Diesel Calibration · EDC17 Guide

The Bosch EDC17 tuning guide: maps through physics

Published April 16, 2026 · Updated April 20, 2026 · 14 min read · By Thomas Pirowski

You open an EDC17 binary in WinOLS^®. Four hundred maps in the left panel. Which ones matter? What do they actually control? And what happens if you change them wrong? The EDC17 family is the most common diesel ECU on the road — Volkswagen, Audi, BMW, Mercedes, Porsche, and thousands of trucks and agricultural machines since 2006. More than half the work of a typical tuning shop passes through this controller. But knowing where the maps live is not the same as knowing what they do. This guide walks every core map through the physics of the engine it controls.

EDC17 architecture: signal flow

You know where the maps are? Good. But do you know how to modify them professionally?

Imagine you open an MD1 or EDC17 binary in WinOLS. These are the most common Bosch engine controllers, and probably more than half your daily work is editing them. Over 100,000 maps and single-byte parameters — yes, a hundred thousand and more. Which ones matter? What do they actually control? How do they depend on each other? And what happens if you change them wrong?

If you get it wrong, you get problems. Sometimes the catalyst dies. Sometimes the engine. Sometimes anything that transmits torque. Or — the car drives fine and nothing breaks, even though your file is full of mistakes. Sometimes obvious ones.

There is someone who knows where the maps are and has a rough feel for what they do — changing them by some arbitrary percent without recalculating AFR. That is not enough.

We, professional tuners, also need to know why the ECU uses them and what happens when the values change. More than that — we need to change them so the whole system stays logical and consistent with the physics of the engine and combustion. That is what separates a real tuner / calibrator from someone who just presses buttons in a memorised sequence.

The EDC17 family is wide. MD1 is also growing, appearing in more and more variants. EDC17 has its “retirees” — CP02 (VAG 2.0 TDI — the example file we reference throughout this article), CP04 (BMW 2.0d), CP14 (VW 2.0 TDI CR), CP20 (VW V6 TDI), CP44 (Porsche, Audi) — and its heavily modernised units: C64 (Mercedes, VAG), C74 (VW, Audi, Seat, Skoda), and further. Looking at the code, the C-series are already early MD1 hidden in EDC17 clothing, running on a cheaper processor (TriCore instead of Aurix). But that is already reverse-engineering knowledge from Ghidra.

All of them control a diesel engine using the momental model. It works like this: convert the driver’s wish into a torque request, cut it with limiters, convert it to fuel mass (accounting for injector parameters as a function of fuel temperature), calculate injection duration (accounting for mechanical losses in the engine), and finally calculate the air parameters (boost pressure, variable turbo geometry).

For the ECU, torque does not really exist. We — humans — think in torque (and some people, less familiar with the topic, think in horsepower). For the ECU, only fuel mass and air flow exist. Yes, torque exists as a concept, but it is a secondary parameter. It is a requested value that has to be converted into physical control parameters. You do not tell the engine what torque to make. The combustion process needs air and fuel in the right amounts, at the right moment.

The fuel loop: from pedal to injector

The fuel loop is the core of diesel calibration. Four independent systems (groups of functions with their corrections and sometimes feedback), in this order:

Driver’s Wish → Torque Limiter → FMTC → Injection Duration.

Change them in the wrong order, or without controlling the final result, and one of two things happens: you either hit the maximum allowed torque (or fuel quantity, or injection time) and your changes do nothing — or you damage the engine. Sometimes after a while. Sometimes immediately.

Driver’s wish — pedal to torque request

The first operation is the simplest. The driver makes a request by pressing the accelerator. This is a simple map, axes: engine RPM and pedal position (%). The pedal position is linearised, and this map — Driver’s Wish (DAMOS folder AccPed_DrvDemDes, map AccPed_trqEng on EDC16/EDC17/MD1) — returns the torque request in Nm.

Example: in an EDC17CP02 on a 2.0 TDI, at 2000 RPM and 50% pedal, the request is around 180 Nm.

If you raise these values — especially in the lower pedal range — the car feels more responsive, but fuel consumption rises. And if you go above the torque limiter, nothing changes at the top end. Maximum torque and power stay the same, because the limiter catches it.

I often see obvious mistakes in other tuners’ remaps — even Driver’s Wish alone is enough to ruin the car. Push torque above a logical value and you get “pedal compression”: the car responds to the pedal up to about three-quarters travel, then it goes dead. A seemingly small mistake, but it raises fuel consumption. Why? Because the pedal is “shorter” and therefore less precise.

