MAT correction issue - leaning out on hot starts
 

Hi guys,

I've been having trouble with really lean hot starts and noticed that my MAT correction table is just 0% for all the values. The car runs great once its moving a bit after a being restarted.

Is this table still used in the most recent updates? I'm using DIYPNP for a 1990 1.6.

Should I use brain's table as a baseline?

Attachment 184888

Unless MAT is either falsely or actually high, then MAT Corr. will not help. I don't think it's the right solution anyway. MAT Corr. should be used for MAT Corr.

High or low impedance injectors?

Yes. Additionally, move the sensor to the cold side of the IC if possible.

Pretty sure nobody uses low impedance injectors with MS. Nobody smart anyway.

This. Makes a huge difference and it cools off much quicker once you start moving.

derp post.

I will explain. When I was N/A, I had heat soak that gave lean hot, due to false reading of AIT sensor to values much higher than ever actual. In that case, adding MAT correction was helpful and valid.

Now that I am boosted, and the AIT is in the cold side of the FMIC, there is no heat soak, and the Intake Air temps are correct, low, close to ambient even on hot restarts. Under that condition, I still get lean hot restarts. If I added MAT correction, it would never take effect, as the MAT does not go high, even though AFR goes lean.

In other words, Lean hot restart, AIT is still 70*, same as original start, same as after things re-stablilize. Hence, adding fuel via MAT correction at, say 100* will have no effect, as that part of the curve is not used.

Same as you said before, with AIT sensor located in intercooler, there is no heat soak. That has been my experience.

So, I stand by the statement, that if AIT is not going high, either falsely, or accurately, then adding fuel at the higher end of MAT correction will have no benefit. If one does have heat soak, such that AIT is reporting high temps, then MAT correction will help, as there is a variable that relates to the problem.

DNM

i will explain:

the ideal gas law code of the MS is aggressive at best.

If you're reporting intake temps higher than 80°F, real or not, the MS is going to pull a lot of fuel and the only way to counter the effect is to tune out the ideal gas law algorithm by using the MAT corrections table to zero them out and then fine tune them yourself.

This was a huge issue when MS3 first came out as they change the implementation from MS2. It was such a sore point for so many installs, that the developers went back and reworked how the MAT corrections and ideal gas law code works on the newer MS3 firmwares.

I've also gone as far as implementing 2 ait sensors in my ms3; measuring temperatures in various locations in the engine bay, and even with the AIT sensor directly after the in IC, there's still PLENTY of heat there, enough to cause intake temps to soar without airflow--not falsely.

In your case, you can tune your startup settings (ASE and Warmup); which are based on CLT or AIT temp. But i can't imagine anyone having a lean start issue at 70°F; that just doesn't really happen.

Also no one should ever be running batch injection and wasted spark; that elminates almost all lean issues where you can idle smoothly at 16:1.

Scott. I appreciate the edit.

I am in the process of tuning EAE, which is quite the challenge. It tends to mess with my start, so perhaps my restart. If that is the case, as I get EAE dialed in, my issues will likely be fixed.

Re: Lean re-starts when AIT is 70*. I agree it does not make sense to me either. Hence my looking at other causes than MAT correction.

My MAT is flat from 60* to 90*, no fuel pull at 80*.

But, I don't want to hijack this thread, so I will bow out.

I'm using 550cc rx7 injectors.

I'm not sure how big of a difference it will make moving the sensor from it's current location. The IC pipe it's on doesn't get much cooler than the end tank, they are both downstream and within 6-8in or so of each other.

During a hot start - the car will go all the way to 19.0:1 and stall if i don't give it a little gas. Once it's moving/cooled slightly it will richen up and run fine.

you need to tune the MAT corrections to compensate.

Was this approach successful? I recently had problems with the engine going very lean on hot restarts. This is with one of the pre 1.3.1 builds (upgrading to latest 1.3.1 update in near future). We were using the Miata for drifting, so very high throttle for maybe 5 minutes with relatively low to road/air speeds to cool things down between runs. IAT temps were 150-160 on restart. I was running the "50%" MAT table to begin with, but then cut that in half again.

I noticed very little change in behavior after changing the MAT table. The car would run lean for a few minutes or until the driver drove around the pit area, at which point it would settle back down to normal range AFR's.

What we eventually noticed was a lot of pressure build up in the gas tank. If we cracked the gas cap to release the pressure, the lean start condition would clear up. This is with a NB2 VVT engine swapped into a 97 NA with no charcoal canister (vent line is connected directly to the OEM line which runs into the front subframe, previously used by the charcoal canister). I haven't confirmed, but believe that line isn't actually allowing pressure to release.

