2024+ Mazda MX-5 ND3 — OBD-II PID Research

Confirmed working PIDs and implementation notes for the 2024+ MX-5 (ND3 platform).

ND3 is a different platform. All prior ND1/ND2 PID work (berumiya DBC, timurrrr RaceChrono DB, agronick CSV) was built on the old architecture. The ND3 uses the CX-60 generation electronics with isolated CAN buses; the OBD port no longer exposes raw CAN traffic — only diagnostic request/response (Mode 01, Mode 22, etc.) is accessible from the port.

Status

Validated on: 2025 MX-5 RF Club (Cal ID PFD2EB000PEP6010, Engine Family STKXV02.0FFB) with OBDLink MX+ over Bluetooth Classic SPP.

Methodology (2026-06-23 update): mixed Mode 01 + Mode 22 candidate list probed via auto-probe-all with raw response logging. Decoded values validated against:

• Sanity ranges (e.g., RPM 700–7000, throttle 0–100%)
• Cross-PID correlation (throttle ↔ engine load ↔ MAF ↔ fuel rate)
• Real-world dynamics during 2-km drive captures (idle → cruise → WOT)
• Temperature warming curves (cold start → warm idle)
• Odometer drift matching actual distance driven (8698.8 → 8700.8 km = 2.0 km ✓)

Confirmed Working PIDs (36)

Mode 01 (Standard OBD-II — should work on any car responding to 01 00)

#	Signal	Header	PID	Formula	Units	Observed range
1	Engine RPM	7DF	01 0C	((A*256)+B)/4	RPM	806 → 2367
2	Vehicle speed	7DF	01 0D	A	km/h	13 → 58
3	Coolant temp	7DF	01 05	A-40	°C	80 → 89 (warming)
4	Coolant temp 1 & 2	7DF	01 67	B-40 (byte B)	°C	mirrors 0x05
5	Throttle position	7DF	01 11	A*100/255	%	13 → 92
6	Engine load	7DF	01 04	A*100/255	%	14 → 99
7	Absolute load	7DF	01 43	((A*256)+B)*100/255	%	11 → 55
8	Intake air temp	7DF	01 0F	A-40	°C	31 → 35
9	Ambient air temp	7DF	01 46	A-40	°C	27 → 35
10	MAF airflow	7DF	01 10	((A*256)+B)/100	g/s	2.5 → 43.8
11	Intake manifold pressure	7DF	01 0B	A	kPa	19 → 94
12	Barometric pressure	7DF	01 33	A	kPa	101 (constant)
13	Ignition timing advance	7DF	01 0E	(A/2)-64	°BTDC	7.5 → 34.5
14	Fuel/Air equiv ratio (λ)	7DF	01 44	((A*256)+B)*2/65536	—	0.91 → 2.00
15	Fuel rail gauge pressure	7DF	01 23	((A*256)+B)*10	kPa	9540 → 25110
16	Fuel rail pressure abs	7DF	01 59	((A*256)+B)*10	kPa	10080 → 25590
17	Fuel tank level	7DF	01 2F	A*100/255	%	matches dash gauge
18	Engine fuel rate	7DF	01 5E	((A*256)+B)*0.05	L/h	0.9 → 10.5
19	Engine/Vehicle fuel rate	7DF	01 9D	((A*256)+B)/50	g/s	0.45 → 6.05
20	Engine exhaust flow rate	7DF	01 9E	((A*256)+B)/20	kg/h	11 → 120
21	Catalyst temp B1S1	7DF	01 3C	((A*256)+B)/10-40	°C	365 → 591
22	Battery voltage	7DF	01 42	((A*256)+B)/1000	V	12.5 (off) / 13.8 (on)
23	Actual engine torque	7DF	01 62	A-125	%	10 → 78
24	Engine reference torque	7DF	01 63	(A*256)+B	Nm	250 (constant)
25	Engine friction torque	7DF	01 8E	A-125	%	7 → 10
26	Accel pedal position D	7DF	01 49	A*100/255	%	16 → 47
27	Accel pedal position E	7DF	01 4A	A*100/255	%	7.8 → 22.7
28	Relative throttle position	7DF	01 45	A*100/255	%	1.96 → 30
29	Absolute throttle position B	7DF	01 47	A*100/255	%	12 → 39.6
30	Commanded throttle actuator	7DF	01 4C	A*100/255	%	0 → 17
31	Transmission gear ratio	7DF	01 A4	((C*256)+D)/1000	—	1.594 / 2.035 / 2.991 / 3.127
32	Vehicle odometer	7DF	01 A6	((A*256³)+(B*256²)+(C*256)+D)/10	km	drift matched real

Mode 22 (Manufacturer-specific)

#	Signal	Header	PID	Formula	Units	Observed
33	Accel pedal % (encoding B)	7E0	22 03 2B	A/2	%	0 → 43.5
34	Engine oil temperature	7E0	22 13 10	(((A*256)+B)/100)-40	°C	66.7 → 78.4
35	Brake position	760	22 2B 0D	max(0, signed_int16(A,B))/2.3	%	0 → 0.43 (light tap)

Static-by-design (correctly constant — not bugs)

• 01 33 Baro = 101 kPa (sea level pressure; doesn't change in a 2-minute drive)
• 01 63 Reference torque = 250 Nm (vehicle constant per SAE spec — NOT a runtime value)

Gear Detection — 01 A4 (Engine-Running Required)

The standard SAE J1979-2 PID 0xA4 (Transmission Actual Gear) is supported on ND3. Critical quirk: requires engine running.

