AFRs going lean when engine is hot

[aceswerling]

I have a 1995 Miata with lean AFRs once then engine gets warm. I've worked with everybody I can think of, including (but not limited to) my tuner, Dimitris (aka Reverant) who built my Megasquirt, the Miataturbo website, Nigel from Flow Force who sold me the injectors, and Flyin Miata. I'm very grateful for all the help I've received so far but since we haven't gotten to the bottom of this yet, I'd sure appreciate any ideas you can offer. Also, I apologize in advance for the length of this post but I wanted to share all the available information.

As background, my car is running the original 1.8L engine with 98K miles, although the head was replaced about 20K miles ago with a unit that was built to OEM specs. I've got a Flyin' Miata turbo kit, an MS3 Basic from MSLabs running v1.4.1 firmware, an Innovate LC-2 wideband O2 sensor, and Bosch EV14 660cc injectors sourced from Flow Force (http://www.flowforceinjectors.com/). All this equipment was new when it was installed 8 months ago.

Soon after receiving my car back from the tuner, I experienced stalling on warm restarts and then once I got the engine going, saw lean AFRs of ~17.5 for the first few minutes. The air temps were quite cool when the car was being tuned but I picked up the car just as we were getting warmer weather (air temps around 60*), which affected the fueling. The stalling and lean AFRs on startup were much more prevalent during the hot weather, which is why my tuner didn't see the problem when he was working on it.

The Miataturbo folks suggested that warm fuel in a heat soaked fuel rail and injectors could cause lean AFRs on a restart due to vapor lock until the warm fuel was burned off. Consistent with prevailing wisdom, we bumped up priming pulse and cranking pulse while substantially increasing both ASE percentage and taper.

This increased ASE kept me going for a few weeks until the outside air temps got to the mid-80s. Once the engine got good and heat soaked I experienced more stalling problems and huge amounts of EGO correction. In fact, EGO reached its limit of +20% correction but that was insufficient to hit the 14.7 AFR target. You can see this in the attached log. Later, I removed the 20% limit on EGO correction and saw upwards of +35% correction required to hit the AFR target. My tuner and Miataturbo said this was a ridiculous amount of EGO correction and I needed to try something else.

I started thinking about the vapor lock again and researched volume changes to gasoline based on temperature. I figured that hotter fuel would be less dense, giving it less energy for a given volume, and potentially explaining the need to add fuel to maintain consistent AFRs. This article https://answers.yahoo.com/question/inde ... 130AA79rH6 says there's a 1% volume change for every 10*C (or 18*F). Assuming a worst case where the fuel temps change from 80*F to 140*F, that 60* swing would result in a 3% volume change. That's significant, but doesn't seem sufficient to account for a 35% change in fuel demand by the engine. I kept looking for another explanation.

The Miataturbo folks suggested a heat soaked AIT sensor could be the problem, guessing it was reporting higher temps than actual, which would cause the Megasquirt to lean the mixture. I was confused about this since the MAT density table was basically flat after 77*, meaning the Megasquirt shouldn't lean the mixture at those higher temperatures.
https://drive.google.com/open?id=0B8JS8 ... XRzbkU4YUE

But as an experiment I added fuel at higher temps for both WUE (to address CLT-based temperatures) and MAT density, consistent with the increased ASE from earlier. Examples can be found at the URLs below.
https://drive.google.com/open?id=0B8JS8 ... FBJYkY4M28
https://drive.google.com/open?id=0B8JS8 ... 05VYU5WVmM

This helped the situation by removing the need for massive EGO correction at higher temperatures because the tune was baking in the enrichment. However, subsequent observations indicated the lean AFRs were not directly connected to high MAT or CLT. Last weekend, for example, was cooler with air temps around 60*F. I saw MATs between 80* and 100*, with most of the driving around 80*, and CLTs between 195* and 205*. As I drove more, meaning the engine became increasingly heat soaked, MAT and CLT stayed pretty constant but EGO correction increased. I don't have a log of this behavior but I could get one if you need to see it.