My plan is to go ahead and install a new fuel HP/flow fuel pump, new pressure regulator/gauge, etc and check that the vent line is able to actually vent.

This may not have any connection to the OP's problem, but thought it was worth sharing in case it "sparked" some ideas for the OP.

I have the same issue with hot-starts, or even just idling with a hot engine on my '99. Car goes very, very lean in idle after an Autocross run.
In my case it's NOT the MAT correction- the intake air temp reads close to ambient and I dialed the impact of MAT correction waaaay down anyway.

My latest theory is that the injectors heat-soak and their dead time increases a little as a result.
Since the '99 has a no-return system, there's no cool gas from the tank cooling the injectors. Not a big deal on the stock injectors, where 0.1ms or 0.2ms difference in dead-time don't have a huge impact on fuel delivered, but really important on my rather large (850cc) injectors.
Will eventually convert to a return system to confirm.

I have same issue.... Air Temp sensor is located on cold side of intercooler and reads ambient on hot re-start.

O2 readings go from the regular 13AFR at idle to somewhere near 16-17AFR...car stalls at idle for the first few minutes.

Injectors are DW1000 and dead time is very high (1.62ms @13.2v) to begin with.

I'm thinking injector heat soak on returnless system as well but i'm currently just running a richer idle to compensate.

how you logged a good idle vs. the hot idle to see if there's any difference?

any enrichments? same VE%? etc. etc.

My Megalog viewer won't open any logs from Tunerstudio...haven't logged. Something about X axis not defined...haven't looked into it yet... Haven't noticed any compensation setting differences in TunerStudio though...

In my case there's no fuel being pulled, rpm-range is the same, MAP is the same, pulse-width is the same. It's not the MS pulling fuel.
At identical settings less fuel is being delivered at high temperatures under the hood, or for some reason the car needs more fuel (doubtful).

Opening the hood actually helps- presumably because hot air extracts a little better with the hood open and the injectors can cool down.

Maybe I'll throw some ice cubes on my injectors and see if that helps:party:

Just checked and here's the result:
At 14.5V, which is where my car idles, my programmed injector dead-time is 1.21ms.
Pulse-width is 2.3ms to 2.35ms in normal idle.
Doesn't seem out of the ordinary to me. I usually idle around 12AFR, but going to 17 and even stalling when it get's nice and toasty under the hood.

does the purge solenoid allow the gas tank to vent on a returnless?

Tank is currently vented to air (purge deleted and hose points towards ground to vent tank)

Stock has a relief in the back off the canister that relieves pressure, but I'm not sure at what pressure. My line at engine is plugged, so I am relying on that same relief to control tank pressure. Purge valve has gone missing.

Logs
 

Looks like this thread has turned from MAT correction lean to Returnless Lean. So, I'm back in.

I am also, and still, under the impression that I have restart lean due to high fuel or injector temperature. I have attached logs of:
1) Idle after having run
2) A restart in about 1 minute of shutting down
3) A restart after about 30 minutes

Synopsis:

Restart After
Variable Idle 1 min 30 min
MAP 30.9 28.7 26.7
RPM 895 898 900
EGO 90 94.1 110
AFR 15.0 14.9 16.6
PULSE 1.756 1.742 1.886
CLT 198 197 178
MAT 87.2 89.5 87.7

As you can see, a 20% swing in EGO is not enough to control the AFR
Yellow 425cc injectors with dead time set at 0.8mS.

Note: I have large ASE and ASE time at above 130 CLT to allow a rich enough mixture until the EGO takes over on a restart. Kind of works, but not fixing root cause.

End of synopsis. Start of thoughts, assuming that evaluation of the information I have uploaded does not lead someone a cause I am over-looking.

If indeed the issue is injector heat soak causing long dead times, then I see (3) benefits to going to a return system:
1) Removal of damper and loops means less volume of hot fuel to burn before issue goes away at each restart.
2) Manifold reference of regulator means that the idle pulse widths are longer and the dead time variance makes less difference.
3) Fuel pump priming and running pushes some of the hot fuel through the rail and back to the tank.

DNM

Braineack, JasonC, anyone see anything in my logs or MSQ that shows something I am missing that can be fixed with tune?

I have posted this before.

