lbatalha

For an NA car, at least, is there any reason I cant just use EGO correction on full time, even at the top load cells and 100% TPS? This, of course, over nicely tuned VE table that gets very close to targets under the average operating conditions.
The way I see it, say if i have a 13.0 AFR target at 100kpa (all rpm range), and im hitting 12.8-12.9, leaving EGO on there would not cause any issues.

Also, how close do you guys have your VE tuned, how much AFR error do you usually see during normal use. Ive seen few videos from nice tunes from this forum and ive seen variations on the order of 0.1-0.3 AFR at WOT on NA cars. What are your usual deviations at cruise, WOT? (anyone have a megalogviewer histogram table from a recent log showing AFR error?)

18psi

iTrader: (76)

Just about everyone uses full time EGO.
Unless you're tuned to the very limit of det, you're not gonna notice much difference even if your afr swings half point at wot. at part throttle it's even less important.

lbatalha

Much thanks for the reply, I've searched for quite a while looking into this, and most of what I saw (not that many results) was people afraid of it due to possible wideband malfunction/miscalibration, "works okay for me" or "OEMs go open loop anyway".
I'm a big fan of "try it and see" but I also like to extensively research what has been done before to avoid wasting too much time re-inventing the wheel (potentially with catastrophic results due to non-obvious edge cases or just my ignorance).

I have a rev MS3 with digital wideband module, doesn't account for calibration drift but if it stops working it will just output 14.7, which ego will try to correct to run richer.

My plan is to set authority at high loads to 3-5% or thereabouts, but have it at 20-25% for the idle, low load/ low rpm cruise cells, since I've found that the injector heat soak in my returnless system is quite the pain to handle in city driving, confirmed by older threads, and takes a good 3-5 minutes to be completely gone at 3k rpm cruising, returning within a minute of siting at idle.

Spent a tuning session or so chasing my tail due to the injector issue, trying to figure out if it was MAT correction, but ruled that out quickly in megalogviewer after filtering out high mat entries, or removing MAT correction factor entirely from VE values.

18psi

iTrader: (76)

your plan should work just fine

Savington

iTrader: (31)

I am not a fan of EGO at WOT. Too much reliance on an O2 sensor which will eventually fail/stop reading correctly (as they all do, eventually). I sleep better with a ~90kpa limit on EGO and ~10% authority everywhere else. YMMV

18psi

iTrader: (76)

You can limit it per the authority table like he proposed, which would essentially be a fail safe. I've not a lot of experience on track cars, but I've seen less than a handful fail out of hundreds of street cars with years of use, so I tend to be more optimistic of their lifespan.

Savington

iTrader: (31)

They last forever in street cars, much less than forever in track cars. I mostly tune and work with track cars or outright race cars.

MX5RACER

Quoted for the truth! On my racecar and others I have tuned, I try to tune for worst case scenarios. I get the motors HOT on the dyno and try and simulate what the car will experience at the end of a race, hot air, hot fuel, etc. I have seen too many O2 sensors overheat (READ: Innovate Motorsports) on the dyno and become useless for tuning on the dyno, let alone on a track.

For me, I tune for power on the dyno in poor conditions and use the numbers the motor likes to build my target AFR's. I then will run VEAL as much as possible making as many varing conditions as possible. I do not use any EGO correction at all as the O2 sensors are too unreliable. I would hate to blow a $3,000+ motor on track because the $200 O2 sensor was not reading correctly.

Get the fuel table dialed in and there should not be large changes needed unless you have large changes in elevation that cause the air to thin out. My fuel tables are purposely on the rich side for power, and they fall within the acceptable range, lean for power, when the weather turns cooler (Disclaimer: I live in S. FL so cooler is still pretty warm, like 40-50's, lol).

aceswerling

OP, you said in your first post that you have an NA and then you mentioned your fuel system is returnless. Are you sure about that? I'm guessing you have a return system, although I don't doubt that you're running into heat soak issues.

I'm inclined to agree with mx5racer that EGO correction shouldn't be doing much. Even so, I see your EGO wants to add fuel when the engine gets hot and it's independent from MAT (and CLT too, I'm guessing). I'm experiencing the same problem and seen the same posts about injector heat soak being the problem. I've made some progress investigating the issue although I haven't found a complete solution. Would you be interested in collaborating on this?

lbatalha

I said "for an NA car" as in naturally aspirated. Sorry if it was not clear, but this is a 2001 NB2.