I also noticed AFRs in boosted conditions that were richer than the autotuned targets. An example would be 11.0-11.2 instead of the expected 11.6-11.8. I'm guessing that's because the extra fueling added by my tune at higher MATs and CLTs was overcoming the overall tendency toward lean AFRs.

Either way, the increased EGO correction with constant MAT and CLT values tells me the change isn't driven by those settings. That implies the problem isn't in the tune, but in the hardware instead.

By process of elimination, I next considered that the problem could lie with the injectors. Perhaps it's because they're responding more slowly when they get hot (more dead time?), which means the Megasquirt needs to send a longer signal to hold them open for the same amount of time? That kind of PW variation with temperature would seem to be a design flaw. The fact that I seem to be the only guy with this problem tells me that's an unlikely explanation.

I compared my dead time curve against the instructions at http://www.flowforceinjectors.com/instructions/. I'm running the OEM standard 43.5psi of pressure at the fuel rail so my configuration is consistent with the reference documentation, and Nigel agrees it's configured properly. That means the Megasquirt should be running the injectors as intended.

It's possible I've got a bad set of injectors and Nigel has graciously offered to replace them in an attempt to troubleshoot. I'm unconvinced the injectors are defective because that seems possible, but unlikely.

So overall, it seems my car is running in a bizarro world where the engine is running lean when it should be tending towards rich. Addressing the problem with MAT and CLT enrichment isn't getting the job done because the lean AFRs appear independently of those corrections and the injectors and fuel pressure appear to be configured properly in the tune.

I'm really stumped and wondering what else could be the problem. Any ideas?

[elutionsdesign]

I think the last thing I would mess with is the MAT curve, in speed density mode you depend on accurate IAT to calculate density so don't feed it bad information. Also why do you have MAF enabled? Not that it has anything to do with your issue but I'd turn that off. No, you need to play with the MAT/CLT correction, what you describe is what that is made for, read the little tooltip under that feature.

[ntstlgl1970]

What is your fuel pressure like when the symptoms occur - the logfile shows max fuel pressure as 17psi? Is that correct?

[aceswerling]

@elutionsdesign, that's interesting about the MAT table. I understand it's supposed to be set to 75% fueling when hot based on the ideal gas law. The consensus in the Miata world appears to be that causes overly lean conditions when the engine is hot. There was a recent thread at http://www.miataturbo.net/megasquirt-18 ... ike-89026/ where it was explained the Megasquirt code doesn't quite account for the math based on that formula and that table is really about adjusting fueling based on air temperatures. Basically, it said to set MAT density based on what your engine needs and not what science says it should be. On the other hand, I received direct feedback that I shouldn't need to enrich at higher temperatures like my engine has been demanding. Thoughts?

Regarding the MAT/CLT correction, I've looked for documentation online for this feature and it seems like it wasn't completely implemented in the code. I haven't been able to find examples of how to use this feature. Can you point me towards something telling me how to use it correctly?

@ntstlgl1970, I've seen a consistent 43.5psi fuel pressure at the rail every time I've looked at my inline mechanical pressure gauge. I don't believe my Megasquirt is wired to receive a fuel pressure signal so I don't know where it would be getting 17psi. Of course, if there were nothing at all why would it be showing anything other than zero? Also, I've been figuring fuel pressure is sufficient or else I would see lean conditions under boost, not the rich conditions I'm actually experiencing. Is that correct? If I'm honest, I'll admit that I haven't explicitly verified the fuel pressure when the engine is hot.

One last thing to add, which may or may not be a factor. I want to keep us focused on the lean issue I described, but I'm also getting about 19mpg in a mix of city and highway driving. I'll get ~30mpg if I'm only doing highway. This compares with 26-28 mpg combined, which I've heard from Miataturbo and Flyin Miata is typical fuel consumption for turbo/Megasquirt Miatas. We haven't been able to determine why my fuel consumption is so high because the AFRs are generally hitting reasonable targets (~14.7 when unboosted, ~11.7 when boosted) when I'm not experiencing the lean conditions at high temps. I'm confused about why my consumption would be so much higher when I'm running stoich. FWIW, I recently recalibrated the Innovate O2 sensor and I'm using a CAN bus interface so I know that signal is accurate.

