[JHolland]

The automotive standard (ISO16750) calls for operation down to 6V during cranking.

That's an interesting requirement. It realistically means you either must be using a switching regulator or need to run a 3v3 system (Though given most automotive sensors are 5v, that would still be challenging). I can't see any way of meeting that requirement using an LDO and reverse polarity protection unless you throw in something like a buck/boost, which brings it's own resiliency problems

An ideal diode can be used for reverse voltage protection, the spec was written years ago and is basically 4.75V minimum processor voltage + a diode drop + the LDO losses, the 5V doesn't need to be stable it can be allowed to drop as long as it doesn't go below the minimum voltage required for safe operation of the processor, the sensors are ratiometric. I doubt if there are any ECUs that use LDOs for the front end regulator these days and 5V is still the norm for engine management syestems to maintain ruggedness and noise immunity. Buck/boost systems are certainly used in some ECUs, the standard is a minimum requirement and manufacturers often specify more stringent conditions - VW80808 specifies 2.5V for the cranking waveform for diesel engines.

[zoidberg]

What's the sensor that you're using? If it's (for example) a 9v-12v powered hall sensor then that will most certainly be a problem and can cause signal readings like what you're seeing

uhm, i have to look that up, i think i connected it to 5v, if i remember correctly. I switched around various sensors, and glued them into the sensor holder, but i think i used TLE4905L which should work from 3.8 up to 24v.

[zoidberg]

something like a buck/boost, which brings it's own resiliency problems

I thought about that too, but that would have noise problems, wouldn't it?
If a boost converter is used to create a stable voltage, couldn't that be used for the injector power supply as well? To make the injector characteristics settings obsolete?

[Eric H]

When clean and efficient power rails are required usually you have a buck/boost followed by an LDO.
On my digital dash project I use a buck/boost module followed by an actual pi filter since the analog current draw is quite low. It gives great results in my application.

[theonewithin]

something like a buck/boost, which brings it's own resiliency problems

I thought about that too, but that would have noise problems, wouldn't it?
If a boost converter is used to create a stable voltage, couldn't that be used for the injector power supply as well? To make the injector characteristics settings obsolete?

Injector outputs are pulled to ground like all other outputs

Would need a complete rework of the hardware to support

[JHolland]

something like a buck/boost, which brings it's own resiliency problems

I thought about that too, but that would have noise problems, wouldn't it?
If a boost converter is used to create a stable voltage, couldn't that be used for the injector power supply as well? To make the injector characteristics settings obsolete?

Cost and board space is the main issue. Pre and post filters are required to remove noise issues but that isn't a big deal these days with higher frequency controllers. Pre-filters are required to remove conducted emissions, there seem to be a large number of people who believe that a post filter or an LDO fixes all noise problems but they clearly haven't ever designed any electronics for sale. I think all modern powertrain micros now use PMICs with front-end switching supplies.