Why tip-in needs its own logic
Holding a steady throttle, fueling is simple: the ECU reads manifold pressure and RPM, looks up the VE and target lambda, and injects. The trouble starts when the pedal moves quickly. In that moment three separate things work against you, and each one tends to make the engine stumble or go lean for a moment:
- The manifold pressure reading trails behind the real pressure in the plenum.
- On a port injected engine, some of the extra fuel ends up wetting the runner walls instead of reaching the cylinder.
- Cylinders already on their intake stroke were given fuel sized for the old airflow.
The Nexus MAP Prediction / Fuel Film function tackles the first two directly, and correct injection timing handles the third. It is an alternative to the Transient Throttle (Classic) function, not an add-on, so you use one or the other. If you are used to the classic approach, our transient throttle article covers that method.
1. The MAP reading is late
Manifold pressure pulses every time a cylinder draws in a charge, so a single instant reading is useless for fueling. The ECU averages it over one or two cylinder periods, chosen with the Averaging Window setting under Engine Synchronous on the Sensors page of the tree. That averaging is what gives you a stable number, but it costs time.
Now picture a snap from idle. There is a huge pressure difference across the throttle plate, and with a ratio above roughly 2:1 the flow through the throttle chokes and moves at the speed of sound. The plenum fills far faster than the averaged signal can show. For a short while, the ECU is fueling off a pressure that no longer exists.
MAP Prediction solves this with a Predicted MAP table, indexed by RPM and throttle position, that holds the manifold pressure the engine settles at for each combination. When the throttle moves quickly enough, the ECU uses whichever is higher, the measured MAP or the predicted value, for the period set by Predicted MAP time. It trades a little precision for a lot of speed, which is exactly the right trade in that half second.
2. Fuel sticks to the port walls
With port injection, a share of every injection lands on the runner and the back of the valve instead of going straight into the cylinder. That wet layer evaporates and gets drawn in over the following cycles. When conditions are steady, fuel joining the layer equals fuel leaving it, so you never notice it and the base fuel map looks correct.
Change the load and the balance shifts. Higher load means the layer wants to be thicker, so part of the next few injections goes into building it up and the cylinder runs lean. Lift off and the opposite happens: the layer shrinks, releases its fuel, and the engine runs rich. The fuel film model keeps a running estimate of that layer and adds or removes fuel to cancel the effect. Two tables drive it: Film Pooling Percentage, which sets how much of the injected fuel is assumed to go to the wall, and Film Evaporation Time Constant, which sets how quickly it comes back off.
3. Some of the fuel is already committed
If idle injection is timed to finish while the intake valve is still closed, the cylinders whose valves happen to be open when you tip in get fuel for the old, low pressure and air for the new, high pressure. That is a lean hiccup no enrichment table can prevent, because the fuel was delivered before the ECU knew anything had changed. On a four-cylinder it affects one cylinder, and on a V8 it affects two.
The reliable fix is to move the injection event later, into the open-valve window, so the ECU is still deciding the pulse when the air changes. Haltech's guidance puts idle injection timing somewhere around 270 to 330 degrees BTDC, with about 300 degrees a common result, instead of the roughly 400 degrees typical of closed-valve timing.
A lean spike caused by early injection timing tempts people to crank up the enrichment until it goes away. It hides the symptom and leaves the engine too rich everywhere else. Get injection timing right first, keep pooling low, and only raise it once timing is optimized.
Setting it up in NSP
- Check the inputs. You need a working throttle position sensor (cable or drive by wire) and a MAP sensor, internal or external. Watch the TPS channel: it should sweep smoothly as you press the pedal and sit perfectly still when you hold it.
- Swap functions. In the Fuel Tuning menu, disable Transient Throttle (Classic) and enable MAP Prediction / Fuel Film. Click the underlined function name to open its settings.
- Start from defaults. If the map has been edited and you want a clean slate, use Restore Defaults in the File menu and pick the MAP prediction / fuel film settings.
- Set idle injection timing for the best response off idle, in the open-valve range described above.
- Let the ECU learn the Predicted MAP table, then review it.
- Adjust pooling and evaporation last, only once everything above is right.
Long term predicted MAP learning
Filling the predicted table by hand is tedious, so the ECU can learn it. The trick relies on two rules that always hold: at a fixed RPM, opening the throttle further never lowers manifold pressure, and at a fixed throttle opening, higher RPM never raises it. So whenever the ECU takes a real MAP sample outside of a prediction event, it knows every cell at higher RPM and lower throttle must be at or below that value, and every cell at lower RPM and higher throttle must be at or above it. It nudges those cells into line. Learned values are capped at 101.3 kPa absolute (about 14.7 psi, one atmosphere) unless the aspiration method in Engine Configuration is set to Supercharged.
