Miata Woes: Random Lean Condition

I apologize in advance for a very long initial post. I've been trouble shooting this issue for several weeks now, so this is a compilation of what I know and what I've tried. I'm wasting the summer troubleshooting this problem instead of being on the track.

{paging Keith)

Here's what we're working with...

1992 Miata 1.6L. Greddy tubo, starion IC, 315cc injectors, stock fuel pressure regulator, Walbro 190 fuel pump, EMB with olderguy O2 clamp & Autotune. Zeitronix WB.

I'm data logging with the EMB and Zeitronix WB.

Problem:

Non-boost cruising, the car will go randomly lean. Steady rpm/load (or even at idle), AFR will go from ~14.5 to 17.5-19.5. Will typically stay that way and then may come back down randomly and stay. Sometimes when this happens, I can watch the AFR oscillate in a slow wave format, typically in the 15.5-17.5 realm.

The problem only happens once the car is warmed up, never in cold open loop. What's weird is that there's almost a "click" and all of a sudden the engine changes, runs smoother, and the AFR goes back to 14ish.

The problem isn't limited to just non-boost, it happens in boost too, though obviously that's a much bigger problem, so I've put in-boost tuning on hold until this is resolved.

Things I've tried:

Originally, it was idling very lean (17-19) with the injector compensation adjusting for 205 to 315. My current tune has the injector size at 250cc. Cold start AFR is very steady. Initially it's 12.5 and then slowly works up to 14.5 as the engine warms (open loop).

When the car isn't going lean, steady cruise AFR will be in the upper 13s typically.

I replaced the heated 4 wire NB O2 sensor which is in the downpipe, just below the turbo.

I also ran a new O2 signal wire and a new ground wire, thinking there could be a short somewhere. I've also tried a couple of different grounding locations, but no change.

I opened the AFM and cleaned the contacts. No real change.

I got another AFM and installed it. No change.

I installed a new fuel filter back in Nov 2013.

Confirmed that the injectors are high-ohm units.

I wired the O2 signal directly into the ecu, essentially bypassing the o2 clamp unit. No change. I also set the injector compensation in the EMB to zero and did a cruise. While significantly richer, it still had the random lean spike. (For further testing I've gone back to the original setup with the O2 signal first passing through the o2 clamp)

Confirmed 12v at the fuel pump.

Last week I discovered a short in my ground wire. The boost controller, autotune, NB O2, Zeitronix unit, boost sensor, and pressure switch (for the autotune), all share the same ground and they were all spliced into each other. The previous owner used an excessive amount of wiring, so I was suspicious that I had introduced the short by having to pull out all the excess wiring during my testing. I felt it was "A" problem, but not convinced it was "THE" problem.

Any ways, I basically rewired everything and changed the common ground location, just in case there was an issue. So the wiring is much cleaner, and no more electrical short, but the lean condition continues.

I also got a chance to drive it with the mechanical fuel pressure gauge hooked up. Obviously I can't see the gauge while driving but I was able to observe the fuel pressure at idle.

At least at idle, there's no change in fuel pressure, even when it's running lean. During warm up, open loop, the fuel pressure is 36 psi. After driving and getting it fully warmed up, the random lean condition kicks in. At idle, 17-19 AFR, the fuel pressure is still 36psi.

I bought a Zeitronix fuel pressure sensor that I've wired into my Zeitronix unit to data log my fuel pressure. I'm having some trouble getting the signal to display properly but at this point, I don't think that's the issue.

The charcoal canister is disconnected from the intake, and plugged. The PCV lines are routed to a remote canister.

I pulled the throttle body and cleaned it (probably the first time it's been off since 1992.). Replaced the tps. No change.

I read where some people had issues with the transmission neutral switch (how the ECU exactly interprets the neutral switch is a bit fuzzy but it was suggested that it would indicate the computer go into a leaner idle fuel map). However, I discovered getting to the trans neutral switch was going to be a pain. But after some research, I found that basically the ecu sends a voltage to both the neutral switch and the upper clutch switch. If either switch is engaged, it pulls the circuit to ground. So disconnecting both sensors would essentially be telling the ecu that the car is never not in gear.

So that's what I did. Cranked up the car and let it warm up. Got to operating temp, and the AFR was still right at 14.7. Feeling good so I took it for a drive. All the way out of the neighborhood it was mid 13's afr. Finally get out on the road and by the time I get a mile down the road, it goes super lean again and stays there. Afr's are kinda all over the place but always very lean.

