ive been using the forum for a while for help on stuff so its time i gave something back.
i had an issue when the engine was hot, long crank time and sometimes seem to "catch" during the crank. i thought this pointed to a fuel delivery delivery issue, checked out the lift pump , pressure and delivery OK so not that.
ive read about the issue of a minimum crank speed needed when hot, tried he check of unplugging the CTS, this gave a long glow plug time, as it should but did not affect the starting.
i then noticed that the rev counter was bouncing around, showing up to 3000 rpm during the problem starts. i also got a CEL at last
code for G28 implausible and cpu checksum error. could be a failing G28 then, nope, too easy, changed it with no affect on the problem
i decided to take it the garage i use when things get beyond me, they are very helpful and willing to take time to discuss faults with them. they scoped the G28 signal line and it was just mush when the starter was running, new starter and all was good.
fast forward a couple of years and it started to do the same thing, extended crank and rev counter going loopy.
i don't believe in co incidences when fault finding, the first fault mode, starter motor interfering the the G28 was rare enough that the chances of the second starter doing the same are non existent.
this time i tried something different, i could get the car to start by doing a rolling bump start, so whatever it was had noting to do with the fueling system, but was linked to the starter motor running
had a chat with the garage, they were as intrigued as me. first thought was a break in the shielding on the G28 cable, so stripped it out of the loom, routed it away from the starter, wrapped it in silver foil to block any electrical interference, no change
then load tested all the wiring fro the crank and cam sensor, basically putting a few amps down the cable using a stop/tail lamp as a dummy load and measure the voltage drop, all good
then measured voltage drop from the battery to the ECU during cranking, to see if CPU was seeing a low voltage, all good.
checked and cleaned all the earth points, no change
i was running out of things to check. then i had a think about the sequence of when the fault first appeared. abut 4 years ago i put in a new clutch/DMF. i also bought a second hand gear box with the 1st/2nd selector fork issue( for peanuts), as the original was starting to get a bit whiny. fixed the selector fork. i'm sure the issue started about 6 months after this, but not regular enough to link the two.
thinking about how a starter works, it has a 12v feed and the return path is through the body of the starter, through the gearbox casing to the main earth return cable at the top of the gearbox. grasping at straws, it took the starter off and had a look at the mating surface where the starter mates onto the gearbox. it was fairly white and cruddy, as it has been exposed to the atmosphere for several years whilst waiting to go on the car.
i cleaned it all up with sand paper so it was nice and shiny, re assembled SUCCESS!
it now starts first compression stroke, hot or cold.
the made no sense to me till i looked at where the engine speed senor is,and where the starter bolt are in the block.
so this is my theory. normal electrical return path is through the gearbox casing, keeps the current flow away from the G28 and aluminium is not a good material for generating magnetic/electrical fields.
the oxidized aluminium of the gearbox where the starter motor mounts presents a increased electrical resistance, the return current preferentially flows down the bolts holding the starter in, then through the block, which is right by the G28, the block is iron, which is a good ferro magnetic material. this then induces a voltage in the G28 which we see as the bouncing rev counter. ECU cant correlate cam and crank so no engine start.
ive seen quite a few postings on various forums describe the same fault, but none with any resolution. there is one on this forum recently
i'm posting this as it may help others when searching for the fault, and its a free fix!