Sega vs Namco vs Taito: JVS I/O Boards Compared (JVS-003)

A board identifier is not a compatibility certificate

JVS-001 describes the protocol and JVS-002 follows bus discovery. This third reference asks a different question: what do actual Sega, Namco and Taito I/O boards identify themselves as, and what do they say they support? Two devices can use the same E0/D0 framing and both claim JVS 2.0 yet expose different controls, revision bytes, outputs and recovery behaviour.

The reliable way to compare boards is to preserve the command-10 identification string, all three BCD version replies (11/12/13), the raw command-14 feature tuples, board model and physical configuration. A feature advertised by firmware is not proof that a cabinet harness, peripheral or host application actually uses it.

This is a technical comparison grounded in one reconstructed Sega firmware/PCB family and public third-party JVS Information Extractor captures. It is not a universal list of supported games or an invitation to swap incompatible hardware.

How the evidence is classified

FIRST-PARTY FIRMWARE/BOARD means we decoded a specific SP5001-B ROM path and, where stated, traced conductors and components on an identified Sega 838-13683-93 Rev.B. It is source and continuity evidence, not a fresh live serial capture for this publication.

THIRD-PARTY HARDWARE CAPTURE means somebody else queried an actual identified board and published its responses. Their raw identity/feature bytes are strong evidence for the configuration tested, but we cannot infer the MCU's internal logic, the firmware of another revision or the state of the cabinet's connectors.

PROVENANCE OPEN means a public captured ID exists, but factory firmware attribution is unproven. Emulators and community profile presets are additional research aids, not replacements for a real trace or primary firmware.

Version bytes: compare these before assuming sameness

Commands 11, 12 and 13 return independent BCD-encoded values for command-format, JVS-standard and communications versions. For example 13 means version 1.3, not thirteen; 30 means version 3.0, not thirty. This matrix transcribes our Sega firmware findings and the published extractor results without conflating the three fields.

The important contrast is that Namco FCA-1 and FCB both report JVS 2.0 (20), yet FCB reports command format 1.3 (13) and communications 3.0 (30). Taito's K91X1130A also reports command format 1.3, but JVS 3.0 and communications 1.0. Version similarity is not a sufficient interoperability test.

Index of exact reported JVS identification strings

Command 10 returns an identification string. These values come from the primary SP5001-B firmware study or attributed public captures of identified boards. Original punctuation and revision fields are preserved so diagnostic tools can compare reports directly.

An ID alone never proves electrical or software compatibility. The 98701-labelled converter is explicitly tagged as having unresolved factory provenance.

Capability 14h: preserve bytes before translating them

The function list consists of four-byte records CODE P1 P2 P3, ending with one standalone 00. Common examples are 01 switch input, 02 coin inputs, 03 analog inputs, 04 rotary encoders, 06 screen position, 12 general-purpose output and 13 analog output. Parameter meanings depend on the function, and byte values are hexadecimal.

Compare the exact records. The tuple 01 01 10 00 on Namco FCA-1 describes one player and sixteen switch positions (10 hexadecimal), while 01 01 13 00 on Taito describes one player and nineteen positions. The analog tuple 03 07 00 00 in the FCA-1 public capture must keep the raw effective-resolution byte 00; we cannot replace it with an invented bit depth.

General-output code 12 deserves special care. A published command reference describes P1 as a bank count of eight-bit banks, while our Sega source audit associates 12 08 00 00 with an eight-bit P0 driver-output variant. Those two descriptions need original-spec and implementation reconciliation: in this article we publish the raw tuple but do not translate it into an unverified number of physical output pins.

FEATURE  PARAM1 PARAM2 PARAM3
01       01     10     00     Namco FCA-1 : 1P, 0x10 switches
01       01     13     00     Taito K91X1130A: 1P, 0x13 switches
03       07     00     00     Namco FCA-1: 7 analog, width unresolved
03       07     0A     00     Namco FCB / Taito: 7 analog, 10-bit parameter
06       10     10     01     FCB / Taito: 16 X, 16 Y, 1 position channel
12       0D     00     00     Taito general-output declaration (raw)
00                              END OF LIST (one byte)

These are excerpts from separately sourced profiles, NOT one packet.

Raw command-14 capability inventory by configuration

Each four-byte record is one CODE P1 P2 P3 feature; a real response ends with one additional 00 byte. Every list belongs to a specific hardware/firmware configuration. The two Sega SP5001-B modes were decoded from firmware, not captured live here.