[nismoautoxr]

Being a miata owner and the tuner of the car I can tell you right now.that the mat correction is what is likely causing the problem. I am in the process of trying to learn how.to tune the mat/coolant table but until that is accomplished I'd take and put the corrections at 100 above 80f. Bad info is 1 thing but if you are leaning the engine out to the point of driveability symptoms....which is worse? I'm not saying do it permanently but for sure it's a viable work around untill then.

My car runs like poo with the mat correction curve set as default. It's 90 here now. I think there needs to be some more docs added for the coolant/mat code as well because it's so imperical right now.that it's a total shot in the dark. You take and tune that mat curve to 100 and the car runs great. No one can argue that it dont. If the afrs are good then where is the harm?

I'm essentially in your boat. 1.8 na non vvt with 650cc ev14 injectors with stock fuel system and stock motor. I'm prepping the car for turbo but it's not installed yet. Fm clutch and a begi coolant reroute. I'm trying to work out all these little bugs before the turbo gets installed but the air density thing is pretty much an issue that's holding that progress. Please disregard the periods in wierd places...I'm doing this post from my phone lol.

As to your fuel.consumption...it was all mazda could do to get 260 miles to a tank on that motor and I can assure.you it was way leaner than stoich. I'm sure you've noticed the car runs so much nicer in town now. It's because your afrs are at 13.5 to 14.7 instead od 16 :1 which is where the stock one runs putting around at mid throttle. My car with no turbo is getting about 225 miles to a tank full of fuel now on an ms3 but so far I've not had a whole tank of gas without at least 1 or 2 tuning sees ions in it which will definitely gobble some fuel due to the car not moving. I've been working on idle enhancements with ac lately so a lot of idle time. Before ms3 my car only got about 260 miles of in town miles to a tank. It's a 97 all stock car right now all the way down to the stock air intake and the Maf sensor. I'm not using g the Maf but I am using the it inside the Maf so it's all still in place. I figured if I tuned it as produced ...that was a good baseline since there are no excuses . The car runs way better now than 9t ever did with the stock ecu in place but that is at a cost of fuel economy. As I tweak my tune I'm sure my mp3 will get better. I figure mine is mostly in accel enrichment and maybe in being a bit too aggressive between 50 and 100 kpa on my afrs.

[elutionsdesign]

I've tuned a half dozen or more Miata's with turbos, NA and superchargers and I've never touched the default MAT correction, its the law. I read that thread and I think...., well, I won't say what I think. Go to a professional.

Back to the MAT/CLT correction, I have used that with good success, not on a Miata but on the this thing: http://www.dailyturismo.com/2015/04/rea ... -with.html, the IAT was buried in the back after a long run of fiberglass intake. Realize at very low flow like a hot start the intake air is very likely close to coolant temp or worse but that air is a lousy heat transfer medium so as the intake air starts to move through it cools to much closer to exterior ambient temps. The IAT sensor will be slower to cool and will lie to the ECU about the actual air temp. Using the MAT correction table to compensate is, in my opinion, the wrong way to attack the problem since you need that for the other 95 to 99% of your driving experience. Tuning those low flow situations with the MAT/CLT correction is the way to go.

And this pops up on a lot of original OEM MAF based systems transitioned to speed density, the MAF based system just doesn't care a whole lot about IAT since they have already measured actual mass air flow but a speed density based system needs an accurate air temp because its.... AIR DENSITY. PV=nRT, its the law. In this case I move the IAT to a better location, typically it's a turbo conversion so you can stick it between the intercooler and the throttle body. I've seen more than one mod motor IAT record coolant temp better than the coolant temp sensor.

[nismoautoxr]

That's all good and I will not disagree that it's the law but I'm running an na miata with the iat in the stock location which is in the Maf sensor. It is pulling in the engine compartment temperature....just like it was when mazda tuned it.I'd be thrilled to tune the coolant/mat table If someone would put together a.decent theory and operation with some examples and visual aids to help a brother to understand what the goal is. I have another thread in the tuning forum so feel free to share there too.

