Idle, cold start and injectors on Hondata KPro and FlashPro

This guide covers port-injected K20 and K24 engines: cable-throttle KPro with an idle air valve, and DBW FlashPro with throttle-controlled idle air.

K20C, L15 and K24W direct-injection engines, B-series, D-series and s300 are outside this guide. Their fuel systems or software controls differ. FuelTech on a K has its own fuel model and guide.

Separate no-fire, catch-and-stall, warm-up and hot-restart complaints before choosing a table.

A log that begins after the engine is running cannot show cranking voltage or time to fire. A short warm-idle clip cannot show a cold-soak transition. Record from before the key to at least the point where the engine holds a steady idle.

If the complaint is a stall, the useful part of the log is the seconds before it, not the minutes after.

One reading is not a diagnosis. A lean wideband during a rough idle can come from a misfire or an exhaust leak. Oxygen that did not burn shows up as excess oxygen in the exhaust. Adding fuel to chase that reading makes the mixture richer without fixing the misfire.

Check pulse width, voltage, combustion and exhaust integrity before you move a fuel cell.

An injector has a flow rating at a pressure, a dead time against voltage, a short-pulse behavior and a spray pattern. The software scalar covers the flow rating. It does not cover the other three. A larger entered flow shortens the pulse and leans the mixture.

A smaller entered flow lengthens the pulse and richens it. Either change lands on every load and on cranking, which is why a size change can look right at one load and wrong at idle.

More details

In KManager, changing the injector size compensates the fuel tables and the cranking fuel. It does not compensate the specific characteristics of the injector. The voltage compensation table is the dead time against battery voltage. In FlashProManager, the size change compensates main fuel, cranking fuel and closed loop.

The flow is nominally at 3 bar (43.5 psi). Where exposed, the stock/current fuel-pressure and injector-test-pressure fields compensate main fuel, cranking and closed loop. Do not also hand-scale those tables for the same pressure change.

Static flow can be estimated with the square root of the differential-pressure ratio. Take a 1000 cc/min injector rated at 3.0 bar as an arithmetic example. It moves 1080 cc/min at 3.5 bar. The arithmetic is 1000 × sqrt(3.5 / 3.0). That is arithmetic for the same injector, not a recommendation to raise pressure. The regulator reference matters. A manifold-referenced regulator holds differential pressure across the injector. A fixed-pressure system does not, so the same rail gauge number is not the same delivered flow.

More details

The sheet rates this part at 1000 cc/min at 51 psi / 3.5 bar differential pressure. Its calculated flow at 3.0 bar is about 926 cc/min. If FlashPro exposes a current-injector test-pressure field, enter 1000 with 3.5 bar there and the actual operating pressure separately.

Alternatively, use the converted 926 figure with a 3.0 bar test pressure. Do not convert the flow and also leave 3.5 bar as its test pressure.

• Keep the other pressure settings unchanged. Entering 1000 instead of 926 against a 3.0 bar test pressure shortens the calculated delivery pulse by about 7.4%. The latency portion is separate. The DW dead-time row remains data at 3.5 bar; converting static flow does not convert latency to another pressure.
• For a simplified example, a 2 ms total command with 1 ms dead time leaves 1 ms of effective delivery. A 0.1 ms error is 10% of that window. A 10 ms total command leaves 9 ms; the same error is about 1.1%. Real short-pulse nonlinearity can depart from this subtraction model, so it cannot generate a missing injector curve.

1. Confirm the part and the pressure. Record the injector part number, the rated pressure, the rail pressure and the regulator reference. Check for leaks, excessive pressure, a leaking injector and the wrong impedance or driver arrangement.
2. Enter flow at the pressure the software uses. Then enter the latency curve for the exact part. Keep the data sheet revision and date with the calibration.
3. Verify the measurement path. Check wideband scaling and warm-up. Confirm the ECU recognizes closed throttle using the channels for that application. Fix vacuum leaks before you change fuel.
4. Warm the engine and log two steady points. Use idle and a steady part-throttle point. Compare the error at short and long pulse widths. A similar percentage at both points points to flow or pressure. An error that is large only at idle points to dead time, an air leak or combustion.
5. Run the electrical-load test before you trim the size scalar for an idle-only error. The next section is the method.
6. Make the smallest change and repeat. A dead-time change moves delivered fuel at every load. Recheck the fuel map after it, not only the idle.

More details

Injector phase is a separate setting. In KManager, the phase values determine the end of the injection event, when the injector closes. The same help gives the typical opening point at 400 to 500 crank degrees before TDC compression.

There are four phase values: idle, cruise low load, cruise medium load and wide-open. The first three are fixed; the wide-open values are indexed by rpm. A compensation table moves the event with the intake cam angle. Phase can change mixture formation.