Come back home, car idling in the garage... Afr is mid 13's.

I'm wholly and truly baffled and I'm running out of things to guess at. I can't figure out if the ECU is commanding the lean condition because it's bad, or because a faulty sensor is telling it to go lean. Or if there's a vacuum leak somewhere.

The idle is pretty consistent though, aside from running a tad rough because of the extreme lean condition, so I don't think it's a vacuum leak. All the rubber hoses are in good shape.

Thanks in advance for any suggestion or thoughts on the matter.

What's an EMB?

I've seen that exact behavior from a fuel pump on the way out. Then it got worse.

The ECU uses the neutral switch to tell the difference between closed throttle deceleration (injectors turn off) and idle (injectors fire).

If you have a fuel pressure gauge, use a long bit of rubber hose and tape it to the windshield. Go drive. Check fuel pressure first.

Rob R.

I assume this is an internal wastegate turbo? have you checked to make sure nothing is pinning the wastegate op rod?

EMB = Greddy Emanage Blue (piggy back system)

If it is the fuel pump, wouldn't I see a drop in fuel pressure? For example, I've seen 36psi at idle, at both 14 AFR and 19 AFR. Also, wouldn't the ECU see the lean condition and try to compensate by increasing the injector duty cycle? At idle my injector duty cycle is 1% and cruising around 15%. No crazy spikes.

Yes, it is an internal wastegate turbo. If the wastegate were stuck, shouldn't I be able to see that on my boost gauge? Either I'd have a hard time building boost, or I'd have run away boost. ??

I'm going to go back and recheck the inj duty cycle at 14AFR vs 19 AFR. If they're the same, then either the fuel pressure/volume is dropping off or unmetered air is being introduced. Right?

Thanks for all the input and being my sounding board for thoughts and ideas. It is greatly appreciated. And if I've expressed an incorrect understanding of something, please don't hesitate to correct me. Knowledge is good.

Dumb question: Are you smoking at all when you go lean? Does it smell like fuel? An excessively rich mixture can read as lean..

Next dumb question that I didn't see listed above: Did you use a stethoscope (or a long screwdriver or other solid rod) to listen to the injectors firing when you go lean?

Along those same lines, do you have access to a IR thermometer? Check the header at the exhaust for each port. When you run lean it should get hotter, so get a baseline under normal conditions and then see what happens when it's lean. If all ports are the same temp, you've got a wiring or computer problem (all injectors are doing the same thing). If one or two of the ports vary, you've got a wiring or injector problem.

I haven't noticed any excess smoking.

I do have a stethoscope. I'll give the injectors a listen.

I don't have an IR thermometer, but I am running an EGT gauge. While I can't check individual exhaust ports, I can say the EGT's are much hotter when it starts running lean.

Just a follow up after reviewing my fuel pressure data logs from tonights test drive.

Essentially zero change in fuel pressure during the cross over to the lean condition. 13 AFR to 16.5-17 AFR.

Fuel pressure ranges from 36 psi at idle to around 43 psi approaching zero vacuum at 3300 rpms, mild acceleration.

It's a little tricky to correlate the injector duty cycle and AFR because my Zeitronix unit logs the WB O2 while the EMB logs the duty cycle. But the best I can tell, the injector duty cycle drops from 12% to 10% when it kicks over to lean. Assuming all other factors held steady, how large of a change in AFR would you expect to see in a 2% drop in duty cycle?

I hate to be that guy, because I hate those guys, but throw away the emanage and the rest of those band aids and get a Megasquirt, throw away the AFM and go MAP. Otherwise you're just throwing so much time away troubleshooting what is essentially a system that will continue to have a variety of problems that most people either don't remember how to solve or have no interest in solving.

Also the MS will run the car way better in the end and you are just better off putting your time into tuning that.

I doubt that throwing a bunch of electronics at the car will make it run better - I owned a Miata with a home made turbo kit on it that was controlled by an EMB a while back and you could get that to run OK. Yes, it would run better with an MS and MAP, but I'm not convinced yet that the problems are with the piggyback and not somewhere else.

BTW, how many O2 sensors are you running?