You could keep listing examples. Raising the Torque Limit too early, causing a puff of smoke at full throttle. Changing Duration too much and decalibrating the dashboard fuel-consumption readout. And so on — hundreds of small and large errors. Sometimes they add up into one big one that the engine cannot win. And then: failure.

Torque limiter

Now the Torque Limiter. Input: engine RPM. Output: maximum allowed torque.

For EDC17, the scaling is usually ×0.5 for RPM and ×0.1 for Nm. Factory limits are set so the torque stays compatible with the gearbox, the rest of the drivetrain, the way torque reaches the wheels, and emission standards. The limiter is actually split across two protection maps on EDC17: engine mechanical protection (DAMOS folder EngPrt_PrtLim, maps EngPrt_trqLim for torque and EngPrt_qLim for fuel quantity) and gearbox protection (Gearbx_TrqLim / Gearbx_trqDrive). Both fire at key-on and during runtime.

Raise the limiter and yes — you can request more torque. But injection duration, FMTC, and most importantly the air management system must be able to deliver it. Fuel rail pressure must hold its level even with longer injector openings. Pre-injection must be corrected so the larger fuel charge mixes properly with the larger air charge. Get any of this wrong and your torque request is physically unachievable — or, put simply, the engine and the injection system’s physical limits cap it for you.

When I look at other tuners’ files I think — “one map, but so many questions, and often no correct answer.”

FMTC — fuel measure torque converter

Next in the information flow: FMTC (Fuel Measure Torque Converter). This is a critical map. It converts the approved effective torque (Nm) into fuel mass (mg per combustion cycle). In Bosch DAMOS this lives in folder PhyMod_GenCur with map PhyMod_trq2qBas — but later Bosch revisions use the alias FMTC_q2trq / FMTC_trq2qBas for the same logic. Both names refer to the exact map you are looking at. Scope: EDC16, EDC17, MD1.

Here you have to make a decision. You can extrapolate the map, keeping the torque request realistic. You get accurate CAN readings — but every undetected torque limiter in the file will fire. The extrapolation has to be linear and clean, so the operating parameters do not drift apart.

You can also decalibrate the map, which is often the better option if you do not know where all the limiters live. Most tuners use a hybrid: part of the Nm request is real, part comes from FMTC decalibration.

That is the theory. In practice, a mistake here means the engine gets either too much fuel (smoke, DPF clogging) or too much air (wasting the turbocharger, dropping efficiency, and therefore raising fuel consumption).

Injection duration

Now the data goes to the Injection Duration map (DAMOS folder InjVlv_GetET, map InjVlv_tiET on EDC16/EDC17/MD1 — ET = effective time). Inputs: fuel mass (often stored in cubic millimetres instead of milligrams) and fuel rail pressure. Output: injection time in ms or µs.

This is the central map of injection control. It does not just decide the main injection time — it also decides pre-injection time (critical for the system) and post-injection time. All three use it.

Extrapolation is often needed, especially if you raise rail pressure. The extrapolation is square-root (it comes from Bernoulli’s law), so WinOLS’s linear interpolation is wrong — it leads to serious decalibration.

The fuel axis is in cubic millimetres (mm³). A litre of diesel weighs 0.835 kg. If you do not apply this conversion, your fuel-mass-to-duration error is around 20%. A very common mistake — forgetting the mm³ → mg conversion (factor 0.835).

What does this conversion mean physically? It leads to a serious mismatch between the real engine load and the value the ECU actually picks from the map — which then drives the injector opening time. Tuners often get confused here because some ECUs store milligrams on this axis, not cubic millimetres. Always check which unit the axis is in before you touch anything.

And just like FMTC, bad extrapolation or decalibration here shows up as exhaust smoke or excess air. You also have to remember that the errors from FMTC and Duration stack multiplicatively. A +5% change in fuel mass and +5% in Duration do not add to +10% — it is slightly more. A common downstream error: bad AFR (too rich).

Concrete example: at 1600 bar rail pressure and 30 mg fuel mass, the injection duration in the ECU is 473 µs. That is the kind of anchor number you carry in your head after working enough CP02 files.

Injection precision: rail pressure, SOI, pre-injection

You have now covered the main fuel loop. Let us zoom in on injection precision itself. Common rail pressure, Start of Injection (SOI), and pre-injection decide how combustion unfolds.

Tuners often obsess over fuel quantity and ignore injection timing and pre-injection. They think about the mixture, not about how the mixture is prepared. That is a mistake. With a good AFR it is not a huge mistake — but if you want the ideal calibration, you cannot skip it. You get Stage 1. You could have had Stage 1+.