After a hot restart the injectors are much warmer due to heat soak and no fuel flowing in them. This increases their dead time because their resistance increases and the current rises more slowly at the start of the pulse. Much more significant with big high impedance injectors.

If dead time is 1.0 ms, and at idle the car needs 0.2 ms of fuel, then the electrical pulse is 1.2 ms. If the dead time increases to 1.05 ms, then 0.15 ms of fuel is delivered. That's 25% less fuel.

I suspect that the AIT correction complaints claiming that the ideal gas law is wrong, are probably based on observations at light load where the injector pulses are short, not at full load.

With ms3 there is a dead time curve you can tune, but it's 2D based on voltage. How would you tune out that behaviour Jason?

your EGO at idle suggests your fuel map is poorly tuned. Youre leaning it out 10% at idle, but then adding 10% when it's lean.

add 10% to your VE table at idle. then at 110% EGO, that's 20% more fuel.

so try changing your idle cells from 56% VE to 61% and see what effect that has. then at 110% EGO, you'll be at 67%VE instead of 61...


oh wait... no. shit.

possibly, they change the code because of us from the original implementation, but never did add back in the RPM filter like we wanted to decay the correction back to ideal at higher rpms...

Great info in this thread!

Those of you with returnless fuel systems, I would not expect changing to a return type fuel setup to magically clear up all of your lean on hot restart issues.

I have a NB2 VVT engine, but am running the NA "return type" system with NA pressure regulator and return line to tank. During an "open" track day (drifting) with the return type fuel system, my Miata was going lean on hot restarts until driven around the pit area. Idling in place eventually corrected the problem, but it took quite a while. I am taking 5+ minutes to clear up during the heat of the day. I also had a lot of pressure build up in the gas tank, so I believe that the fuel temp in the tank had risen to the point that it wasn't helping cool the injectors much at all.

When I was having this lean on restart problem, I compared map and injector pw at different AFR's with the engine idling (~950rpm). The map and pw were very close, yet AFR was extremely lean (17+). This supports the theory/claim that there is a mechanical/fuel problem related to heat soak. The increase in injector deadtimes makes sense. However, I also wonder if the very hot gas had become aerated, further reducing the density beyond the reduction from the temp increase. It could even be a combination of the two...

This has to be a larger problem than just Miatas with Megasquirt ECUs.... Do AEM, Motec, etc ECU's have an injector deadtime compensation curve based on CLT and maybe other factors like run time? How about the Miata OEM ECU's?

Yes, but thanks for responding here. As I recall, you were working with Reverant on a possible solution to the issue. Any progress?

I think I am doing about the best I can with my situation, and it appears that scot was going to recommend exactly that solution.

My questions:
1) Is there a better, preferably tuning, solution?
2) Any other comments on potential improvement by going to a return, referenced regulator system? NoName suggests not. Are there any other data points from guys who have run the same car both ways?

I have a circuit idea to drive hi-impedance injectors similarly to lo-impedance injectors to reduce the problem. Seems a bit overkill. No progress.

Other possible solutions include:
1) A deadtime correction curve vs. CLT or intake manifold temperature (lol). Some OEs I think, do have intake manifold temperature sensors.

2) Have a feedback loop recognize that the dead-time is wrong by seeing the behavior of the AFR and correct it with a dead-time offset variable. Or simply bump the dead time when the short term fuel trim hits -15%.

3) Have a model of heatsoak and do the same as (2)

4) have a circuit measure the current waveform and determine the temperature of the injectors and adjust the dead time accordingly.

All of the above seem complicated but...

I have an uninstalled Adaptronic e1280s with its infinite programmability. I can implement either #1 or #2.

My setup is currently naturally aspirated and running on E-85.

I was totally lost on this until I asked Reverent what the hell was going on. I would tune on a cool evening (65F ambient) and get the tune damn near perfect A/F ratios (15.5 at cruise at anywhere from 2000 to 5000 RPM, 13:1 at WOT) and the next day in the heat (85F ambient) I would be running 16.5:1 at cruise and 14:1 at WOT! So I would think I messed up my tune, redo it until it was once again correct... and then at night in the cool evening air I was running 14:1 at cruise and 11:1 WOT! It was driving me nuts until I asked Rev what was going on. Now I am in the process or correcting my IAT table.

I was planning on having my AIT post intercooler after installing my turbo system (Step promises next week :jerkit:) but I was wondering if we would be better off with it in the intake track near the air filter? Any reading over 150F isn't going to have any correct "at cruise" tuning to have a AIT correction applied to it... we will always be WOT to get those temperatures post intercooler. With the AIT in the pre-turbo intake track your tune would look very different than someone with the AIT post intercooler... but which one would be better for drivability, hot starts, changes in weather etc.?