You say where you drive 90 to 95 percent of the time...well my climate.right now is 80 to 90*f so the mat correction is near 100 normally anyways. The only time I see higher right now is in traffic idling and trust me....it sux with air density as the law. At 100 it runs at what I tell it...that is 13.5 to 14 and not 17 and 18

I'm not being argumentative but it's easy for someone to say how it's a hack to mod the air density table but it's yet.another to back it up with real solutions. I have been in the driveability game for oems for 25 years. I understand what we are talking about. The code and how it deals with this is what I don't understand. That is what I'm striving to understand. Right now I'm giving my my motor the fuel it needs....it's not surging or backing or stalling. It may be using more than it needs at times but.that is not in excess. It's getting right about 20 combined mpg.

How many of the miatas that you have tuned drive every single.day in stop and go real life traffic with the ac running and never stall or surge or stumble? I've got this one doing that but the air density table had to be model to get that done. The best tuner for any car is someone who drives the car everyday.and has a firm understanding of how things work . I'm such a person but I don't understand the coolant /mat feature and how to arrive at real numbers. I could PLAY with it sure...but how do I get there mathematically and scientifically?

[ntstlgl1970]

I only asked about the fuel pressure because it Looked weird in the datalog.

[aceswerling]

Many thanks to everybody for responding. I sure appreciate the help.

Regarding fuel economy, it's interesting that the OEM ECU runs so lean. My car is running quite smoothly with the Megasquirt and I'm happy about that. I asked about fuel consumption because feedback from Miataturbo indicated folks were getting 26-28 mpg with Megasquirt/turbo Miata setups. Flyin Miata reports similar results for their Hydra/turbo setups and I was getting 23mpg with my previous Link/Jackson Racing supercharger setup. I was expecting improved fuel economy after going turbo since we've removed the parasitic losses. I'm surprised to have a decline to 19mpg instead of an improvement to 26. That's a pretty big difference and makes me wonder what's going on. I'll let it go if 19 is really the best I can expect but the engineer in me wants to square all the information I'm receiving.

This issue combined with the lean conditions when hot have made me question whether the Megasquirt is the right choice of ECU. But I've got one now and I'd like to get to the bottom of this.

@nismoautoxr, it sounds like we're in similar situations, wondering how to tune MAT/CLT correction. If I read your comments correctly, you're also running MAT density near 100% at higher temperatures to correct for lean conditions. And those lean conditions are worse if we stick to the curve that finishes at 75% at MATs of 248*F. So maybe there's something about Miatas that doesn't work well with the default MAT density curve? Or maybe there's a collective misunderstanding?

@elutionsdesign, I'd to investigate the MAT correction a little more. I can understand the need to set the MAT density table to reflect PV=nRT as I remember all those Boyle's Law calculations from high school chemistry. For the record, we're talking about a curve that looks like this, right? https://drive.google.com/open?id=0B8JS8 ... GUzdnlSdjg

The science of that makes total sense. But if I understand the MS behavior properly, a curve like this would lean out the mixture quite a bit at higher temperatures, which would move me in the wrong direction. I figure that behavior is why Miata folks - rightly or wrongly - adjusted the table to look like https://drive.google.com/open?id=0B8JS8 ... XRzbkU4YUE instead. Thoughts?

The bit about IAT placement makes some sense and it's consistent with guidance on Miataturbo. My IAT is placed in the standard location for Flyin Miata kits, which is on the hose between the intercooler outlet and the throttle body. That's inside the engine compartment and I can see how it would be affected by heat soak. You can see pictures at https://drive.google.com/open?id=0B8JS8 ... GR3VjdNM3M and https://drive.google.com/open?id=0B8JS8 ... WMtRkZHSFk. The Miataturbo folks suggest an installation on the intercooler outlet so it's less affected by high ambient temps, and therefore less likely to indicate an inaccurately high signal. I asked Flyin Miata about this and they said the default placement isn't a problem with the Hydras they sell but they could see how it could be an issue. They concurred with a relocation to the intercooler so it appears that's the consensus best location for an accurate air temp signal.