There are only 4 solutions I see for the injector heat soak issue:
1. Just go Closed loop like OEM
2. Install temperature sensor in fuel rail and have injector PW/dead time(?) compensation based on fuel temperature
3. Live with it and/or somehow wrap the fuel rail to reduce the rate of heat transfer.

1. Requires no extra hardware, works very precisely but can have the potential edge case of sensor failure (can be mitigated by carefully picked authority cells, having a rev digital wideband box or using CEL/Limp mode feature to pick up on maxed out EGO
2. Is more involved, requires you waste an extra analog input, requires more tuning time, requires compatible EMS.
3. For people that dont care, they just live with it. Wrapping the rail might be effective in delaying the issue (doubt it does much after being parked for 30m), but I dont know what happens if there is some fuel leak and you have a bunch of fiberglass wrap soaked in it. Probably doesn't matter much.

18psi

iTrader: (76)

option 2 doesn't work. someone here tried that a while ago. option 1 still works perfect for me

DNMakinson

iTrader: (3)

Rather than have the CAN box give a false Stochiometric at sensor fail, I prefer to actually signal the fail and go to some form of limp mode. Available on MS3. Otherwise, I agree with running a highly stepped authority.

aceswerling

@lbatalha, I misunderstood the point about NA being naturally aspirated. Now I understand about the returnless fuel system in your NB. I have the return system in my NA Miata but still have the problem.

You've clearly done some research of this. I know the forum consensus is that heat soak changes the electrical performance of fuel injector solenoids, which leads to changes in dead times. FWIW, I've had problems with lean AFRs when the engine gets heat soaked with three sets of different injectors. My latest, the ID1000s, perform the most consistently but still strongly demonstrate this behavior. The AFR variability is always independent of MAT and CLT values so that told me it's separate from Megasquirt's fuel calculations, all things being equal.

I've bounced this off Tony at ID, who says it's technically correct that heat would change electrical performance in EV14 injectors. He says it's to such a small degree that it would result in a negligible change in dead times. I also asked Jeremy at FM, who says they don't have this problem to a large extent with the ECUs they sell. (OK, fine.) He also said to let EGO take care of the problem. Aftermarket ECUs don't have the same sophistication as OEM ECUs so there's no way they can be as accurate. For whatever that's worth. Neither of them suggested where the problem might lie.

More recently I noticed the behavior appeared to correlate to either fuel temperature or pressure. I'd installed a flex fuel sensor a while back to adjust fuel PW for temperature. It's pretty easy. I needed the fuel temp correction because the lean AFR behavior was messing with hot restarts and I couldn't rely on EGO for that. I tuned for fuel temperature density changes based on data from BP's customer information department. That helped but I still saw lean AFRs when hot.

More research indicated that fuel can cavitate (?) when hot, causing bubbles in the liquid fuel. That implies the density practically changes at a rate that's different from the numbers BP gave me. I substantially turned up the temperature-based correction from BP's numbers. I'll see high fuel temps immediately after startup, which will drop pretty quickly once the engine starts running. The fuel temp correction has helped and gotten hot restarts pretty reasonable. I'm still seeing lean AFRs for several minutes until everything stabilizes. Like you, I also see them when idling and it takes a few minutes to even out after I start moving again.

So ultimately that means fuel temperature correction helps but there's still something else going on.

Reverant

iTrader: (10)

Set the failure value to 7.5, then enable the EGO warning in the Check Engine Light parameters to activate when the AFR is less than 8.0. Now if the wideband fails, you'll get a CEL and EGO will default to 100.

lbatalha

Awesome! I had the CEL setup already to warn below 8 iirc. This is probably the best failsafe yet. The digital wideband module is by far my favorite part of the setup
Rev, I dont remember reading in the module manual how to change the failure value, can you share here or via PM?

Ian

iTrader: (2)

I'm also wondering where the failure value is located.

Reverant

iTrader: (10)

You need to load the ini file for the module in TunerStudio, then you get the additional configuration menus for the module.

lbatalha

Fantastic, I will give it a shot next time I go out to tune. Just grabbed the manual from the car, and yep, on the second page there are instructions to do this >.> . Sorry to waste your time

BTW is there any link to download both the ECU and Wideband module manuals, or is that sent on a per-customer basis (found older manual for v3.0+ in TSE site)