For output feature 12, the exact unit requires reconciliation of implementation records with the primary specification. We preserve raw hexadecimal and do not silently multiply an advertised value into a physical wire count.

Sega 838-13683B: why a ROM is more revealing than a label

Our SP5001-B reconstruction identifies 14 non-default JVS command handlers: F0, F1, 10 through 14, 20 through 22, 2F, 30, 32 and 35. It also establishes 115200 8N1, E0/D0 decoding, checksum handling, the F1 unassigned-address gate, and a P7.2-controlled SENSE sink traced on a specific Rev.B PCB.

Command 14 does not always return the same functions. In the mode where FF73 equals zero, the firmware advertises 01 02 0F 00, 02 02 00 00 and 03 08 00 00, with no 12 output feature. When FF73 is nonzero it advertises 01 02 0B 00 and adds 12 08 00 00. The mode flag is sampled from P7 bit 3 and also changes P0CR (00 versus FF).

Consequently, even two devices with the same board identification can expose 15 versus 11 switch positions per player and different output features depending on the original firmware's hardware configuration. That is a real conditional path in the recovered source, not proof that every retail PCB offers a user-selectable switch.

Sega 838-13683D: similar family, distinct evidence

A public hardware-capture session identifies another Sega node as SEGA ENTERPRISES,LTD.;I/O 838-13683D ;Ver1.09;99/11. Its version replies are 11/20/10. Its raw functions are 01 02 0B 00, 02 02 00 00, 03 08 00 00 and 12 08 00 00.

Those data resemble the SP5001-B output-capable profile, but B and D are not interchangeable firmware revision names. The 838-13683D values are third-party observed replies; our first-party firmware and Rev.B continuity evidence concern the earlier B-family study.

This is exactly why a catalogue entry should include the complete model string and report provenance rather than a generic 'Sega JVS I/O' badge.

Namco FCA-1: rotary, analog and an unresolved parameter

The third-party Namco FCA-1 Ver1.01 capture identifies a multipurpose board with rotary encoder capability. The version trio is 11/20/10; the feature records include 01 01 10 00 (one player, sixteen switches), 02 02 00 00 (two coin slots), 03 07 00 00 (seven analog channels), and 04 02 00 00 (two rotary channels). It also reports 12 06 00 00 and 13 04 00 00 output-family records.

An analog resolution argument of 00 is not permission to assume eight, ten or sixteen meaningful ADC bits. Keep it as raw 00 until firmware or controlled physical tests explain the board's effective width.

The public record is real-board evidence, not an Atelier Rocky firmware disassembly of FCA-1. A future ROM reconstruction should reproduce the observed ID, version bytes and command-14 tuples for the matched hardware configuration before making broader claims.

Namco FCB: same JVS revision, different capability map

The publicly captured Namco V288/JV2RS FCB1 identifies as FCB Ver1.02, described as TouchPanel&Multipurpose. Its three versions are 13/20/30, rather than FCA-1's 11/20/10. Both report JVS revision 2.0 while exposing distinct protocol and control capabilities.

The FCB list includes digital inputs 01 01 10 00, coin slots 02 02 00 00, analog 03 07 0A 00, rotary 04 02 00 00 and screen-position 06 10 10 01, plus output-family tuples 12 06 00 00 and 13 02 00 00. The position tuple advertises sixteen bits for both coordinates and one channel.

A feature code is not a guarantee that a specific touchscreen, light gun or cabinet monitor will work. The known public FCB main-MCU content is incomplete for much of the JVS handling, making the third-party real-board trace particularly useful without making it our own measurement.

E0 00 20 01 01
01 01 10 00    one player / 16 switch positions
02 02 00 00    two coin slots
03 07 0A 00    seven analog channels; raw resolution 0A
04 02 00 00    two rotary channels
06 10 10 01    X=16 bits, Y=16 bits, one position channel
12 06 00 00    output-family feature; raw P1=06
13 02 00 00    two analog-output channels
00             feature terminator
A6             checksum

Source: Arcade-Projects JVS Information Extractor (2016).
Not a trace made at Atelier Rocky.

Namco V185 TSS-I/O: evidence of another kind

A separate published JVSCore test of a physical Namco V185 TSS-I/O Ver2.02 reported the ID namco ltd.;TSS-I/O;Ver2.02;JPN,GUN-EXTENTION. The operator summarized one player, twelve switches, one coin input, one screen-position channel and three general outputs.