Harvey - I do agree a MS would be better and I may end up there (and I do appreciate the honest suggestion). But the EMB has worked for so many people, and my goal with this car is cheap track time, so I'm trying to make what I have work. As BoxheadTim mentioned, I'm not yet convinced the issue is with the EMB. During non-boost, the EMB isn't changing any of the fueling. Really the only thing it is doing is the injector size compensation and the timing adjustment table. I'd hate to spend the money to convert to a MS and still have issues because of an OE sensor, etc.

BowheadTim - I've got two O2 sensors. One is a 4-wire heated narrow band that feeds the stock ECU. The other is a Zeitronix wideband that feeds the gauge and data logging unit.

I would first try swapping the injectors. You say it goes lean without any boost just steady state cruising. See if it still does it with say the stock injectors. Turn off the boost while you do it.

If it still does it, swap to a known good fuel pump.

After that, swap to the MS.

I've never tried running a four-wire O2 sensor on a stock 1.6 ECU. It's possible you're seeing different results than the ECU expects from the original unheated unit. I'd swap back to see.

I'm not opposed to swapping the injectors or the fuel pump. However, I have don't have spares for either, so that will involve buying both new. (Again, if that fixes it, it's money well spent).

However, I will also mention that I initially started working on the tuning because I was going very rich during boost conditions (9.9 AFR). So I have seen evidence that both the injectors and fuel pump can put out a lot of fuel.

I'll also mention that I found posts by the previous owner on miataturbo.net discussing an extreme lean condition at idle, which I also have/had (lying about the actual size of the injectors in the EMB injector compensation settings allowed me to richen the mixture across the board). So I assume the source of this problem has existed for years.

So I guess I have a few questions...

1) If the fuel pump or fuel injectors were faulty, would I not see evidence of that in the data logs?

2) Is a 2% reduction in the injector duty cycle enough to cause a change in AFR from 13:1 to 16.5:1?

3) If we assume that 2% change in injector duty cycle did cause that drastic of a change in the AFR, then it would appear this was a commanded change by the ECU... either because the ECU itself is faulty or there's a faulty sensor somewhere sending a bad signal. Right?

You can probably find a stock pump and injectors off Miata.net or on miataturbo.net for very little.

If you aren't using an adjustable FPR for fueling then you don't need anything other than the stock pump.

bigmackloud wrote: I'm not opposed to swapping the injectors or the fuel pump. However, I have don't have spares for either, so that will involve buying both new. (Again, if that fixes it, it's money well spent). However, I will also mention that I initially started working on the tuning because I was going very rich during boost conditions (9.9 AFR). So I have seen evidence that both the injectors and fuel pump can put out a lot of fuel. I'll also mention that I found posts by the previous owner on miataturbo.net discussing an extreme lean condition at idle, which I also have/had (lying about the actual size of the injectors in the EMB injector compensation settings allowed me to richen the mixture across the board). So I assume the source of this problem has existed for years. So I guess I have a few questions... 1) If the fuel pump or fuel injectors were faulty, would I not see evidence of that in the data logs? 2) Is a 2% reduction in the injector duty cycle enough to cause a change in AFR from 13:1 to 16.5:1? 3) If we assume that 2% change in injector duty cycle did cause that drastic of a change in the AFR, then it would appear this was a commanded change by the ECU... either because the ECU itself is faulty or there's a faulty sensor somewhere sending a bad signal. Right?

The more things that come out the less enthused I would be to continue down this road.

The fact that you are changing settings in the ECU intentionally to technically improper settings (IE telling the ECU it has different sized injectors) to compensate for issues with fueling is just very worrying overall and tells me that anything you do on top of what you have already done is just guess work. The fact that the original owner had similar or other fueling issues that he could not work out while he owned the car is also pretty worrying.

At least at idle, there's no change in fuel pressure, even when it's running lean. During warm up, open loop, the fuel pressure is 36 psi. After driving and getting it fully warmed up, the random lean condition kicks in. At idle, 17-19 AFR, the fuel pressure is still 36psi.

Okay, going just by this from your OP the fuel pump seems to be working and it would point to either the injectors are bad or the ECU is just plain not working right. I'm going with the ECU being the problem. Did you check all the wiring for this thing into the injector harness?

If we're talking small throttle openings then 2) is a possibility. After all, you're not talking about a 2% change here, going from 12% to 10% is more of a 16% change.