Rail pressure

Higher rail pressure (DAMOS folder Rail_SetPoint, map Rail_pSetPointBas on EDC16/EDC17/MD1) improves fuel atomisation. That means better combustion — same fuel mass, more power.

We cannot raise pressure freely. Manufacturers have already pushed common rail pressure up to meet stricter emission standards. Raising it further without understanding the pump means rapid pump wear, and eventually a mechanical failure. The upper safety ceiling lives in a separate map — Rail_SetPointAddCor / Rail_pSetPointMax — which caps peak pressure as a function of RPM and fuel quantity.

It is also worth looking at the maximum pressure used across generations. For EDC16 it is 1600 bar. First-generation EDC17 runs 1800 bar. Newer EDC17 variants hit 2000 bar and above, and MD1 goes higher still. Each generation raised the ceiling to meet emission norms — which means the hardware margin on top shrinks with every generation.

There are also rail-pressure limiter maps, and they are usually a function of fuel temperature. Pushing these blindly to maximum can have serious consequences — up to seizing the fuel pump when fuel temperature is too high. If you do not know what the limiter map is doing, do not touch it.

Start of injection (SOI)

Start of injection — for both main injection and its pre-injections — is defined by the crankshaft angle relative to TDC. In Bosch DAMOS, the main injection SOI lives in folder InjCrv_MI1, map InjCrv_phiMI1Bas on EDC16/EDC17/MD1 (phi = angle, MI1 = main injection 1, Bas = base). This is the map you edit. The phase-split timing maps nearby are not the SOI map — don’t touch those.

Sign convention: a negative value is injection delay relative to TDC; a positive value is injection advance relative to TDC.

Rule of thumb: 1° of advance ≈ 1% more efficiency. Sounds good, but is it always true? Can we improve mixing with pre-injection and advance combustion that way — getting the same or better effect without side effects like louder engine noise?

Over-advance raises NOx. Over-retard means power loss, more soot in the exhaust, and higher exhaust temperature (because part of the combustion energy goes to the exhaust instead of doing work).

We have to balance the desire for efficiency against the engine’s hardware limits and noise. Classic protection zone.

Real-world example: a typical SOI for a 2.0 TDI CR (EDC17) at full load and 3000 RPM is about 10° advance. Shift it by 5° (to 15°) and you get roughly +5% performance — but also about +10 dB more engine noise. (That depends on the engine’s sound insulation and how drive is transmitted — AWD spreads noise through the body more, so the extra sound is more noticeable.) You have to find the compromise.

Pre-injection

Pre-injection is the Holy Grail of tuning. A lot of people talk about it, almost nobody modifies it. The knowledge just is not there.

Typical pre-injection dose: 1–2.5 mg per combustion cycle — stored in DAMOS folder InjCrv_PiI1, map InjCrv_qPiI1Bas (q = fuel quantity, PiI1 = pilot injection 1) on EDC16/EDC17/MD1. It reduces combustion noise while raising performance. It also raises peak combustion temperature for a moment.

Worth modifying — if you want more than a plain Stage 1. A truly powerful method. Very underrated and poorly understood.

In DENSO ECUs, you will find a noticeably different approach to pre-injection setup, which delivers higher engine efficiency and lower fuel consumption. Worth studying how DENSO does it.

The air loop: turbo, VNT, MAF

You have added fuel. Now the engine has to burn it — completely. The loop that controls air quantity and variable turbo geometry takes over. This is where air mass and VNT position are set.

Boost target (target boost pressure)

Target boost pressure (DAMOS folder PCR_DesValCalc, map PCR_pDesBas on EDC16/EDC17/MD1) — the target pressure in the intake manifold, in millibars, measured relative to vacuum.

How to read real values: a map value of 2200 = 2200 mbar absolute = 1200 mbar of boost (2200 − 1000 = 1200 mbar above atmospheric). On your boost gauge: 1.2 bar. Simple as that.

Air mass is tightly linked to good combustion. Too little air → black smoke and a cascade of bad consequences for the DPF, turbo, and engine. Too much air → efficiency drops.

The turbo can only deliver a specific mass of air. And if you drive up into the mountains, that mass goes down. You cannot compress thin air up to the same boost pressure safely — the turbo has to spin faster. You are not allowed to “patch away” the turbo’s limit and correction maps and just set everything to maximum. Because your customer’s breakdown is going to happen on their ski vacation in the Alps. You really do not want that.

