Supercharged car: relay K18 earth fault and part-throttle misfire (case study)

Part 1: no start with P1512

After an oil and filter change (which involved unplugging the throttle body) the car turned over but wouldn't start, with P1512, no fuel pump prime and the charge cooler pump not running. The code came straight back after clearing it. Unplugging the coil pack made the charge cooler pump run.

• First suggestions: a failed relay (the yellow one in the boot is one owner's first guess), the fuel cut-off switch, the fuses next to the relays, an earth fault or the coil pack. Another owner said he had seen the same fault caused by a dud brown relay.
• The brown relay, K18. The owner found the relay worked when given 12 V from another source, but its socket supplied only about 6 V. That turned out to be a flat battery after weeks of being towed with the wipers on. With a charged battery the fault still came and went. He measured about 60 ohms from relay pin 86 to earth when it failed to switch and about 15 ohms when it worked.
• How K18 is wired, according to another owner: it is switched by the green/pink wire (the negative) and fed by a brown wire. The small brown wire links to the larger brown that feeds the relay outputs (injectors, coil pack and so on), and that comes from the fuse under the windscreen next to the wiper motor. The green/pink wire passes through the plugs behind the coolant tank, so check those. See Throttle failure and the connectors and Relays and fuses (NA).
• Earths. The owner fitted a new earth wire between the ECU and the cylinder head (a decent size, 2.5 mm² or more, said another owner) and a missing braided earth strap between the head and the cam cover. The car then worked for a while, but the fault was not fully cured by that.
• What it finally was (for the no-start). A previous owner had tapped into the harness that carries power to the rear, to feed the charge cooler pump, and broke the fir-tree clip, so the harness rested on the oil cooler. The plastic insulator inside the connector had melted, giving a poor contact. With the harness moved around, a relay could be felt clicking on and off. He cleaned the connector and the car kept working.
• Method that worked: having someone hold the relay in a hand while he jangled parts of the loom, to find the section that made it click on and off. He also noted that intermittent faults often return soon after they seem to be fixed.

Part 2: stalling on a blipped throttle, and misfires on light throttle

• Symptoms. After the new ECU the car cut out whenever the throttle was blipped from idle, and it hesitated and misfired between about 1,500 and 2,500 rpm on gentle throttle, which made pulling out of junctions hard. It had done something similar through two engines, two ECUs and two throttle bodies.
• Blipping cut-out. The tuner noticed the second (post-cat) O2 sensor was reading maximum voltage all the time. The owner unplugged it for fuel learning, since it is only used for checking the catalyst. The base map was later tweaked and the engine no longer cut out when the throttle was applied sharply.
• Tried without success: new O2 sensors (a genuine front sensor and an aftermarket rear one), plug gap reduced from 1.0 mm to 0.7 mm (one owner cured stutter on the Dutch ECU this way, but it made no difference here), repeated fuel learning, a spare throttle body test on loan (not completed because of the no-start).
• Suggestions: try a different coil pack (GM or Denso rather than a cheap eBay one), check the exhaust (two owners found their catalytic converter had come loose from its locating welds and rattled, though a camera inspection of this one looked intact), check the cam timing, and note that some aftermarket cams are said to have base circle problems that cause misfires and idle problems.
• Logs. Misfire detection is normally disabled on supercharged maps, so enabling it was suggested to see misfires in the logs. With it on, the misfires showed mainly on cylinders 2 and 4 and very little on 1 and 3. Another owner pointed out that the coil pack fires cylinders in pairs, so a bad coil would normally affect a pair, but a working coil pack should still be tried.
• MAP oddity. One owner noticed in the logs that the MAP reading dropped slightly when the throttle opened and revs rose, and that 47 kPa at idle seemed high (30 to 40 kPa expected), which could point to an air leak or just the cams.
• Outcome. The thread ends with no fix for the light-throttle misfires. Don't take any of the above as the answer.

Tuning-software notes from the thread

• After changing values, save them to the ECU's permanent memory, or they are lost when the ignition goes off (the owner made this mistake once).
• One owner advises getting the car about 95% right on the road before a rolling road session, and says a wideband is essential above about 250 hp and recommended in logs. The car's owner understood the tuner can't use a wideband signal for the fuel calculation, while the other owner says it can be used for the power runs, so the two views differ.
• Replacing the O2 sensors on a car that is about to be learned is suggested by one owner even without fault codes (see MAP, boost and lambda sensor faults).