Has your tuner broken your torque model?

Applies to: IOcan · Standalone controller. What this means

Your transmission decides how hard to clamp its clutches from one number: engine torque. If that number is wrong, the gearbox is not at fault and no amount of transmission tuning will fix it.

This is the check to run first, before you change anything else. It takes about ten minutes and needs nothing but a datalogger.

Why this happens so often

Engine tuners are not usually thinking about your gearbox. Three routine tuning actions break the torque signal:

  • The torque model is deleted. Some tunes strip torque-structure code entirely to remove factory torque limiting. The gearbox then gets nothing, or gets a constant.
  • The torque model is left decalibrated. The engine now makes far more torque than the untouched model claims. The number is still plausible — it just does not match the engine.
  • Torque is confused with a torque limiter. A tuner sets the reported value to a fixed ceiling to "remove the limit". The gearbox reads a flat line.

All three produce a signal that looks alive in a log. That is what makes this worth checking deliberately rather than assuming it is fine.

What a wrong number actually does

The torque value sets clutch pressure. There are two failure directions, and they are not symmetrical:

In the log In the car
Reads high Above the real curve Clutch pressure higher than the load needs — abrupt engagement, harsh shifts, aggressive creep and takeup
Reads low Below the real curve Clutch capacity below the torque actually present — the clutch slips, heats and wears

Reading low is the damaging direction. A slipping clutch destroys itself. If you must run an uncalibrated estimate temporarily, bias it high.

On a gateway there is a second consequence people miss. During a shift the factory TCU asks for a torque reduction as an absolute limit, and the gateway converts that into a proportional cut demand using your torque value as the reference. So a wrong torque model does not only mis-set clutch pressure — it changes how deep the shift cut is. That is why a bad torque signal so often shows up as bad shift quality rather than as an obvious fault.

Nothing warns you. No fault is raised for an implausible torque value, in any calculation method. The estimation block reports no faults of its own.

The check

Log the Engine torque output and drive the four points below. You are not looking for precision — you are looking for whether the number behaves like torque at all.

1. Any running condition — is it alive?

What you see What it means
A constant −123 The configuration still selects the removed Fuel Calculation (v1). Choose a real method — see Renamed in 2.1
A flat line at any value A torque limiter is being reported instead of torque, or the source is dead
A value that never goes negative See step 4
A changing value Good — continue

2. Warm idle, in neutral

Expect a small, positive, steady value.

A negative or jumping value at idle means the drag maps or the torque source are wrong.

3. Steady mid-RPM cruise, then near peak torque at full load

Compare against the best reference you have, in this order:

  1. The ECU's own internal torque channel
  2. A dyno curve for this engine
  3. The manufacturer's published torque curve
What you see Most likely cause
Out by a constant ratio everywhere A scaling input — injector size, displacement, cylinder count
A ratio that changes with load A VE table or drag map problem
Correct at cruise, far low at full load On a boosted engine, a MAP input that saturates before peak boost, or VE tables that stop short of the boosted region

4. Overrun, closed throttle, in gear

Expect a negative value. The drag maps are subtracted from every calculated estimate, so on overrun the result should go below zero.

A value stuck at or above zero on overrun means the drag maps are empty. This is the fastest single tell that the model was never calibrated.

What to do about it

If the ECU broadcasts torque and it is wrong, the fix belongs with whoever tuned the engine. Ask them specifically whether the torque structure was modified or removed — not whether "the tune is good".

If you cannot get a trustworthy signal from the ECU, stop using it. Switch the Torque calculation method to one the device computes itself:

  • Fuel + Air (arbitrated) if the ECU broadcasts injector pulse width. It cross-checks fuel against an air model, so an over-reading injector signal is caught.
  • Air Model if there is no injector signal.

Then calibrate and re-verify with this same check.

⚠️ After changing the torque method, reset adaptations. On the standalone drivers, clutch pressure adaptation runs continuously, so corrections learned against a wrong torque signal stay learned. Re-run adaptation after changing the method or its calibration — otherwise you have fixed the input and kept the error.