Even so, let's assume that my IAT is heat soaked and indicating a higher temperature than actual. A MAT density table that follows the science of the ideal gas law would respond to this inaccurate signal by taking our fuel, resulting in excessively lean mixtures. But if I'm running a MAT table that's at 100% at air temps above 80*, the fueling should stay the same - at least based on this setting - because the Megasquirt wouldn't be removing fuel. If anything, the AFRs should be getting richer and not leaner. If so, the fact that the mixture is leaning out at higher temps implies there's another factor affecting fueling. Am I thinking that through correctly? Either way it seems like I need to move my IAT regardless of what the logic is telling me.

Now let's drill into the MAT/CLT correction table. I like the idea of variable fueling based on air flow velocity through the manifold. It seems like this could add fuel at the lower flows in the unboosted portion of the VE table and then remove the fuel at the high flows in the booster portion. This would certainly be a desired correction to for the behavior I've seen.

But it appears this feature works by adding some CLT% to the MAT number. If we do that, wouldn't it generally result in a higher number? And considering that the Megasquirt generally wants to remove fuel at higher temps, wouldn't that result in a leaner mixture at low air flows, not a richer one like my engine needs?

It would be really helpful to see an example though. Can you share what you did for the Porsche with the Toyota engine? I searched the internet and only found posts like http://www.msextra.com/forums/viewtopic ... 31&t=38619 and http://www.miataturbo.net/megasquirt-18 ... ble-80897/ saying not to touch this setting. That second post actually says this setting will pull fuel when I should be adding it.

@ntstlgl1970, I appreciate your comment about the fuel pressure. I'm not sure where that weird number is coming from but I'll try to find out.

[Dohc Rocks]

I have ran Megasquirt since MSnS-E debuted in my 98 Turbo Neon.

I have been through every MS platform from MSnS-E to MS2 (first ever sequential Neon) right up to my current MS3X setup.

Lean and shitty hot starts and unstable long term idling has always been an area of annoyance.

I have done everything to try to mitigate it including moving the sensor to the intercooler cold pipe. The only partial solution is to dial out the IAT correction but that brings a whole different set of tuning issues with relation to real air temp fluctuations (rich on hot days, lean on cold days)

My gut feeling is that the standard-use GM sensor might be slow to react to changes and compounding the underlying issues.

I think the solution lies in making the IAT correction influence relative to engine rpm.

[nismoautoxr]

I moved away from speed density to Maf on my 240 sx v8 and it runs really well. I may do the same with the miata.

[aceswerling]

@dohc rocks, thanks for your comment about the lean starts. I'm not thrilled to know this is a common problem, but at least I know I'm not alone. I've been able to get my starts reasonably good by jacking up the cranking pulse and ASE, even though that technically shouldn't be required.

I can see how the GM IAT sensor could be the root of the issue. Has anybody investigated a replacement?

Also, very interesting that the MS would be having these issues. I wonder if I should switch to the Hydra...

In the meantime, I'm looking forward to comments on my MAT/CLT correction questions.

[nismoautoxr]

Just to be clear ....I do not have hot restart problems as I have tuned that out with the mat density table. My car runs better than it ever has on the stock ecu. I may be using more fuel than 8 need to but that's not really a huge.concern of mine. I think most of my economy issues are going to subside the more I dial in accel enrichment. SD is working fine at the moment but if need be I'll go to Maf because I got really great results with my LS swapped 240 and was getting about 20 mpg combined on a 400 hp v8 so I think that's pretty good.

Hang in there and keep.trying things . It's bound to get better. I still think that ms is hands down the best thing out there bang for buck and I continue to use them . I learn more.about tuning them every day.

[BimmerSchnitzel]

@aceswerling

I'm dying to find out what your solution here was, if any. I'm having the exact same issues with my 90' miata with a 04' engine swap. You haven't logged in since 2018 so hopefully this finds you somehow.... My testing keeps bringing me back to fuel and injector heatsoak but the car is using OEM fueling components. Both my Spec Miata racecar and the swapped car with the stock 1.6 never had these issues..Could there have been some trick to compensate in the OEM ECUs

[nealric]

I've struggled with this as well on my turbo Alfa Romeo. My band-aid solution has been the flat MAT correction curve advocated by the Miata folks. To avoid leaning out on cold days, you can leave the cold side of the table alone- just dial out the correction for hot. Worst case scenario is running rich under boost on hot days, which may actually be indicated for detonation resistance anyways. Anyhow, the idea of MAT/CLT correction makes sense, but it seems like all it does is further lean things out. How does one set it up to add fuel under the heat soak situation?