It is not dead time, and it does not prove the spray is right.

Air determines the charge available; spark changes the torque it produces. Both help hold idle speed. Fueling must support stable combustion while those controls work.

More details

On a cable-throttle KPro car, the idle air valve meters air past the throttle plate. The hot target is a speed in rpm. The duty-cycle adjustment is a proportional trim on valve feedback. The KManager idle page says to increase the slider if the idle dips when you release the throttle.

Decrease it if the idle hangs high. That is after the mechanical checks, not before them. The same page notes that large injectors may need a richer hot-idle target to prevent lean misfires.

The Disable idle valve switch is an error-check switch. It does not seal the air path and it does not tune idle airflow. If the valve is removed and the passage is open, the engine gets air the ECU did not command. A block-off plate is the mechanical close. The software switch only stops the error check.

On a DBW FlashPro car, the throttle plate is the air actuator. There is no idle valve to clean or delete. Two coolant-indexed tables set the target: an after-start speed and a normal speed. The after-start target is used for 20 to 60 seconds, then the normal target takes over.

That is a target change, not a fuel blend. Do not tune it as if it were cranking fuel.

Idle spark moves to control torque. The FlashPro ignition page gives a normal idle range of −10 to +15 degrees. That movement can be normal when actual rpm follows the target. Persistent retard with overspeed still calls for an air-path check. That band is a FlashPro observation, not a KPro limit.

KManager's ignition-controlled idle uses target-versus-actual speed error tables, so use the table in the calibration you opened. Do not flatten the idle spark table to make the log look steady. Flattening it removes the fast correction in both directions.

The FlashPro help also has an Idle Torque Reserve paragraph. That paragraph is explicitly about Bosch ECUs. Its 0 degree average is not a K-series table target.

Swipe or use the arrow keys to read every label.

1. Warm the engine fully and log the loop. Record target and actual rpm, TPS, MAP, idle valve duty or throttle command, spark, battery voltage and lambda.
2. If the actual speed is high with the actuator at minimum, find the air. Check the throttle stop, cable tension, bypass passage, manifold and throttle gaskets, PCV and brake-booster plumbing. A lower target cannot remove uncommanded air.
3. If the idle dips on return, confirm closed throttle first. Check closed-throttle recognition and that the valve or throttle has capacity. Then adjust the air trim. A dip is not automatically a target problem.
4. Apply the loads the car actually has. Fan, headlights, demister, steering and A/C. Note whether the sag starts with the air actuator, a voltage change, a mixture error or spark correction at its limit.
5. Settle the fuel before you chase the last few rpm. A misfire can make a wideband read lean. More fuel is not the answer to a misfire, and a steady mixture is the foundation for the air and spark check.
6. Confirm hot restart and repeated deceleration to idle. A dyno idle at zero accessory load is one operating point, not the whole job.

A flow error tends to remain a similar percentage at different pulse widths. A dead-time error becomes more visible as delivery time gets shorter. Electrical loads can expose a voltage-related error, but alternator torque also changes the air needed to hold idle.

Compare settled rpm, MAP and pulse width before blaming the curve.

With valid closed-loop feedback, a trim that becomes more positive at lower voltage means the ECU is adding fuel. Too little entered dead time is one possible cause. A more negative trim means it is removing fuel, which can point to too much entered dead time.

Compare the change from the unloaded baseline, including learned correction. In open loop, compare measured lambda with the target instead.

1. Baseline warm, accessories off. Let voltage settle and record lambda, trim, pulse width, MAP, rpm and battery voltage.
2. Add one load at a time. Fan, headlights, demister. Keep rpm and MAP as close as you can, then record the new voltage and the new lambda and trim.
3. Repeat at a steady part-throttle point. Similar errors suggest flow or pressure. An error that shrinks at a longer pulse suggests dead time, but air leaks and short-pulse behavior can give similar symptoms.
4. Correct the region actually used. The logged voltage may interpolate between two cells. Keep the curve smooth, change only the justified region and repeat both load states.
5. Recheck idle and low-pulse cruise. A dead-time change that fixes idle also changes delivered fuel at cruise. Recheck the cruise trim.
6. Recheck the fuel map. Dead time changes delivered fuel at every pulse width, so the map has to be checked after the latency change.

A start is a sequence, not a table. The ECU uses a cranking calculation, then blends to the main fuel calculation. It moves from an after-start idle target to the normal target, then applies warm-up compensation. Fuel blending follows elapsed time; running compensation follows temperature.

A change in one phase can worsen another.

Those thresholds and timers describe the documented FlashPro starting strategy. Read the KPro start settings from the actual KManager calibration.

