Bosch MED17.1 on the EA888 Gen1 and Gen2 2.0 TFSI

Bosch MED17 petrol DI turbo (torque-structure Motronic). Two sub-families cover the 2.0 TFSI/TSI EA888 gen1/2 (and early gen3): MED17.5.5 (transverse PQ35/PQ46 gen1, e.g. CCTA-class) and MED17.1 / 17.1.1 / 17.1.6 / 17.1.27 (longitudinal Audi B8 gen2 and MQB-era gen3). Both are Infineon TriCore, little-endian, full-flash calibrated.

Reference - updated 4 August 2026

The ECU hardware

MED17.5.5 = Infineon TriCore TC1766 (1.5 MB PFLASH) / TC1767 (2 MB PFLASH); full read ~1.5-2 MB, no/limited external flash. MED17.1.1 = TC1796 (2 MB internal PFLASH + 128 KB DFLASH); full image is commonly ~2 MB (TC1796 internal) — external memory is NOT standard on 17.1.1, so do not assume a 4 MB read. MED17.1.6 / 17.1.27 = TC1797 (4 MB internal PFLASH) → ~4 MB image. All 32-bit TriCore, little-endian; PFLASH mapped in segment 8 at 0x80000000 (cached) and segment A at 0xA0000000 (non-cached); internal emulated EEPROM = DFLASH (holds immo/adaptation/component-protection). Confidence: MCU-per-variant mapping is solid; exact image size and any internal-vs-external split vary by hardware rev and must be confirmed from the actual read.

Fuelling and load sensing

Petrol, direct-injection (FSI/TSI) turbocharged. Gen1/Gen2 EA888 = DI-only (single cam-driven Bosch HDP5-class HPFP, stock rail ~120-150 bar class). Gen3 added port+direct (MPI+DI) dual-fuel — relevant only if the target is genuinely a gen3 code. WOT is enrichment-cooled (lambda < 1, ~0.80) for EGT/knock protection; no diesel smoke/DPF path (later EU6 gen3 petrol may carry a GPF/OPF).

How this ECU is read and flashed

Read/write over ISO-TP UDS on CAN (OBD-II pins 6/14). Three access tiers: (1) OBD flash — UDS 0x10 session, 0x27 SecurityAccess (seed/key), 0x31 erase routine, 0x34/0x36/0x37 transfer, 0x11 reset; MED17.5.5 OBD read is often a 'virtual/partial' read (calibration slots only), so a full clone needs bench/boot. (2) BENCH — ECU on the bench powered via boot/bench connector, tool talks to the loader directly and reads full internal flash without opening the die; supported for non-Tprot MED17.1.1/17.5.5. (3) BOOT MODE (TriCore BSL) — solder/probe the BSL-entry test point to enter Infineon bootstrap; required whenever the ECU is Tprot read-protected or RSA-signed and OBD/bench are refused. Tools: Alientech KESS3/KTAG, CMDFlash, MagicMotorsport Flex, ECUHelp KT200 / FoxFlash, PCMflash (has TriCore/BSL boot modules for Bosch MED17/EDC17 — module numbering varies by version), Autotuner, bFlash. IMMO/Security: SecurityAccess seed-key handled by the tool; 'TPROT V08 VAG'), but newer RSA-2048/4096 signature (roughly MED17.1.6+ / 2012+) blocks OBD writes of modified code → boot/bench or a signature-aware loader. Component Protection (CP) + immo data live in DFLASH — matters for ECU swap/virginizing/cloning, NOT for a same-ECU calibration reflash. Always keep an untouched original full read as the recovery/checksum baseline.

Engine codes in this family

• CCTA (EA888 gen1, 2.0 TSI transverse ~200 hp — MED17.5.5; VW CC/Tiguan/EOS/Passat B6, Audi A3 8P)
• CAEB (EA888 gen2, 2.0 TFSI longitudinal ~211 hp — MED17.1; Audi A4/A5/Q5 B8, SEAT Exeo)
• CDNC (EA888 gen2, 2.0 TFSI longitudinal ~211 hp — MED17.1 / later MED17.1.6; Audi A4/A5/Q5 B8)
• CPMA (EA888 gen3 MQB, 2.0 TSI ~220 hp — commonly Simos 12/18, sometimes MED17.1.6; VERIFY the ECU label)
• CPPA (EA888 gen3, 2.0 TFSI, Garrett MGT1752S — typically Simos 18.1/18.10 rather than MED17; VERIFY before assuming MED17)