With learning on you rarely need to edit the table yourself. After some driving, or after visiting the cells during normal tuning, open it and make sure the values match the manifold pressure the engine actually reaches at each RPM and throttle combination.
The settings, and where we start
| Setting | What it does | Our advice |
|---|---|---|
| Transient TPS Scaling | The throttle rate (TPS-d) at which full prediction is used. Slower movements get proportionally less, and below 25% of this value there is no prediction at all. | About ten times the TPS-d noise you see with the throttle shut. If it wanders around 10%/sec at rest, try 100%/sec. |
| Predicted MAP time | How long the prediction is held after it triggers | Around 200 ms. Longer if the MAP hose is long. Don't share the MAP hose with gauges or other sensors. |
| Predicted MAP table | Expected steady manifold pressure at each RPM and throttle point | Let learning fill it, then sanity check against logs |
| Film Pooling Percentage | Share of injected fuel assumed to land on the walls | As low as feels good. Warm gasoline engines with a well matched injector and port can sit near 5-10%. Awkward ports, such as a side-port rotary or a single-spray injector hitting the divider in a two-valve port, often need 25-30%. Cold engines can need around 50%. |
| Film Evaporation Time Constant | How long the wall fuel takes to evaporate, so how long the correction lasts | Roughly 200 ms warm, up to about 400 ms cold |
Pooling and evaporation both climb when the engine is cold for the same reason: cold fuel evaporates less readily, so more of it stays on the walls and it hangs around longer.
Tuning by symptom
| Symptom | Usually |
|---|---|
| Predicted MAP never goes active, or only on a violent stab | Transient TPS Scaling set too high for your sensor. Confirm the TPS is healthy, log TPS-d and rescale to about ten times its resting noise. |
| Predicted MAP goes active while holding steady throttle | A noisy TPS or scaling set too low. Check the sensor and wiring, then raise the scaling. |
| Small hesitation or hiccup as you tip in | Predicted table doesn't match real steady-state MAP, the base fuel map is off at steady state, or idle injection timing is too early. Around 300 degrees BTDC usually gives the injector time to deliver at idle while staying late enough to react. |
| Rich 50 to 150 ms after the throttle moves | Too much film pooling. Bring it down. |
| Lean on the transition | First confirm the predicted table, the steady-state fueling and injection timing. Only then raise pooling. |
Channels to log
- Predicted MAP active: whether prediction is in use right now. It reads "Used because of MAP sensor error" during a sensor failover.
- Unfiltered Manifold Pressure: the raw sensor signal with no averaging.
- Measured Manifold Pressure: the sensor value averaged over one or two cylinder periods.
- Manifold Pressure: the value the ECU actually uses, which can be measured, predicted or a blend.
- Fuel - Film Model Correction: how much the film model is changing fuel, as a percentage. A reading of 10% means 10% more fuel than steady state would call for.
- TPS-d: throttle rate, which you need to set the scaling.
A good tip-in log shows Manifold Pressure jumping ahead of Measured Manifold Pressure, the two converging within the prediction time, and lambda staying close to target through the whole event with a brief positive film correction.
If the MAP sensor fails
When the MAP input has a low-level fault, such as a voltage out-of-range code or no input assigned, the ECU fills the MAP channel from the predicted table. That is enough to limp a naturally aspirated or supercharged car around gently, but it will not let a turbo car make boost. Closed loop O2 control switches off during the failover and reports "MAP sensor failure" as the reason. It also pauses whenever prediction is active and reports "Transient throttle", so short gaps in O2 control during tip-in are expected.
FAQ
Can I use this together with the classic transient throttle?
No. MAP Prediction / Fuel Film replaces it. Disable the classic function when you enable this one.
Do I have to fill in the Predicted MAP table myself?
Usually not. Turn on long term learning, drive the car through its normal range, and check the result. Hand editing is only needed to tidy up areas the engine rarely visits.
My steady-state fueling isn't finished yet. Should I tune transients now?
Get the base map and steady-state fuel tuning right first. The transient model builds on those numbers, and long term fuel trim and short pulse width corrections are easier to judge once tip-in is clean. If you'd like a second set of eyes on your logs, book a free consultation.