Of course, factory boost is already high for emissions reasons. We cannot control that directly — but we do control it indirectly. Set boost correctly and fuel consumption falls while power rises. Set it wrong and you are beating up either the turbo or the whole engine for nothing.

VNT (variable nozzle turbine)

Now the critical point — correct setting of variable geometry (VNT / VGT). We control the position of the exhaust-side vanes on the turbo rotor (PWM %) via DAMOS folder PCR_CtlValCalcMdl, map PCR_rCtlBas on EDC16/EDC17/MD1. 100% = 8192. You have to scale it correctly.

Rule #1: do not skip this map. Rule #2: edit downward, not upward. After tuning, you have more exhaust. More energy hits the turbo rotor. Factory settings will be too aggressive.

You will say — but there is autoregulation. The ECU will correct the vane position itself…

I wish. The problem: the air control loop reacts incredibly slowly compared to the fuel loop. If the VNT map is not accurate out of the gate, the problems come fast. Usually a turbo failure — or at the very least, unpleasant, jerky engine behaviour, strange vibrations, sometimes an overboost fault.

There is a second, valuable takeaway from this map: you can read the turbo’s reserve from it. Is it a generously-sized unit — or did the carmaker save money and fit, effectively, a cheap hair dryer instead of a real compressor? That is a joke. But this map tells us a lot.

This is also where turbo lag gets decided. A skilled trick and you can kill most of it.

MAF — mass air flow

Now the MAF sensor — the mass air flow meter. It measures the actual mass of air entering the engine. Not a target. A measurement.

Many tuners believe that MAF decides the air/fuel mixture. Nothing could be further from the truth. From the Clapeyron equation (ideal gas law), air mass is calculated a different way — using the MAP sensor. MAF is used for EGR control. On the target side, the ECU expresses desired MAF via AirCtl_DesValCalc / AirCtl_mDesBas (AirDesire), and the EGR closed-loop uses hysteresis bounds AirCtl_qHi and AirCtl_qLow in folder AirCtl_Monitor.

Tuners do not modify MAF maps. But with a little knowledge, we could pull a lot of information out of them — without changing a byte. For example, the real air/fuel ratio at full load with EGR closed.

The real AFR comes from the actual manifold air flow, calculated as (MAP × RPM) ÷ fuel mass. MAF can only be used for this calculation in the region of the map where EGR is closed — otherwise part of the air you are measuring is recirculated exhaust, and the number is meaningless. That is why MAF is a diagnostic tool for the calibrator, not a control parameter to tune.

Putting it all together: the +21% Stage 1 strategy

Our foundational course — Stage 1 — modifies a car by +21%. I picked this target on purpose. Not 20%. Twenty-one. Because I want students to learn to calculate very precisely. If you round to 20%, you are already sloppy.

The course teaches the minimum map set for a modification of this size, and how information flows through the ECU. The minimum set:

Driver’s Wish → Torque Limiter → FMTC → Injection Duration + Boost Target + VNT.

This principle applies to every ECU family and every manufacturer: Delphi, Siemens SID, Bosch MD1, DENSO, Bosch EDC17. The map names change. The physics does not.

There are of course differences between controllers — some maps are organised differently, some use different scaling. The EDC16 and MD1 families are the easiest for students because they are very close to EDC17. DENSO and the newest Delphi controllers surprise students a bit — once you have learnt the physics the method still applies, but you need to re-learn where things live.

We do not change SOI, pre-injection, or rail pressure. That is the Stage 1+ goal — and to get there you first need to understand the physics of combustion and engine control itself. Leave it for later. Not everything at once.

That is the philosophy of development: start with the system you understand, not the map someone told you to change.

This approach is safe and leaves no gaps in your personal growth as a tuner / ECU calibrator. In the Shaolin monastery, you do not become a master on day one. But you can be a skilled student who knows a lot of strong techniques. You become a master with time and practice. Exactly like in the film.

This is what separates us from the people who offer you “buy a mappack or a DAMOS and we will tell you where to change the maps and by how much.” Or from the approach where you buy a file and then try to copy it to the next similar car.

“Two tuners, same car, same EDC17, same WinOLS. One rounds to 20% and hopes. The other calculates 21% and sleeps. The hardware does not care which of them you are — but your customer does.” — Thomas Pirowski, 30+ years in ECU reverse engineering

Where to go from here

Understanding the EDC17 maps as you have just read them is Level 2 work. You know the physics. You know the fuel loop order. You can produce a safe +21% Stage 1 on the cars you have been trained on, and you can explain to a customer why more than that is not safe with the factory turbo.

