Boost control on a K20 or K24 with Hondata and FuelTech

The procedure applies to port-injected K20 and K24 engines on Hondata KPro, FlashPro and FuelTech FT450, FT550 and FT600. This guide covers port-injected K20/K24 only. The K20C in the FK8/FL5 Type R and the L15 1.5T are open for Full Custom. This procedure does not apply to them.

The K24W direct-injection Accord, B-series, D-series, H-series and s300 are outside. The s300 A11 output and its 1% then +5% duty steps are s300 material. Do not paste that pin or those steps onto a KPro or FlashPro.

MAP is absolute pressure. Boost is the amount above ambient. At 101.3 kPa ambient, boost psi = (MAP kPa − 101.3) × 0.145038. A 200 kPa MAP reading is 14.3 psi boost, not 20 psi. Use the local key-on engine-off MAP as the ambient reference when altitude or weather is unusual.

Both systems drive a solenoid that changes pressure on the wastegate diaphragm. The difference is which quantity is in the feedback loop. A Hondata KPro or B2/B16 output commands duty. A FuelTech BoostController commands dome pressure. The manifold pressure you see on a gauge is a separate measurement in both cases.

Swipe or use the arrow keys to read every label.

• KPro's gear-based path uses Boost vs Gear to choose a target pressure, then Pressure vs Duty Cycle to choose duty. The second table is a measured characterization. It is not a feedback loop that holds manifold pressure.
• FlashPro's B2 and B16 paths name duty. Fixed duty gives a fairly flat curve. Duty cycle by Rpm & Gear gives a table per gear. Neither table is a manifold-pressure target.
• CANBoost is the exception. It has a target boost table in bar and a duty feed-forward table. PID can add or subtract duty until MAP reaches the target. That is a manifold-pressure loop, and it is only on the CANBoost path described in the 2012–2014 Civic Si help.

One log can tell you which quantity the controller is holding. Log MAP, the command and dome pressure where fitted. Add rpm, gear or speed, throttle, lambda and target lambda. Add injector pulse or duty, fuel pressure where fitted, intake and coolant temperature, and knock or timing retard.

The KManager documented CSV list includes time, map, air and coolant temperature, lambda and target lambda. It also includes injector, ignition, cam angle and target cam angle, rpm and throttle. Load and short and long term trim are listed too. It also lists knockcount. It does not promise boost duty, fuel pressure or every knock signal, so inspect the actual export header.

• FlashPro's sensor list is ECU specific. The CANBoost path can log target boost and duty where the calibration exposes them. The B2 path logs duty if the channel is present. FTManager logs are native .ftml with CSV export, and configured inputs decide what exists. MAP and dome pressure are separate inputs.
• A good trace repeats. The command is steady, and MAP follows with a known lag. Dome pressure holds where it is fitted, and lambda follows target while injector pulse stays stable. A bad trace changes the story: MAP climbs while dome is flat, target is met while duty climbs, or MAP rises at 0% command.
• The channel list is on the Honda datalog page.

KManager runs boost from the purge output. Configuration selects normally open or normally closed, minimum and maximum duty, activation pressure and frequency. Fixed duty is one command. Gear-based uses Boost vs Gear and Pressure vs Duty Cycle.

Fill the pressure/duty table from measured pairs at fixed duty, then let the lookup choose duty. Do not type a curve from another car.

1. Record the hardware and the references. MAP part and transfer, wastegate spring crack pressure, gate plumbing, injector part and data, pump, fuel pressure, wideband and ECU or software version. Stop if key-on MAP does not read local barometric pressure, or if the sensor part does not match the calibration.
2. Set the cut before the command. Hot and cold boost cut, below the sensor ceiling with margin, plus lean protection. Stop if the intended boost is above the sensor range or above the cut you can actually set.
3. Measure the mechanical baseline. For a conventional dual-port gate, connect the lower chamber directly to the pressure reference and vent the top. Isolate the controller pressure feed first. Log the resulting boost curve; it need not equal the spring label. Stop for uncontrolled rise, a missing pressure reference or an actuator that does not open on a regulated-air check.
4. Configure the output and the solenoid. KPro purge, FlashPro B2 or B16, or CANBoost. Set normally open or closed, minimum and maximum duty, frequency and activation pressure. Confirm the safe state. For B2/B16, stop if current reaches 1 A at 14 V, or the circuit has a relay or an added diode. Other drivers need their own current limits.
5. Characterize at a fixed command. On KPro, fill Pressure vs Duty from measured pairs. On B2/B16, fill the rpm and gear duty table. On CANBoost, turn PID off first and set target plus feed-forward. Investigate an unexplained change under matched conditions. Stop for a lean departure from target, fuel-pressure loss or new knock.
6. Add gear targets only after the gear is right. Check the speed signal and the displayed gear. Then set Boost vs Gear or the CANBoost target. Stop if the displayed gear does not match the selected gear, or if a boost-by-gear row moves the wrong gear.
7. Add compensations last. Use the compensations available on the selected route; throttle compensation here belongs to CANBoost. For quick spool, the help recommends 1 psi activation and pressures 1 or 2 psi below target. Stop if the overshoot crosses the cut or the boost does not repeat hot and cold.
8. Check commanded minimum and power loss separately. With the engine off, use regulated air and a pressure gauge on the control assembly. Verify displayed 0% gives the intended minimum-boost routing, then check the valve paths with power removed. Normally open configurations can feed the top chamber on power loss. A bench test verifies routing and actuator motion; it does not measure manifold boost. Keep the top chamber isolated from pressure during any engine-running electrical checks.

