VVT sticking, running rich
 

I've been having an issue of late where my Miata runs super rich after I lug it at low RPM. Normally I wouldn't be doing this, but I'm trying to tune the fuel map on my MS3. If the engine hangs out around 1800 RPM for any length of time, it starts to run extremely rich. Revving the crap out of the engine seems to fix it.

I figure this has to be airflow related, and the only thing that changes airflow outside of throttle and engine speed is the VVT system. The data I pulled today seems to confirm this. I went to the first spot in a datalog where the engine was running rich and found this.

https://cimg8.ibsrv.net/gimg/www.mia...3128decbf2.jpg

The datalog shows it staying exactly at 21.7 degrees for a good length of time. Once the VVT unsticks, it runs much better, but in the meantime it is wreaking havoc on my attempts to tune this and build enough confidence to go boosted. Usually a dash to redline will unstick it, and once it breaks free, there is an immediate surge in power along the with AFRs getting much more lean and closer to target.

Has anyone had any issues with the VVT system sticking? Are there any quick checks or cheap/easy fixes or checks I can do? I previously did have to take apart the VVT actuator because of a stripped screw head to replace the cam seal. It never had any running issues related to this until I went Megasquirt.

Aidan's build thread is super long, but IIRC he had a sticking VVT issue? Vague recollection, but I think it was sticking high.

Yeah... I've tried to read his thread in the past to see all the cool stuff but half of it is local chat. :(

Paging @aidandj, could you be so kind as give a sparknotes of what happened?

Current thoughts are to change the oil from 3k old Rotella T6 5W40 to a lighter weight oil just to see if labor can be avoided, its due for an oil change anyway. Maybe something just needs to be flushed out. I'm also going to take off the solenoid and oil line off the valve cover and look for any obstruction or dirt, but I am hoping to avoid disassembling the VVT actuator spinny bits again.

Which ms3

MSLabs V3.2a 01-05 Basic according to my emails with Reverant. Bought used from VTeckiller2000.

My issue was the cam circuit. I replaced one of the filter capacitors. But iirc reverant uses a different cam circuit than the ms3x.

If it was an ECU side issue, I'd have to have a simultaneous failure of the MS3 VVT driver and the cam position sensor, and that still wouldn't explain why its suddenly running rich. I figure if the cam position is reading constant in the midrange, it has to be physically stuck or I'm having both a driver and sensor failure, and something else reducing volumetric effenciecy. Maybe I'm mistaken, but if a PWM driven device is losing control, it should be forced to either full advance or full retard.

The more I think about it, the more I feel like this is going to come down to tearing it apart again...

Mine was caused loss of sync. Which meant that the ecu thought it was stuck, and cranked the duty cycle to Max or min in order to try and adjust it. In reality the cam was adjusting fine, but the ECU couldn't tell.

Vtechkiller was using a VVT head that was known to be working though. I don't think he was having this issue when he was using the setup.

I only know this because I got head from him.

random thoughts:
1) you shouldn't need much, if any, vvt advance that low. car shouldn't run uber rich with it closed. on many cars I don't even start advancing until past 2k and 50% load
2) if your duty is maxed but angle is still not moving you have an oil control issue or lack of pressure
3) do you have the range defined in ts?

VVT settings and spark advance are exactly as set by Reverant. I can't even get fuelling right yet, not going to dive into more variables.

I'm not ruling out an ECU side issue, but the fact that I also go stupid rich at the same time indicates something is either also reducing airflow or making it dump fuel at the same time as the cam timing goes wonky. Seriously, I go from 14.x:1 AFR when operating normally to 12:1, sometimes 10:1. I can post a datalog later today.

Anyone know if fuel injection control and VVT control share any circuits (on my ECU)?

I'm also running Blackbird Fabworx poly mounts, something could be getting shaken to death at low engine speeds.

both use the cam input don't they?
its easy to see if its commanding more fuel: just look at injector PW

Good point! I'll take a look at the datalogs and the cam sensor after work.

Ugh, just a little bit over the max file size to upload, and it seems you can't cut down a megasquirt log without forking over the cash for the pay version. Guess I need to do that... Anyway, zipped datalog attached, the fuckery starts right at 1136 seconds. Admittedly I haven't looked over this as much as I would like to, ran out of time today with work.

Both the electrical and bolted connections to cam sensor and solenoid were tight, so nothing should be flopping around. Guess I'll aim for getting the oil change and solenoid inspection done tomorrow.

Aidan, was your issue intermittent or constant? My subpar understanding of electronics makes me think a capacitor is either dead or alive.

It was intermittent. I changed to a higher capacitance value for more filtering. Noise was causing it to lose sync.

You're running VVT right? Did your indicated cam position stay exactly constant during this time?

Below is my current line of thinking, making a good few assumptions about things I don't 100% understand. I don't believe the stock 4 tooth crank trigger wheel is accurate to 1/10 of a degree at various engine speeds for 30 seconds at a time.

1. Electrical noise back fed from coils makes ECU wonky at ~1800 rpm.
2. ECU takes last cam position data it trusts as gospel, refuses to listen to new data. For the datalog/event in question, the gospel is 21.7*.
3. ECU sets new cam position target, does not see cam position changing to match target.
4. ECU commands 100% duty cycle attempting to make last accepted position (21.7*) match target.
5. Cam advances/retards all the way, mass airflow decreases, fuel injected remains constant.
6. Situation continues until something about high RPM convinces ECU to re-sync.

I figure its either backfed noise from the coil signal ground, or the cam position sensor signal getting interference from the high voltage from coils. Based on some quick research, I've got two options right now. In order of easiness...

1. Ground both LS coil grounds to block, as per CarPassionChannel video below. don't see any problem with this, as I don't think the Megasquirt is reading anything off of the signals it gets back and is just shunting them all to a common ground.
2. Place capacitors between signal feed and signal ground wires as per Megasquirt documentation here and in picture below.

I don't think shielding the cam position sensor wires all the way from engine to ECU is a reasonable option.

Thoughts?

https://cimg2.ibsrv.net/gimg/www.mia...24e7c966b9.png


Edit: well, attempt to make a text link to the specific time youtube video dun failed. 12:05 is the time in question.

Haven't had a chance to look at your logs but is it always sticking at the same angle? Mine would do the same thing angle would stick and ecu would try to Max the DC to get it to change.

Replacing the cap fixed it

Yes, exactly the same behavior as I have. Given that I didn't build this ECU I'm somewhat less confident in replacing parts on it.

Do you see any reason I shouldn't send both signal and main coil grounds to the same ground? It's the easiest option and I see no reason not to try it before adding four resistors to a wiring harness or doing ECU surgery.

You could try anything. But I'm the second account of the capacitor change fixing it. It's a capacitor on the MS3x. So it comes from diyautotune installed, not by the builder. It's a surface mount part.

I changed the signal grounds to ground on the ECU, and the behavior does persist although it seems better.

Would you happen to have a useful post/guide to the capacitor swap? I'm willing to give it a shot, or at least read more about it.