Level 3 is different. Level 3 is reading the ECU’s code — not just its data — to find maps no DAMOS ships with, to understand the algorithms that process your values, to build features the OEM did not ship: MapSwitch, VIN protection, custom limiter bypass logic, launch control. At that point you are not calibrating anymore. You are reverse-engineering.

Most tuners never need Level 3. Stage 1 and Stage 1+ work on EDC17 pays the bills for years. But if you want to work with ECUs where no DAMOS exists — MG1, newer MD1, Delphi latest-gen — or if you want to offer premium services no other shop in your region can, Ghidra is the way in. The methodology you just learnt transfers directly. You stop looking at the numbers and start reading the logic that writes them.

Between Level 2 and Level 3 is the gap most tuners live in. The Learning Path shows the steps, the prices, and what each level makes you capable of doing.

DAMOS map reference — every map in one table

For tuners using a DAMOS with EDC17 or MD1 binaries, here’s the exact folder + map identifier for every concept discussed above. Scope: EDC16, EDC17, MD1 (the Bosch diesel platform from 2003 onwards). The final column links to our community Knowledge Base, where each map has its own workflow post with step-by-step WinOLS^® editing instructions and real-case troubleshooting.

Concept	DAMOS folder	Map identifier	Deep dive
Driver’s Wish (torque request)	AccPed_DrvDemDes	AccPed_trqEng	KB — no-start after driver wish edit
Engine mechanical torque limit	EngPrt_PrtLim	EngPrt_trqLim	—
Engine fuel quantity limit	EngPrt_PrtLim	EngPrt_qLim	—
Gearbox torque limit	Gearbx_TrqLim	Gearbx_trqDrive	—
FMTC torque-to-fuel conversion	PhyMod_GenCur (alias FMTC_q2trq)	PhyMod_trq2qBas / FMTC_trq2qBas	—
Injection duration (ET)	InjVlv_GetET	InjVlv_tiET	—
Rail pressure base setpoint	Rail_SetPoint	Rail_pSetPointBas	KB — rail pressure +10% safe zones
Rail pressure maximum	Rail_SetPointAddCor	Rail_pSetPointMax	KB — rail pressure +10% safe zones
Main injection SOI (angle)	InjCrv_MI1	InjCrv_phiMI1Bas	KB — SOI Italian Highway zones
Pre-injection 1 fuel quantity	InjCrv_PiI1	InjCrv_qPiI1Bas	KB — pre/post-injection 4-tell method
Pre-injection map encoding (8-bit on CP44)	InjCrv_PiI1	InjCrv_qPiI1Bas (8-bit values, 16-bit axes)	KB — finding pre-injection 8-bit on EDC17 CP44
Boost pressure target	PCR_DesValCalc	PCR_pDesBas	—
Boost pressure protection limit (Protection / error tables)	PCR_Monitor	PCR_pDesMax	KB — setting 0.5 bar boost limit on VW MED17
Boost request encoding traps (BMW C41 factory bugs + C46→C64/C74 substitution)	PCR_DesValCalc / PCR_VGT_Mdl	PCR_pDesBas (mbar) or PCR_chrgDesBas (mg/cycle)	KB — EDC17 boost-map traps
VNT / VGT vane position	PCR_CtlValCalcMdl	PCR_rCtlBas	—
Air mass target (AirDesire)	AirCtl_DesValCalc	AirCtl_mDesBas	KB — AirDesire Stage 1 on EDC17C74
EGR hysteresis upper / lower	AirCtl_Monitor	AirCtl_qHi / AirCtl_qLow	—
Smoke limitation lambda	FlMng_InjMassLim	FlMng_rLmbdSmk	—
Torque-loss runtime monitoring (limp mode)	MoFLos_Co	MoFLos_trqEngLos	KB — Layer 1/2 torque monitoring

Names and folder paths may shift slightly between Bosch revisions — later EDC17 and MD1 firmware sometimes use FMTC_ prefixes instead of PhyMod_ for the torque-to-fuel path, and AirDesire lives in different sub-folders across Euro 5 vs Euro 6 platforms. When you find a map with a near-identical name, check the axes and factorisation before assuming it’s the same function. Physics first. DAMOS strings second.

Frequently asked questions

About the author: Thomas Pirowski — 30+ years in ECU reverse engineering, university lecturer at AGH Kraków. Co-founder of Tuners Guild, where he writes the technical deep-dives that teach calibration methodology instead of interface memorisation.

Related: What Is DAMOS? · WinOLS vs. ECM Titanium · WinOLS Training Methodology · Diesel Fundamental Course · TriCore Reverse Engineering · Learning Path