Throttle closes at WOT after tuning: the MAF overflow trap
Diesel · Gasoline · Methodology
Throttle closes at WOT after tuning: the MAF overflow trap and the linearization fix
The tune pulls clean, the codes are clear, the sensors log in spec — and still the throttle backs off at high RPM. When the throttle closes at WOT after a tune, the cause is usually not a component. It is an airflow overflow check hardcoded into the ECU program. Two well-worn roads lead into it: raising airflow past the threshold, and raising the driver-wish request instead of linearising it. This article covers the mechanism and both fixes.
When the measured air — the MAF flow and the MAP multiplied by RPM — goes over a threshold hardcoded in the ECU program, the throttle closes: the torque curve looks normal, then continues below the factory value. On gasoline, that behaviour at 5,000–6,000 RPM is the overflow signature. An EGR delete trips it earlier, because all flow now passes through the MAF. On the pedal side, writing a request above what the car can produce makes the pedal die at ~85–90% of travel — the limits cut it. The fix in both cases is the same direction: don’t max-request the system, make the pedal characteristic more linear — which also drops fuel consumption.
This article is built from two questions in our 5 May 2026 Q&A. Dmytro asked how the throttle control algorithm works and why the throttle sometimes partially closes during engine operation (~10:50). Denis asked whether to edit driver-wish maps that request more power than the car can produce, because customers keep asking for the dull pedal to be removed (~50:46). The two answers land on the same mechanism — the timestamps below reference that recording.
The overflow signature: normal curve, then below factory
Start with what the symptom looks like, because it is distinctive. This is a typical problem caused by tuning:
“This is overflow — when the measurement of air is over some level, and this level is hardcoded in the program of the ECU. Then it will just close the throttle. We will have something like a normal torque shape, and then it will just go and continue — but below the factory value.”
Q&A 2026-05-05, ~11:18–11:46
So the torque trace climbs exactly as expected, crosses the hidden threshold, and settles into a plateau under the stock output. That two-part shape — normal rise, sub-factory continuation — is what separates the overflow from a sensor fault or a boost leak. Those look messy across the whole range, not clean up to a point.
“If you see such a behaviour, for example in a gasoline engine at 5,000–6,000 RPM — this is just overflow. It means that the value of the MAF flow and the value of the MAP multiplied by RPM is over some threshold.”
Q&A 2026-05-05, ~11:46–12:07
Two signals are in play: what the MAF measures, and the airflow computed from manifold pressure times engine speed. The threshold check sits over them in firmware logic — not in the calibration maps you scale in WinOLS®. That is why hunting for it in the map pack goes nowhere.
Two roads into the trap
The tune raises flow past the threshold
More boost, more fuelling, more air demanded at high RPM — the MAF flow and MAP × RPM values climb until they cross the hardcoded level. Classic appearance: a gasoline engine that pulls clean and then backs off at 5,000–6,000 RPM.
The delete reroutes all flow through the MAF
The MAF flow and the EGR flow together make up the measured total. Close the EGR, and everything that used to recirculate now passes through the MAF — which reaches its maximum flow easier, tripping the overflow earlier in the range.
“The second situation is when you disable EGR — then we have more flow through the MAF, so the MAF reaches the maximum flow easier. Remember that the MAF flow is connected with the EGR flow, and together they make the flow, which is measured by RPM multiplied by pressure. So if you close EGR, you push more air through the MAF — and then you cause the error of the overflow. This is why the throttle is closed.”
Q&A 2026-05-05, ~12:10–12:49
For EGR-off work the consequence is direct: the delete produces two separate downstream effects. The exhaust runs cooler and the DPF stops completing regeneration — that is the +20% post-injection story. And the MAF now carries the full flow, which walks the airflow signal toward the overflow threshold. A complete EGR-off calibration accounts for both.
What the throttle algorithm is actually doing
The throttle is not a dumb actuator between the pedal and the intake. Three details from the same answer explain most of the “weird throttle behaviour” tuners report:
“The control algorithm utilizes two maps which are one in reverse to the other — just to ensure the calculation and protection for the maps. This is the simple linearization of the flow versus the position of the throttle. And then there is some delay: if you press the pedal rapidly, the throttle is not opening immediately fully. And if you remove the leg from the pedal, it closes almost fully, and then very slowly the last part — to enable the idle run. All these parameters can be controlled from the ECU program.”
Q&A 2026-05-05, ~12:52–13:36
- Paired maps, one the reverse of the other. The calculation is cross-checked — the same protective pairing Bosch uses elsewhere, like the driver-wish monitoring shadow that produces no-start conditions when only one copy gets edited.
- Linearisation of flow vs position. Throttle angle is mapped to flow through a calibrated curve, not 1:1.
- Intentional delay. Fast pedal press ≠ instant full throttle; pedal release closes the plate almost fully, then creeps the last part shut to keep the idle stable.
