What long term fuel trim does
Short term O2 control reacts in real time: the mixture reads lean, so it adds fuel right now. Long term fuel trim (LTFT) is the memory behind it. As the engine runs, the Nexus notices which cells of the fuel map keep needing the same correction and stores that correction in its own table, so next time it starts from the right number instead of fixing it again.
That makes it useful for three things:
- Filling in areas of the map you couldn't reach on the dyno or during road tuning.
- Following slow changes as the engine wears.
- Absorbing small changes like a new intake or exhaust without a full retune.
It works with narrowband or wideband sensors, but we strongly prefer a wideband, such as a WB1 CAN wideband kit, because it can target any lambda, not just stoich. The trim table can be mapped up to 4D.
The main settings
| Setting | What it does | Default | Our advice |
|---|---|---|---|
| Enable Long Term Fuel Trim | Turns the function on and adds its tables to the navigation tree | Disabled | Enable once the base map and short term control are sorted |
| LTFT Max Enrich | The most fuel, in percent, that LTFT is allowed to add | 15 | Limits the damage if a sensor misreads |
| LTFT Max Disenrich | The most fuel it is allowed to remove | 15 | Keep this cautious; taking fuel away is the risky direction |
| Long Term Fuel Trim Rich Bias | Makes the learned value sit slightly richer than target. Short term control then pulls it back to target | 3 | Leave it on. Set 0 to disable |
| Long Term Fuel Trim Min Temperature | Coolant temperature needed before learning starts | 60 | Stops it learning cold-engine corrections into a warm map |
| Disable During Post Start Corr | Stops learning while the post start fuel correction is not 0% | Disabled | Turn on if start-up enrichment is skewing what it learns |
| Disable During Coolant Corr | Stops learning while the coolant fuel correction is not 0% | Disabled | Useful if you want it to learn only at full operating temperature |
Why the rich bias matters
If the O2 sensor fails or the system drops out of closed loop, the ECU falls back to whatever is in the base table plus the learned trim. With a rich bias, what it learned was slightly rich of target, so that fallback errs on the safe side. While everything is working, the short term controller trims out the extra, and you still see target lambda.
Learning speed: LTFT gain
The gain table sets how fast corrections get written into the trim tables, and it can be mapped in up to 3D. The default is 30. Raising it learns faster but can start to oscillate, overshooting and correcting back. Lowering it learns more slowly and steadily. We'd rather have slow and stable, especially on a street car that spends thousands of miles learning.
The trim tables
Bank 1 Long Term Fuel Trim (and Bank 2, if you run a second bank) holds the percentage correction learned for each cell. The axis channels are fixed to match the fuel base table, but you can change the breakpoint values. Changes to the axis values carry across to Bank 2 automatically. You can also edit cells yourself while the ECU is learning. It supports up to 4D.
Reset and Apply to Base Table
- Reset clears every learned correction back to zero. Do this after any change to the base map or hardware that would make old corrections wrong.
- Apply to Base Table writes the learned corrections into the fuel base table, then zeroes them. With one bank this is straightforward. With two banks there is only one base table to write into, so the Nexus moves across what both banks agree on and leaves the difference between the banks in the trim tables. Cells that matched get zeroed.
How we use it
- Get the base map close first. LTFT is for fine correction, not for building a map from nothing.
- Confirm short term O2 control is stable, and check the proportional and integral gain settings.
- Enable LTFT with the default limits, then drive a range of conditions once the engine is warm.
- Review the trim table. Consistent corrections are a sign the base table is off in that area. Scattered, random ones point at sensor or mechanical issues.
- When you're happy, apply to the base table so the next round starts from a better map.
Troubleshooting
| Symptom | Usually |
|---|---|
| Trims hitting the max limit | Base table too far off, a fuel pressure problem, or an exhaust leak fooling the sensor |
| Trims keep swinging back and forth | LTFT gain too high |
| Learning never happens | Engine not reaching the minimum coolant temperature, or a disable option is active |
| Cold starts richer or leaner than before | Cold running corrections learned in; enable the coolant or post start disables and reset |
FAQ
Is long term fuel trim the same as auto-tune?
It is often called that, but it is a slow learning correction bounded by your limits, not a way to tune an engine from scratch. See our short and long term fuel trim explainer for how the two work together.
Can I use a narrowband sensor?
Yes, but a wideband is preferred because it measures across the full mixture range rather than just rich or lean of stoich.