FuelTech separates three functions that are easy to call the same thing. A generic duty cycle output sends PWM. A boost activated output is an on/off switch based on MAP. Wastegate boost pressure control is the BoostController, and it regulates pressure at the top of the wastegate.

The generic duty output can run a boost solenoid the way a Hondata duty table does. The manual recommends 20 Hz for most PWM 3-way valves. The output can be scheduled by rpm, time, speed, MAP, gear and rpm, or gear elapsed time.

It is open loop unless you add a feedback device elsewhere.

1. Draw the pneumatic system. Pressure source, gate bottom, gate top, fill path, vent path, sensor and the failure position of each valve. Decide between generic duty and the BoostController. Stop if you cannot name the chamber and the failure position for every valve.
2. Allocate outputs and sensors. In dual-valve mode, use same-color outputs for the increase and decrease solenoids. Verify driver current and suppression from the FuelTech documentation. Keep MAP and dome pressure on separate inputs. Stop if dual-valve outputs mix colors or the dome sensor shares a line.
3. Calibrate both pressure channels. Key-on reference, then a regulated known pressure on each sensor. Stop if dome reads manifold, if either channel does not hold a known pressure, or if the sensor is not the type configured in FTManager.
4. Establish spring-only boost. Verify that the top chamber vents and the lower chamber has the intended pressure reference before the baseline run. Record the mechanical boost curve rather than expecting an exact match to the spring rating. Stop for uncontrolled rise or a failed reference line.
5. Characterize dome to MAP at one target. Use single-target mode and log both pressures. Stop if dome is flat and MAP climbs. Fix the gate, the plumbing or the backpressure before you touch gain.
6. Choose the mode and verify the timer. Time after 2-step, gear and rpm, gear elapsed time, single target or rpm. The time-based test works with the engine off. With the engine running, a valid launch requires 2-step plus TPS above 50% or reaching the 2-step limiter. Verify the launch state and timer reset before interpreting the curve. Stop if the intended mode never starts.
7. Set gain and the safety limits. Raise gain if dome is slow, lower it if it overshoots. Set max MAP and max MAP on 2-step as backstops. The limits do not cut fuel or spark; the pressure bounces around the setting. Stop for oscillating actual dome pressure or MAP exceeding the intended limit. Verify limit operation with suitable signal simulation; do not deliberately overboost the engine.
8. Recheck fuel and spark at the new boost. Lambda against target, injector pulse, fuel pressure and knock. Stop for a lean departure from target, falling fuel differential pressure or knock. An absent knock channel does not establish knock-free operation.

Boost does not create a fuel target. The target lambda table and the fuel model do. Read the active fuel model, the injector data and the differential pressure before changing a cell. On FlashPro the model can be AFM, MAP or hybrid. On KPro the raw values are not VE percent everywhere.

KManager boost cut has a cold limit and a hot limit. It cuts fuel and restores when the throttle is released or engine speed falls below 2,000 rpm. With a stock MAP sensor the maximum cut is 10.5 psi.

Set it below the sensor ceiling with margin, and set it before the first boost run.

• Lean protection uses Maximum Lambda, Engine Load and Trigger Time. It needs a working front wideband. It cuts and restores at throttle release or below 2,000 rpm. The help warns that lean protection is disabled after launch when the launch-control rev limiter is active. Verify this interaction in the installed version. These are conditions to verify on the car, not numbers to copy from a table.
• FuelTech's max MAP and max MAP on 2-step are safety limits. The manual says they do not regulate to a target and the pressure will bounce around the limit. They do not cut fuel or spark. Use them as a backstop, not as the boost curve.
• Configure FuelTech engine protection separately from the BoostController pressure limits. Overboost and fuel differential-pressure alerts can use a configured safe mode, including an rpm limit or engine shutoff. Verify the selected response and required sensor inputs in the actual file.
• A cut is a backstop, not a control. A normally open solenoid that loses power or wiring can go to maximum boost. A normally closed valve can return the system to its mechanical baseline when the plumbing vents the top and preserves the lower reference. Know which one you have, and test the failure state with the gate safe.
• With the engine off, test displayed minimum and power loss separately using regulated air. Confirm the pressure paths and gate motion. For engine-running electrical checks, isolate pressure from the top chamber first. Test protection inputs through suitable signal simulation, keeping the wiring and sensor transfer valid. A valve-routing test does not prove that the ECU's boost or lean cut works.