All of it is calibratable — which cuts both ways. A tune that fights this structure instead of working inside it ends up in a protection path.
The wrong fix: raising the driver wish
Which brings in Denis’s question — the pedal side of the same trap. Customers ask for the “dull pedal” to be removed, and the tempting move is to raise the driver-wish request. Thomas ran the numbers on a live file:
“Let it be the torque of this car is 275… if you just write something like 300, then of course you will cause the trouble that the pedal will stop working at some level — maybe 85, 90 percent, or 87 — and it will stop working, because then the limits will cut it. It is just a strange behaviour: the pedal works some percent, and it is not working over some position.”
Q&A 2026-05-05, ~51:09–52:01
The mechanism is the ceiling again: request more than the system will deliver, and the request saturates before the pedal does. Everything past ~87% of travel maps to the same clipped output — the customer feels a dead zone at the top of the pedal, which is the opposite of what they paid for.
The right fix: make the pedal curve linear
The answer from the same demonstration — when the values are already big enough, do not increase them:
“The best idea in that case is not to increase the pedal, if the values are big enough — or even make it more linear, which is somehow very good for the fuel consumption… because the change of the position of the throttle by vibration — when your leg moves a bit, even when you don’t want to move the leg, but the car vibrates — the change will be smaller, because of making the characteristic more linear.”
Q&A 2026-05-05, ~52:01–52:47
In the live demo, the linearisation went downward — Thomas reduced the over-aggressive part of the pedal curve rather than lifting anything (“as you see, I reduced the pedal”, ~52:17). Factory driver-wish calibrations often front-load the response for a lively showroom feel; flattening that skew is what makes the pedal honest. What the customer gets:
“It will be a very nice feeling at the road: if you press a bit, it goes a bit. If you press half, it goes half. If you press three-fourths, it goes three-fourths.”
Q&A 2026-05-05, ~52:47–53:01
And a measurable side effect: fuel consumption drops. On a non-linear curve, the small involuntary pedal movements from cab vibration translate into disproportionate throttle movement. Linearise the characteristic, and the same physical wobble produces a smaller flow change. The customer asked for feel and gets economy as a bonus.
| Approach | What the driver feels | What happens at the limit |
|---|---|---|
| Raise the request (275 → 300) | Sharper low pedal, dead zone from ~87% up | Limits cut the request — overflow-style saturation |
| Linearise the curve | Press a bit → a bit; half → half; full → full | No saturation; smaller vibration-induced flow changes, lower consumption |
The full calibration method behind this
The overflow check, the paired throttle maps, and the driver-wish structure are chapters of one system — the torque model. Diesel Practice covers it on EDC17 and MD1; Gasoline Practice covers the MED17 side.
See Diesel Practice →Frequently asked questions
Why does my throttle close at WOT after tuning?
Airflow overflow. When the measured air — the MAF flow and the MAP multiplied by RPM — goes over a threshold hardcoded in the ECU program, the ECU closes the throttle. The torque curve looks normal, then continues below the factory value. On gasoline, that behaviour at 5,000–6,000 RPM is the classic signature.
Can I edit the overflow threshold in WinOLS?
The level is hardcoded in the program of the ECU — it is not a calibration map you scale. The practical fix is to keep the airflow signals under the threshold: account for the MAF carrying full flow after an EGR delete, and linearise the driver-wish curve instead of raising its maximum.
Why does the throttle act up after an EGR delete?
The MAF flow and the EGR flow together make up the measured total, checked against RPM multiplied by pressure. Close the EGR and all of it passes through the MAF — the MAF reaches maximum flow easier, the overflow error triggers earlier, and the throttle closes.
Should I raise the driver-wish maps to fix a dull pedal?
No. If the car produces around 275 Nm and you write 300 into the request, the pedal stops working at roughly 85–90% of travel — the limits cut it, and the customer gets a dead zone instead of response. When the values are already big enough, make the characteristic more linear instead.
What does linearising the pedal change for the driver?
Press a bit — it goes a bit. Press half — it goes half. Press three-fourths — it goes three-fourths. Fuel consumption also drops, because the throttle movement caused by cab vibration gets smaller on a linear characteristic.
Related reading
- EGR off and DPF won’t burn: the +20% post-injection rule — The other downstream effect of the same EGR delete, from the same Q&A.
- Bosch EDC17 tuning guide: maps through physics — The torque-model structure the throttle algorithm plugs into.
- Find idle maps in EDC17 and MD1 — The slow-close-to-idle end of the same throttle behaviour, from the same session.
- KB-08 — EDC17C46 no-start after a driver-wish edit — The paired-map protection tripping in its hardest form.
- KB-13 — Boost pressure limit on a VW MED17 — The boost-side ceiling in the same protection family.
- Related discussion — Why does my pedal stop working past 87%? — The forum thread for the pedal side of this article: post the ECU, the factory curve, and what you changed.
- Tuners Guild pricing — All diesel and gasoline courses, bundles, and upgrade paths.
