Megasquirt Support Forum (MSEXTRA)

Here's files, starting from 27 F, might not be representative as the car started after just a few tries.
"Cold" files are as I first cranked, others are at a cold idle

Rest of files. TunerStudio recorded 32 seperate tiny .msl files while I idled a bit. Started a new.msl with every synch loss, you could string them together and view as a whole. I assume this is because I am triggering the SD with RPM>300, and synch loss = no RPM,

Looks like I am still set on Wasted COP from earlier cam diagnosis.

Snugged up pin sockets on both cam and crank sensors, then tried one cold start each with a hot air pre-heat of cam then crank sensors. No joy.
Won't have new sensors to test for a bit, not sure what else could be temperature (or time?) sensitive to give me the synch loss on the 32-2-2-2 code.

Datalog is a 29 F start with preheated crank sensor, a couple of start attempts, then cold idle with synch losses until time= 34 seconds, then all is well.

Replaced cam sensor and did a couple cold starts, looks like it helped, now getting synch loss only with initial cranking (or when I stall it engaging clutch on cold tranny)
Hopefully this will now start with two tries rather than 6 or 8 or 10, and quit the stalling. Not sure why the sensor could fail cold, but warm to working temperature in 30 seconds, as it's sensing the back of the cam pulley in air. Perhaps a vibration component as well? Time will tell.

Crank sensor came in today, I'll try that next. It's in the crank oil and I would expect it to warm up faster than the cam sensor.

It wouldn't be the first time I've seen a sensor have a temperature dependent failure with VR sensors. I'm not sure if it is due to thermal effects on the magnet or some sort of expansion / contraction changing the air gap just enough to make trouble.

Hoping that the cam sensor took care of the issue- still a drop or two while cranking, but not the continuing missing and stalls I had before. Coincidental that it expressed with the swap to the SW0205?

Just cause there are a trillion settings combinations for cold start, don't assume it's a software settings issue until you check the hardware.

Now on to why my first crank never registers a good AFR on the preheated O2 sensor.....

snowrx wrote:Hoping that the cam sensor took care of the issue- still a drop or two while cranking, but not the continuing missing and stalls I had before. Coincidental that it expressed with the swap to the SW0205?

We don't have quite the same opportunity OEMs have for putting in testing our ECUs with worn, marginal, or partially failed sensors. So the stock Subaru ECU may have had some sort of hardware or software features to compensate for a sensor that was beginning to fail that our system didn't.

Just cause there are a trillion settings combinations for cold start, don't assume it's a software settings issue until you check the hardware.

Now on to why my first crank never registers a good AFR on the preheated O2 sensor.....

The tune has a 40 second delay after start for applying the AFR readings. It is also possible that the supply voltage to the sensor drops far enough to make the controller reset while cranking; many wideband controllers also have a built in delay before they start sending a signal.

Thanks again Matt, I'll look at those O2 thoughts

It's Baaack! Well, the cold temperatures are, so I'm getting the initial cold start weirdness again. Cyclic synch loss at initial cold (0-20 degrees F) start, with the engine at fast idle but the ECU is losing synch and showing RPM going to zero rpm briefly, then picking up rpm signal and heading for fast idle over and over. The colder it is the worse it is. Now that my cold fuel numbers are getting close, the engine has no problem coasting through the synch loss at fast idle, but you can hear the engine missing and RPM surging. As the SW0205 is seeing low rpm with the losses, the PW jumps to cold cranking numbers with each cycle and it's fighting all the fuel, and if the O2 comes on line it's showing lean I assume from rich misfires. After 15 to 30 seconds the losses go away and everything is fine for subsequent warm starts. We're talking around -10C to 0C for CLT and maybe -12 to -5C MAT for these starts. The issue goes away long before either temperature starts to rise in the datalogs. I'm no longer getting the odd discrete rpm levels though, as soon as the synch comes back I have good rpm numbers. I replaced Cam and Crank sensors with new ones, and tightened the connector contacts.
I notice that the next_spark.tooth values change, going from 2,17,32,47 to 3,17,32,47 and occasionally skipping the second value without synch loss, or swapping a 33 for the 32. If I get a loss, it when to code does not see the transition from 2 to 17. I don’t know what those numbers represent, just was looking at them to try to see if it was cam or crank wheel related. Wiring is the OEM harness from sensor to PnP ECU.

Perusing older MS/Subaru threads I see that some have added resistors to their VR sensors to adjust the signal, but that appears to be to avert a high speed synch loss, don’t know if that still applies to newer hardware? Then again users of some other ECU’s are shimming the sensor 1/16” farther away from the cam wheel to get an signal level that they can read. Can’t find any mention of very cold temp sensitivity anywhere. Not sure how many cold nights I'll have left this year to puzzle this out - any ideas out there?

The moving next spark tooth is just a result of the timing moving - the code continuously changes what tooth to start the hardware timer on depending on the amount of commanded spark advance.

Can you get a tooth log and composite log of what it's doing at this point?

I can try, but usually by the time it occurs to me to get the logger open, it's quit doing it! I'm going to try this AM w/wasted COP and Semi seq again to see if goes away.

Ok, it's a balmy 17*F, so not much time to capture logs before it quit, but here's what I got. The first tooth log is repeated start attempts without getting to idle, so not so great, by the time I tried the second tooth log I was out of the "cold" window and all was fine.
First log was wasted/semi then I changed back to COP/Fully seq as not incorporating the Cam did not seem to help.

My starter hates me....

I have to assume this is a physical manifestation of something other than the MS. I don't know if I should be looking at the crank or cam sensor at this point. The crank sensor is mounted directly in the crankcase very close to the small diameter trigger wheel, so I find it hard to believe there's a dimensional difference to the air gap with such small temperature differentials between losses and no losses. Not sure which would warm up faster, the cam pulley from belt friction inside the timing cover, or the relatively massive crankcase from the oil starting to move around. I'm not wild about hogging out the crankcase hole to add a Hall sensor, I would prefer to make the OEM sensor work.

The cam sensor bolts right to the cylinder head, so not much metal to expand between it and the cam pulley, unless I'm getting cam end float at cold temperatures only!
Maybe I should be looking at ECU temperature, how fast can it warm up?

I'm seeing it lose a tooth on the cam. There isn't any noise filtering currently active on the cam trigger - is there any way to get a tighter sensor gap?

Crank signal looks OK.

Thanks for isolating it to the one signal, that gives me a path. I'll mill down the flange on one of my (three) cam sensors and see what that does.

Should I be looking for some sort of filtering setting for the cam input?

The filters are all off, which is what I recommend when there's a possibility they could be filtering out real signals.

I shaved a cam sensor down to almost touching the pulley, and the synch loss was persistent enough not to start readily, although it would of had I persisted. I then added a ~.030" shim under a standard sensor to open up the gap to wider than OEM, and still got the same issue, even at CLT of 4.6C. 10-15 seconds of Synch losses, then just fine. Does not seem to be the air gap. So weird, maybe I should put the engine in a convertible so I don't need cold starts....

I am having a very similar issue did you ever figure this out?

No, it just became less of an issue as other settings got closer to optimal. Car just acts very cold blooded and I wait it out after a restart. Rust has caught up with the old girl after 20years and I think it's time to start on the new equivalent, a GR Corrolla.

Oh man that stinks. But a new project is always nice. Thanks for the reply and good luck. If i happen to figure anything out ill send the info your way.