home / articles / spark-cut limiter without coil codes
Build series · Engine intelligence
A rev limiter that cuts spark without setting coil fault codes
I replaced the car's crude standing rev limiter with a smooth one. It worked on the first try. Then the check-engine lamp came on and the engine dropped a cylinder.
The limiter I wanted to replace
At a standstill, the limiter on this car was all or nothing. At 3,500 rpm it cut the fuel completely. The engine died, the revs fell, the fuel came back, the engine relit and overshot, and the cycle repeated. It is hard on the engine and useless for a controlled launch.
The new one
From 6 November 2026 the car has a progressive limiter. Below 20 km/h, as the revs approach the limit it starts dropping individual sparks, a few at first and more as it gets closer, while the fuel stays on. The engine settles against the limit instead of bouncing off it.
The original fuel cut is still there behind it as a backstop, in case the revs overshoot or the cut goes on too long. The normal 6,500 rpm limiter when the car is moving is unchanged.
Then the lamp came on
After one run on the new limiter the car set P0351, a fault on ignition coil number one. The check-engine lamp lit and the engine ran on three cylinders.
I cleared the code with the engine off and everything was normal. A second run set nothing.
A coil that fails once and then works perfectly is not a failing coil.
What had actually happened
Honda's ECU watches every ignition coil. Each time a coil fires, the ECU expects a confirmation. If one coil goes 25 engine cycles in a row without confirming, the ECU decides that coil has failed, stores a code, and switches off fuel to that cylinder to protect the catalytic converter. It stays off until the codes are cleared.
That is a sensible rule for a factory engine, where a coil that misses 25 times in a row really is broken.
My limiter had held the engine at about 3,400 to 3,550 rpm for roughly 0.8 seconds. Twenty-five cycles at that speed takes 0.88 seconds. By chance, the pattern of dropped sparks had landed on the same coil nearly every time, and the ECU had counted.
The coil was fine. My limiter had imitated a dead coil well enough to convince the ECU, and which coil got blamed was pure luck.
The guard
The fix was a rule inside the limiter: no coil may be skipped more than a set number of times in a row, well short of the 25 that trips the monitor. If a coil is about to reach that count, it fires, whatever the limiter wanted.
With the guard in place the longest run of skips on any coil is 11, and the limiter still removes about 92 % of sparks when it needs to. It lost almost nothing.
The guarded version has been on the car since 7 November. No coil code has appeared since.
The lesson
The factory ECU is full of monitors like this one, each with its own idea of what a healthy engine looks like. When you add a new behaviour, you have to know which monitors it will look like a fault to.
The usual shortcut is to switch the monitor off. I would rather keep it, because the day a coil really does fail, I want the car to notice. The same guard now protects the spark cut used during gearshifts.
Two small related changes
- Earlier radiator fan. The fan now starts at about 88 °C, against 92.6 °C from the factory, and the high speed comes in earlier too. On a long climb the coolant still reached 103 °C, so cooling is not finished.
- Misfire counting switched back on. The tune the car arrived with had it disabled. Turning it back on let me see misfires on one cylinder at 2,500 rpm, which led to that cylinder's own fuel trim.
By Haris, Hex and Boost.
Have a car the usual tools cannot touch?
Tell me the car, the engine, the gearbox and what you want it to do. I reply to every serious enquiry.