaneously around the button press.
🧪 Protocol: Facts Before a Decoder
Pin 2 carries buffered meter-display information, but no examined manual states its voltage thresholds, idle level, symbol rate, frame structure, bit order, or update cadence. The right first question is what the waveform is—not which UART baud rate makes it look plausible.
H1 · Repeated full frame
Each update may contain all eight tracks.
Falsified if traffic occurs only when values change.
H2 · Status in each track field
Level, record, and input may travel together.
Falsified if flags alter another frame or channel.
H3 · Quantized level code
The 15-segment height may be sent as a compact value.
Falsified by direct 15-bit segment masks.
H4 · Static PM/PH state
PM and PH may be mutually exclusive mode signals.
Falsified by pulses, both asserted, or another Off code.
H5 · Clear as a control edge
Peak Clear may briefly change PM or PH.
Falsified if pins 3–5 remain unchanged.
H6 · Instantaneous meter values
The RMB may apply all decay and peak processing itself.
Falsified if pin-2 values visibly decay or hold after input drops.
What a valid decoder must explain
- eight independent levels and all 15 visible steps;
- record and input state for each track;
- a repeatable synchronization rule and channel order;
- steady silence, abrupt level steps, startup, and mode changes;
- captures from more than one physical recorder where possible.
Firmware evidence has a limit
An ADAT v3.05 27C512 image contains valid 8051-family code, but the service manual assigns display work to a VU ASIC. Firmware analysis may reveal register writes or mode handling without exposing the pin-2 serializer itself.
🔬 The Bench Plan
The shortest route is passive observation of one working eight-track link. Four-port replication comes only after the first link is electrically and logically understood.
Unpowered continuity and resistance
Verify the straight-through cable, confirm pins 6–9, map shell/chassis relationships, compare all four ports, and record hardware revisions.
Powered electrical baseline
With high-impedance scope probes only, measure pin 1, pin 2 idle and swing, control-line levels, edge rates, startup, and power-order behavior.
Capture the control truth table
Record pins 3–5 together through Fast, Slow, all peak modes, Peak Clear with and without a held peak, and a power cycle.
Controlled pin-2 captures
Collect startup, silence, every visible channel-1 level, channels 2 and 8, unequal two-channel states, step-down, record, input, mode-cycle, and Peak Clear cases.
Infer and falsify a frame model
Require one decoder to explain every capture. Reject hand-picked offsets, one-channel coincidences, and framing that cannot resynchronize.
Protected one-port transmitter
Use replaceable buffering, measured thresholds, current limiting, test points, and independent control inputs. Validate static bars before dynamic audio.
Controlled capture set
Meter-data captures
powerup, silence, ch1-levels, ch2-levels, ch8-levels, two-channels, and step-down isolate framing, ordering, and decay ownership.
Status and control captures
record-status, input-status, mode-cycle, and peak-clear isolate non-level fields and correlate the three return lines with visible behavior.
Capture metadata
Every file records recorder model and firmware, sample rate, transport state, stimulus level/frequency, RMB mode, instrument thresholds/sample rate, timestamp, visible result, and SHA-256 hash.
Observation comes before drive
No USB serial adapter, logic output, or bench supply should drive the RMB connector during the observation stages. Confirm analyzer voltage tolerance after measuring the waveform on a scope.
🧩 Intended Adapter Boundary
Audio / source state
32 instantaneous levels plus record/input flags.
Protocol engine
Measured framing, deterministic timing, four 8-track streams.
RMB
Local decay, peak modes, clear event, persistent user selection.
The initial design should treat the bridge as authoritative for visible decay and peak behavior. The host can mirror those selections, but it should not pre-apply decay or peak hold unless step-down captures prove that the RMB expects preprocessed values.
| Software concept | Meaning | Hardware source |
|---|---|---|
DecayMode | Fast | Slow | Decoded pin 3 state |
PeakMode | Momentary | Continuous | Off | Decoded pins 4–5 state |
PeakClear | Edge/event, never a polled level | Measured transition on control returns |
TrackState[32] | Instantaneous level, record, input | Host/audio source |
Hardware design remains intentionally open
The MCU, buffer family, threshold network, isolation, and ground strategy depend on measurements still missing. The first PCB should expose pins 1–5 and each ground domain as test points and avoid tying signal ground, USB shield, chassis, and protective earth together by assumption.
📍 Where It Stands
Research status as of 16 October 2026.
| Part | Status | What is known / needed |
|---|---|---|
| Product identity and compatibility | Documented | Original blackface ADAT; four eight-track links; XT omitted the output. |
| DB9 pin assignment | Documented | One data signal, three control returns, +5 V reference, four grounds. |
| Decay and peak behavior | Documented | Fast/Slow; Momentary/Continuous/Off; Peak Clear in Continuous. |
| Control-line polarity and event shape | Capture needed | Measure pins 3–5 through every local button sequence. |
| Pin-2 electrical interface | Capture needed | Idle level, swing, impedance, edges, and output topology unknown. |
| Meter-data framing | Unknown | No baud rate, timing, synchronization, or bit map established. |
| One-port transmitter | Blocked | Unsafe to implement before electrical and framing captures. |
| Four-port 32-channel source | Planned | Replicate only after one link survives bench validation. |
📚 Sources and Evidence Policy
Primary manufacturer documentation carries the most weight. Secondary sources are used only for corroboration or to record earlier speculation. Original manuals and firmware are not republished by this project; the repository contains our analysis, links, hashes, diagrams, and plans.
Evidence labels used here
Confirmed means stated by manufacturer documentation or directly measured. Inferred means the simplest testable explanation consistent with confirmed evidence. Open means unresolved and unsafe to assume.