Byte layout (data after 41 A4 prefix):

Offset	Bytes	Meaning
0	A	Support bitmap (always 0x03)
1	B	Gear nibble — high nibble varies with engaged gear
2–3	C, D	Gear ratio = ((C*256)+D)/1000

Engine-off behavior: PCM holds the last-known gear ratio (often 5.087 = 1st gear default, the literal ND2/ND3 1st gear ratio). Gear is computed from RPM ÷ output-shaft-speed, which is undefined when RPM=0. Shifter Hall sensor alone is NOT what drives this PID.

Observed ratios during real drive (engine on, 2026-06-23):

• 1.594 (likely 4th-5th)
• 2.035 (likely 3rd)
• 2.991 (likely 2nd)
• 3.127 (likely 1st or transitional)

Byte B nibble observed varying 0x00 / 0x20 / 0x30 / 0x40 during drive — likely encodes gear number directly (nibble = gear), but mapping needs confirmation via parking-lot creep through 1st → 2nd → 3rd.

Neutral / Gear Status — 01 65 (Byte B, NOT Byte A)

Common misimplementation: reading byte A which is always 0x10 (just the SAE J1979 I/O support bitmap). The actual neutral/gear state lives in byte B, which was observed varying 0x30 / 0x00 / 0x60 / 0x40 across captures during a real drive.

Bit-to-gear mapping still TBD — needs correlated shift testing (neutral ↔ 1st ↔ 2nd, engine running) to identify which bit encodes neutral and which bits encode each gear number.

Steering Cluster — DSC 0x760/22 20 XX

Range scan revealed 4 PIDs in this cluster on the DSC module (not the EPS module at 0x730, which is silent). Currently pending engine-on turn-test validation.

PID	Hypothesis	Stationary value
22 20 33	Steering angle (signed 16-bit / 10 = deg)	00 00 centered (engine on); FF FF/00 00 no-data marker (engine off)
22 20 34	Steering rate or alt angle (signed 16-bit / 10)	00 00 static
22 20 1D	Status flag (DSC armed?)	0x04
22 20 1F	Status flag (fault indicator?)	0x00

Engine-off test (5 snapshots, clutch in, shifter moved through gears): none of these varied → SAS/DSC does NOT actively measure steering with the engine off. Engine-on + parked + wheel center → full-right → full-left is the required validation procedure.

Mode 22 PCM Cluster Discovery

Range scan of 7E0/22 XX XX clusters returned a massive trove (256 PIDs × 6 clusters = 1536 probes in ~6 minutes):

Cluster	Hits	Description
22 09 XX	8	Cam / VVT / valve cluster
22 13 XX	1	Thermal (just oil temp 13 10)
22 F4 XX	60	Fuel / boost / MAP — biggest trove
22 DA XX	36	Service / fuel learning (includes VIN + Cal ID fragments)
22 03 XX	38	Engine control (includes accel pedal 03 2B)
22 2B XX (DSC)	11	Chassis (includes brake 2B 0D)

Full byte-level results: pids/range-scan-2026-06-23.md

Dead Ends (Don't Re-Probe)

These were thoroughly tested and returned silence; do not waste time re-trying:

• EPS at 0x730 — all subranges (22 20 XX, 22 06 XX) silent across 512 PIDs. EPS does NOT respond to 0x730 header on ND3. Workshop manual address may be wrong, or EPS requires different addressing (try 0x7E2, 0x720?).
• 760/22 21 XX — yaw rate / lateral G cluster NOT on DSC at this range (0 hits in 256-PID scan). Use IMU/GPS instead (RaceBox provides at 25 Hz).
• 760/22 06 XX — wheel speeds NOT exposed via DSC at this range. The DSC module knows wheel speeds (it has to for ABS/traction), but does not publish them at this Mode 22 sub-address.

The Response-Prefix Fix

If you are building an OBD app that mixes Mode 01 and Mode 22 PIDs, see docs/prefix-fix.md.

TL;DR: positive response SID = request SID + 0x40. Do not hardcode 0x62.

The bug: apps that only ever tested Mode 22 PIDs hardcode 62 as the expected response prefix. Mode 01 responds with 41, Mode 09 with 49, etc. The generic fix:

val serviceId = pidHex.substring(0, 2).toInt(16)
val responseService = (serviceId + 0x40).toString(16).uppercase().padStart(2, '0')

This is not ND3-specific — it affects any app mixing modes against any car.

Next / Unconfirmed

See pids/nd3_candidates.csv for the current probe queue (updated with 2026-06-23 status). PRs with test results welcome.

Open questions:

• Bit-to-gear mapping for 01 65 byte B
• Confirm 01 A4 byte B high-nibble = gear number (parking-lot creep test)
• Confirm 760/22 20 33 is steering angle (engine-on turn test)
• Decode the 60 PIDs in the 22 F4 XX cluster
• Find correct EPS address (try 0x7E2, 0x720, physical addressing)

Related

• berumiya/CAN_DBC_6thGenMazda — ND1/ND2 raw CAN DBC (does not apply to ND3)
• timurrrr/RaceChronoDiyBleDevice — ND1/ND2 CAN IDs for RaceChrono
• miata.net — 2024 CAN bus research thread

Contributing

Test a candidate PID from pids/nd3_candidates.csv and open a PR updating the status column. Include your car trim and model year in the PR description.