Just some more salt to add to the wound of dealing with the MS3 air temperature compensation :)

Keith

no. think about it...

Would a coolant temperature sensor surface mounted to the metal of the fuel rail be a good measure of heat soak for the fuel injectors? If so, would the MS3 have any way of using this information to affect injector dead times or would we need a new version of the code?

Keith

I get dangerous when I start thinking :)

What I am getting at is that any AIT correction we have above 120F or so is just numbers pulled out of our ass, and then we do the actual tuning based on what our A/F ratio is in the real world. So, I can have the AIT sensor down stream of the intercooler and have compensation based on numbers in the AIT correction table that I pulled out of my ass.... or have an AIT correction table that is correct at the intake pre-turbo and have my tune based on that. I know that under boost the curve is very non-linear, where raising pre-turbo intake temp by 10F will raise manifold temperature by 15F or more... but as long as it is consistent you will still end up with a good tune. My question was will this make any difference in daily driving / weather compensation.

In summery, already did the dangerous part (thinking) and I don't see how it would make much difference in daily driving / weather change compensation but I was wondering if there was something I missed :)

Keith

PS: I am a data junkie, I have the canbus 20 channel data box from Reverent that I will be using for temperature at the air filter, pre-intercooler, and post intercooler. If I go to an air to water system (drag racing with ice tank, NOT road courses) I will add ice tank temperature, pre and post IC core, and pre / post intercooler radiator core temperatures to my logging. I may add a coolant temp sensor surface mounted to my fuel rail just for giggles.

well this is why i wanted a MAP or RPM filter to remove AIT corrections, but no one listened to me.

I like to zero out the AIT corrections, then fine tune them based on real life data to keep AFR where they should be based on real conditions; never had an issue after that.

But the corrections will be wrong in a newly warmed up motor vs. a heatsoaked newly started one.

I can see the approximate deadtime when I put a current probe on the injectors. Indeed it lengthens when heatsoaked. And increasing the deadtime made it idle and act properly. Changing the VE or AIT corrections was a kludge.

There is actually a 2nd, smaller effect that screws up fuel calcs at hot idle, and this is what what Braineack's suggested fixes attempt to address (and which Reverant AFAIK already does).

The AIT sensor does not track the actual air temperature at the valves at low airflows. This is because the air will heat up somewhat as it flows past the hot intake manifold runners.

One other ECU solves this by "blending" the CLT temperature with the IAT temperature as a function of RPM (which really should be a function of airflow which is RPM * VE). i.e. at say 5% power (or 1000 RPM if RPM is the X axis), realIAT = 80% CLT + 20% IAT.

Ms has this function iirc, but its not documented.

Quelle surprise

It's the %CLT used in MAT vs. Airflow table.

Where airflow = load*rpm

The problem remains that at the same CLT and MAT (AIT), the injectors respond differently after a hot restart than under continuous conditions. Hence another variable must come into play to be used in adjusting the dead time change due to fuel / injector winding temperatures.

For now, I will continue with tuning steady state with rich idle and let EGO fix hot restarts. MAT correction will be used to keep high kPa AFR consistent over varying AIT as measured at the cold side of the FMIC. If we get some data points on the same set-up with return vs returnless, then I may make the change

One further question: is the problem solved with low impedance injectors driven by proper circuits? Just looking for more learning.

Is the output of the flex fuel sensor zero to five volts? If so, we could replace the flex fuel sensor with a temperature sensor down in the fuel rail area or a surface mounted fuel rail temperature sensor as an input for fuel enrichment on hot starts?

Hell, if that isn't an option how about a simple map switch based on this temperature input? Have your alternate fuel map 5% richer than your standard map and have either a manual switch in the drivers control area to activate it on hot starts, or have a temperature sensor in the fuel rail area to do automatic switching based on a break over temperature.

How does the OEM ECU handle this issue? We all badmouth the Band-Aid piggyback controllers... but they don't have issues like this :(

Keith

you could easily tableswitch fuel maps (or just the req_fuel) based on the fuel temp. that's incredibly simple.

the mazda ecu probably uses the pressure sensor in tank after starts to control the pump in a better fashion, not just on constantly. maybe pwm in certain situations. i dunno. returnless never made sense to me.

Returnless is used to prevent heating of gas in tank and therefore decreases the evaporative emissions.