Staging - what each stage really gives

• stage1 - Software only, 100% stock hardware, requires higher-octane fuel (98-99 RON / 93 AKI). Raise LDRXN + boost setpoint/WG duty, lift KFMIRL/KFMIOP + torque limiters in lockstep, optimize KFZW timing, set WOT lambda/rail pressure, disable only irrelevant DTCs. Typical: CCTA-class 200 hp → ~240-260 hp; CAEB/CDNC 211 hp → ~260-280 hp with torque to ~370-410 Nm. Gains ≈ +40-65 hp / +80-120 Nm. Stock K03/IHI turbo, stock injectors/HPFP, stock clutch. NOTE: even Stage 1 torque already meets/exceeds the DQ250 ~350 Nm clutch limit — watch DSG/clutch slip.
• stage2 - Bolt-ons + software: high-flow/catless downpipe, intake, upgraded intercooler/FMIC, sometimes upgraded diverter valve. Recalibrate boost, load model, EGT and fueling for the freer exhaust; disable cat monitor (P0420) ONLY if the cat is actually removed. Realistic on the stock K03/IHI turbo frame: ~285-310 hp / ~420-450 Nm — the stock turbo is near its efficiency/surge limit here, and fueling (rail pressure/enrichment) plus EGT become the ceiling. Torque is well past DQ250 (~350 Nm), so clutch/DSG headroom is the practical limit.
• stage3 - Turbocharger upgrade (hybrid K04 or big single) + supporting fuel (larger injectors and/or upgraded HPFP; aux/port fuel sometimes added on gen1/2 DI-only) + clutch/DSG upgrade + intercooler/exhaust. Full recal: injector characterization, MAF/load model rescale, boost/wastegate, torque model & limiters, ignition/knock, rail pressure. K04-class ~330-360 hp; big-single 400+ hp but FUEL-SYSTEM LIMITED on DI-only gen1/2 (~350-400 whp before HPFP/injectors max out). Mandatory clutch (DQ250 ~350 Nm / DQ500 ~550-600 Nm limits) and TCU torque-limit calibration; big-single power also approaches the stock rod/piston-ringland ceiling — confirm internals.

Safety limits

• Torque-monitor consistency: modeled vs delivered torque and KFMIRL/KFMIOP/limiters MUST stay coherent — mismatch = torque-deviation DTC, torque intervention, limp. This is a hard functional-safety limit, not a comfort one.
• LSPI / low-speed pre-ignition (super-knock): DI turbo engines are prone to LSPI when high load is commanded at very low rpm (lugging). Do NOT raise low-rpm load/boost aggressively, keep enrichment in the low-rpm/high-load zone, and shape the torque request to avoid full boost below ~1800-2000 rpm.
• WOT AFR: target ~11.5-12.2:1 (lambda ≈ 0.78-0.83) under full boost on gen1/2; never lean out at high load — leaner = knock + EGT spike.
• EGT (pre-turbine): keep below ~950 °C sustained; treat ~980 °C as a hard ceiling — turbine/manifold and (gen3) exhaust-valve damage. Enrichment is the primary EGT control.
• Knock: hold low/zero knock-controller retard activity at WOT; if the ECU is pulling large timing, the tune is over-advanced for the fuel — back off KFZW, do not mask it.
• Rail pressure: do not command beyond stock HPFP capability (~120-150 bar class on gen1/2) — commanding past pump authority causes lean-out; raise only with an upgraded pump.
• HPFP cam-follower wear (gen1/2 EA888): higher/sustained rail-pressure demand accelerates wear of the cam follower that drives the pump — inspect regularly and upgrade the follower/pump for Stage 2/3.
• Boost/turbo: respect compressor surge and overspeed of the stock K03/IHI frame (practical peak ~19-22 psi); sustained over-boost destroys the turbo and can trip the torque monitor.
• Clutch/DSG torque: DQ250 ≈ 350 Nm, DQ500 ≈ 550-600 Nm input-torque limits; manual stock clutch is similar order — exceed only with an upgraded clutch AND raised TCU clutch-torque calibration, else slip/torque-cut. Note Stage 1 already reaches the DQ250 limit.
• Engine hardware ceiling: gen1/2 EA888 rods/pistons are reliable to roughly ~380-420 hp / big-single territory; beyond that (and under knock) ringlands and rods are at risk — forged internals for high-boost Stage 3.
• GPF/OPF (later EU6 gen3 petrol only): if the target actually carries a particulate filter, do not disable its monitoring or remove it without recalibration — backpressure, regen and emissions logic must stay coherent.
• Charge/intake air temp (IAT) & heat-soak: on stock intercoolers, timing must be pulled with rising IAT — keep the IAT-based timing/enrichment corrections intact so repeated pulls don't run over-advanced.
• Checksum/signature integrity: every modified block must be re-CRC'd (and the RSA/tprot signature respected or handled via boot/bench) before flash — an invalid checksum bricks the flash or prevents boot. Always keep an untouched original for recovery.
• Never disable core torque or knock monitoring to 'unlock' power — those are the ECU's protection layers; only disable diagnostics whose hardware you actually changed.

Every car using this platform (30)

Stock power and typical remap figures per car. Stage results assume a healthy engine on good fuel.

How confident is this?

SOLID: TriCore/little-endian architecture and MCU-per-family (TC1766/67 for 17.5.5, TC1796 for 17.1.1, TC1797 for 17.1.6/.27); the layered checksum concept (per-block CRC + descriptor table, RSA/tprot signature on newer/post-2012 variants); the three flash-access tiers (OBD/bench/boot-BSL) and Tprot/CP behavior; the Bosch torque-structure map roles (LDRXN/KFMIRL/KFMIOP/KFZW/lambda/rail/limiters) and their lockstep dependencies; staging philosophy; and the safety ceilings including LSPI, EGT, AFR, knock, rail, clutch/DSG and hardware limits. MEDIUM: exact staging power figures (vary by market/fuel/HW rev and specific SW); the full-image SIZE for MED17.1.1 (TC1796 is 2 MB internal — treat ~2 MB as the baseline and confirm from the read, since some revs differ); which precise sub-variant (17.1.1 vs 17.1.6) a given part number carries; and whether a given gen3 code (CPMA/CPPA) is MED17 vs Simos 12/18 — read it from the actual ECU/part number. THIN / DO-NOT-ASSERT: any specific per-software flash ADDRESS or offset for a map or checksum block — legitimately obtained only from a version-matched DAMOS/A2L or a validated def plus on-file validation, never memorized. The single MED17.5.5 add