The reported coordinate format was sixteen bits each for X and Y, but the tester cautioned that observed scaling did not span the full nominal range. That is a functional summary rather than the complete raw 14h response seen for FCA-1 and FCB; inventing missing tuple bytes would misrepresent the evidence.

Screen-position capabilities also depend on application code, gun calibration and display technology. A high-level summary alone does not qualify a replacement board for any particular arcade game.

Taito K91X1130A: a broad profile with a precise part number

The public 2016 extractor recorded the Taito K91X1130A as TAITO CORP.;U JVS PCB;FJ-type;Ver2.23;2007.08.07. It reported command-format 13, JVS 30 and communications 10. The product firmware version Ver2.23 is not the same thing as its JVS revision 3.0.

Its capabilities include one player with hexadecimal 13 (nineteen) switch positions, one coin slot, seven analog channels with resolution parameter 0A, one 16/16-bit screen-position channel and general/analog output entries. The general-output 12 0D 00 00 is preserved as raw bytes pending a definitive interpretation of the unit.

Atelier Rocky separately acquired and researched a Taito K91X1215A board with an H8S/2138A microcontroller. K91X1215A is not K91X1130A, and we have not proven equivalent firmware, connector maps or features. There is no legitimate reason to copy the earlier capture into the newer board's compatibility row.

E0 00 1C 01 01
01 01 13 00    one player / 0x13 (19) switches
02 01 00 00    one coin slot
03 07 0A 00    seven analog channels; raw resolution 0A
06 10 10 01    one 16/16-bit screen-position channel
12 0D 00 00    general-output declaration, raw P1=0D
13 06 00 00    six analog-output channels
00             feature terminator
A9             checksum

Source: Arcade-Projects JVS Information Extractor (2016).
This is NOT a capture of K91X1215A.

The 98701 / Capcom-labelled converter: factory identity unresolved

An extractor user labeled a tested converter CAPCOM JAMMA JVS, but command 10 actually returned No Brand;NAOMI Converter98701;ver2.0;. The same person questioned whether it was factory-original after seeing that result. Its recorded versions were 11/20/10.

The specific captured mode advertised two players with twelve switch positions each (01 02 0C 00), two coin slots (02 02 00 00), and eight analog channels with raw resolution 00 (03 08 00 00). Other reports discuss configurable four-player modes, so one capture cannot establish every available profile.

We cannot call these bytes 'original Capcom firmware', but neither can we declare a clone without a controlled primary-hardware comparison. The accurate label is an externally captured 98701-associated converter with unresolved factory provenance.

Host firmware matters: Namco WAJVIO Ver2.04 versus Ver2.1.1

Our reverse engineering of Namco System ES/WinArc WAJVIO Ver2.04 found an eighteen-state discovery sequence and a steady-state multi-node reply-association limitation. Its common default services only one node, which avoids much of that edge case.

In Ver2.1.1, we verified a different schedule: complete send, receive, parse, account and publish for one node before servicing the next. That structurally resolves the older multi-node association problem. These are original software findings about particular DLLs, not generic guarantees for every Namco host.

The presence of Ver2.1.1 in a preserved Time Crisis 5 ES3X package does not show that Time Crisis 5 relies on WAJVIO for its active cabinet I/O. Our separate TC5 analysis identifies USIO/WinUSB as its main path. Package provenance and runtime use must remain separate.

How to test a replacement board without overclaiming

For every candidate, save the physical PCB marking, photos and firmware revision; the exact raw command-10 ID, 11/12/13 version bytes, 14 capability records and any configured switch/DIP modes. Preserve physical escaped frames and validated logical frames as separate artifacts.

Then test reset, address assignment, one-node and multi-node enumeration, representative controls, analog ranges, output safety and disconnection/recovery on the actual host, game software and cabinet wiring. A profile that answers identification commands is not a full compatibility pass.

These are research records, not purchasable Atelier Rocky adapters. JVS connectors can resemble USB but use a different electrical protocol. Do not attach conventional USB gear or unqualified interfaces.

• Keep the original hex tuples alongside interpretations; 0x10 equals sixteen.
• Treat analog-resolution 00 and output feature-12 unit discrepancies as unresolved.
• Do not interchange Taito K91X1130A and K91X1215A evidence.
• Do not present community captures as our own bench measurements.
• Only claim verified compatibility for an explicitly documented host, PCB revision and firmware.