• The fuel blend can finish while the after-start idle target remains active. Log target and actual rpm together. If the target steps down and actual speed follows, inspect the target transition before cranking fuel.
• The FlashPro starting-fuel page gives two cranking checks for that strategy. Battery voltage during crank should be about 10.5 V and should not drop below 9.2 V. Cranking speed should be over 150 rpm cold and about 200 rpm hot. Those are checks for that starting strategy, not proof that a starter or battery passes every electrical test on a swap.

1. Begin from an overnight soak. Record ambient, coolant and intake-air temperature before the first key cycle. Record the fuel blend if it is known.
2. Log before you crank. Capture minimum voltage, crank rpm, injector pulse, time to fire and time to a sustained idle. Note the coolant temperature at each event.
3. If it never fires, stop adding fuel. Verify sync, spark, fuel pressure and the electrical cranking conditions first. A no-fire is not a cold-enrichment problem until those are sound.
4. If it catches and stalls, inspect the immediate blend and air capacity. A start that catches then dies after several seconds is not the same as one that never fires.
5. If it fails at one temperature, change that region. If it is hard at every cold temperature, change the cranking trim. A hot-only failure stays on the hot-restart checks: too much cranking fuel, voltage, pressure or vapor lock. Do not use a global change for a one-temperature fault or a hot start.
6. Repeat a genuine cold soak and check the hot restart separately. A restart after a few seconds is not a second cold test. Valves are warmer and the fuel film has changed.

Warm-up fuel is a compensation against coolant temperature, not a second main map. KManager warns that both air- and water-temperature compensation edits may wait for an ignition cycle and several minutes of running. The water reading is buffered to reject fluctuations from air pockets.

Follow the application service procedure if using the service-connector shortcut for that water-temperature response. If you edit and retest immediately, you can conclude a table does nothing when it has not been applied yet.

• Negative compensation is not automatically wrong. Read the baseline and the delivered-fuel arithmetic. The FlashPro help gives the exact form for a negative cranking trim:
• new trim = old trim + ((100 + old trim) × (change% / 100))

• At −70%, changing to −60% raises the remaining fuel factor from 30% to 40%. That adds 33.3% relative fuel, not 10%.
• For the documented FlashPro diagnostic, test +10%, +20% and +30% relative fuel changes from the starting baseline. If adding fuel worsens cranking, test −10%, −20% and −30% from that same baseline. These are separate tests, not successive multiplications. Use the formula above when the compensation already has a negative trim. Keep a temperature-specific fault in that region; do not use the global trim for a hot-only problem.

1. Confirm the warm fuel map and the injector model first. A cold-start change on a bad warm calibration is a guess.
2. Find the temperature where the fault occurs. Log coolant at the stall or the lean dip. Do not change a row you did not reach.
3. Change only that region. Keep the rest of the compensation curve and the other starts intact.
4. Cycle the key and run for several minutes before you judge. Both air- and water-temperature compensation edits can be delayed. An immediate retest can mislead you.
5. Repeat a cold soak and a hot restart. A change that fixes the cold start can hurt the hot start, and the hot start is part of the same delivery.
6. Do not use a global change for a one-temperature fault. It moves every start to fix one point.

A cam changes the air the engine wants at idle. More overlap lowers idle vacuum and makes the engine less tolerant of a lean mixture or a late spark. The target may need to be higher, and the idle air authority may need to be larger.

The cam target itself belongs to the VTC page. This page covers the idle air and the fuel that supports it.

• Raising the target can settle a cammed engine at idle. It cannot correct cam timing, a vacuum leak or a stuck valve. No stage table of idle speeds is printed here. Set a speed that holds stable with the actual cam, oil pressure, charging and accessory load. A target that only works with the lights off is not a finished idle.
• Big injectors make the short pulse the problem. A large injector at idle may be near its minimum pulse. Dead time, spray quality and cylinder-to-cylinder distribution matter more than the rated flow. A part with a clean spray and correct data can idle better than a larger part with a poor pattern. Do not enter a false size to chase an idle. That masks a pressure or model fault and changes the whole range.

• The K-series CAM channel help describes the intake cam as locked fully retarded for the first 10 seconds after startup. The VTC and VTEC guide covers that lock. The cam may therefore be at a different angle from its later warm-running command. Log the cam state during that window. Evaluate starting fuel there, then compare warm running after the cam follows its normal command.
• E85 changes cold-start behavior. Test at the actual ethanol content and the actual temperature. A fuel-flow correction alone does not prove the fuel vaporizes. The E85 guide covers blend measurement and separate start and running compensations.

None of these is a target for your engine unless the row says so. The application column is part of the number.