Interestingly enough, it wasn't really an issue until I modified the head (bigger cams, oversized valves, port and polish). The stock Alfa head would flow very poorly compared to the more modern Miata head design, but the modified one is probably closer (if not better) than a stock Miata head. Perhaps MAT correction based on the ideal gas law is more accurate if air not flowing through the head as well. I also found my motor needed more air and fuel to idle compared to stock.

[MalcolmV8]

My gut feeling is that the standard-use GM sensor might be slow to react to changes and compounding the underlying issues.

I think the solution lies in making the IAT correction influence relative to engine rpm.

I read this thread with great interest as I've been fighting the same hot start issues on my V8 engines. It was great to see this comment in particular pointing out that IAT correction should be relative to engine RPM. That is spot on and exactly what I need. I've requested this multiple times but it never seems to catch on with the developers. If I could stop the ECU pulling timing at idle on hot starts because of artificially high IAT readings and only pull timing under load in boost / higher RPM it would solve a lot of these issues.

[DGA]

In Basic/ load settings, MAT/CLT correction is one solution, and also Startup/ idle, Cranking / startup settings, Ignore MAT correction during ASE turned on..... Should solve your problems. But placing the AIT sensor in a realistic place is really what you need to do first. Odds are that the rest of your tune is off somewhere as well, causing your start issues.

[MalcolmV8]

In Basic/ load settings, MAT/CLT correction is one solution, and also Startup/ idle, Cranking / startup settings, Ignore MAT correction during ASE turned on..... Should solve your problems. But placing the AIT sensor in a realistic place is really what you need to do first. Odds are that the rest of your tune is off somewhere as well, causing your start issues.

Relocating the IAT sensor is not possible when you have a roots type blower sitting on the engine. It needs to read the air in the lower intake manifold under the blower. It's not like a turbo, centrifugal or NA car where you can place the IAT sensor on a pipe somewhere out of the engine bay.

I've played with MAT/CLT correction and that made things worse because in addition to fueling changes it pulls timing. Pulling timing is exactly what I'm trying to avoid so it makes the situation worse.
I do use ignore MAT correction during ASE but ASE doesn't last very long so it's a few seconds at most after a hot start up. Where as a heat soaked IAT sensor sitting under a roots blower can take a few minutes to start showing accurate temps.

The stock factory ECU approached this problem by adding an axis to the MAT / Timing table so it would take into account engine load. That allows you to only pull timing under boost or heavy throttle situations and ignore the heat soak at idle on warm startup so the car idles correctly. Hence my request. I didn't figure it out, I'm just copying how OEM solved it as their fix works quite well.

Thanks
Malcolm

[DGA]

Looks like you can have up to 50 seconds of ASE if you want, that should be enough time to pull the AIT down to reasonable levels. Plus, you don't have to pull timing at MAT based timing retard if you don't want to, change those settings to zero.

[MalcolmV8]

Looks like you can have up to 50 seconds of ASE if you want, that should be enough time to pull the AIT down to reasonable levels. Plus, you don't have to pull timing at MAT based timing retard if you don't want to, change those settings to zero.

It takes about 2 to 5 minutes to un heat soak the IAT sensor. I think we're crossing fuel and spark too much here. On a warm start my big issue is timing pull. I need the ECU to not pull timing at idle because it makes it idle terrible. Especially on my cam'd motors that are sensitive to timing at idle. I do have to use MAT based timing retard as it must pull timing under boost if my IAT temps climb too high to avoid detonation. IATs can climb really high on a roots blower with repeated heavy throttle or an intercooler fluid pump fails etc.
The problem MAT retard is that table is a simplistic 2D table no idea if the engine is under heavy load and boost or just sitting there idling. It only looks at MAT and then pulls timing accordingly. It needs to take into account engine load, or RPM etc. And actually it really should be flex fuel aware too as you need it to pull different amounts of timing at different temperatures for pump gas vs E85.