Re 3), I don't think it's a problem with the ECU, my money would be on a sensor. As Keith said, it might be a good idea to stick an OEM 1.6 sensor in there. Also, are you 100% sure you don't have some sort of occasional vacuum leak or similar?

When does the EMB engage to override the stock ECU signal? It sounds like the thing doesn't even engage at idle right? If that's the case then the stock O2 sensor might be the culprit like BoxheadTim and Keith are saying.

So here's a quick run down of how the system works.

The factory O2 signal feeds into the O2 clamp unit. The O2 clamp is run off a boost switch. Under non-boost, it routes that O2 signal to the ECU, unchanged. At the same time, still non-boost, it sends a synthetic signal to the EMB, indicating no fueling change needed. This allows the factory ECU to run as normal during non-boost conditions.

Under boost, the O2 clamp now sends a synthetic signal to the factory ECU, telling it everything is at 14.7 AFR, and no fueling change is needed. It now sends the real O2 signal to the EMB. This signal is input as a TPS signal and we now build a fuel map based on this.

The theory is that the narrow band oxygen sensor is very repeatable in the area we would like our AFR’s to be under boost, and that the Emanage can be used to add fuel to maintain that AFR based on a setpoint and deviation from that setpoint.

The factory computer has no ability to know that the injectors are larger than stock. So the EMB has an injector size compensation setting. Essentially it alters the AFM voltage to compensate for the injector size. Basically, it tells the computer that there's less air flow than there actually is, so the factory ECU slows down the injectors.

To this point, I've tried setting the EMB injector size compensation to zero (ie no change). As expected, this makes everything much richer (the factory ECU is running the injectors as if they flow 205cc, but I'm actually flowing 315cc). However, the lean condition still happens, albeit at a richer setting. For example idle might be at 12 AFR, but then it goes "lean" to 15 AFR. It just shifts the scale up or down, but the "lean" condition still happens.

I'll also mention that one of the trouble shooting tips when running the O2 clamp/Auto Tune set up, is to convert to a 4 wire heated O2 sensor.

A single wire sensor is relying on the threads and bung in the exhaust pipe to provide the proper ground to complete the circuit. A 4 wire sensor has it's own ground wire. And since it's heated, it comes up to temp faster, and stays at that temp more consistently. (or so I've read. I make no claims of being an expert).

I thought the O2 clamp was only neutralizing the signal from the narrow band O2 sensor after a certain configurable boost pressure is reached so as to prevent lean tip in prior to the stock ECU switching to open loop mode? Essentially it is there to prevent the stock ECU from pulling fuel as the car goes into boost. So basically in boost the clamp is always sending the same voltage to the ECU, but after it goes into open loop the ECU doesn't care because it is relying on internal maps.

At least that is what I remember from when I had one and used an AFPR back in the day to run my 1.6 turbo off of along with the Bipes timing controller all of which also ran the thing pretty E36 M3ty overall compared to the MS.

All that said, at idle the thing is running in closed loop off the O2 signal from the narrow band using the stock computer, so if it runs super lean at idle then it's the stock computer and narrow band sensor to blame.

Harvey, you are correct, the O2 clamp in it's basic form is configured that way. When used in conjunction with the "Auto Tune", it is also used as a switching mechanism for the NB O2 signal, since I'm using that to tune from while in boost.

http://www.miataturbo.net/e-manage-10/autotune-your-emanage-2572/

In the first post OlderGuy has attached an info sheet about the auto tune setup, if you're in the mood for some reading.

And just to reiterate from my initial post, I did try taking the O2 clamp out of the system and feeding the NB O2 signal straight to the factory ECU. No change. I also replaced the NB O2 sensor with a new sensor.

I didn't give it a ton of thought at the time, but in replacing the O2 sensor, I could have sworn the previous owner had it wired wrong. He had the signal wire going to ground and the ground wire going to the ECU. Regardless, I'm confident with my wiring now since I ran a new signal wire and ground wire.

Ah, I forgot about that whole thing with the Emanage and the O2 clamp auto tune.

What are we talking about though at this point? You're saying that it warms up and then will randomly go way lean at idle. Idle means stock ECU and NB O2 sensor are the only things controlling fueling. So, either something is messing with the signal that is going to the stock ECU and it thinks it has to pull fuel for some reason or your wideband isn't reading properly, but since you say you notice it running rough when it shows lean I'm thinking it probably is reading properly.

And you say you swapped NB O2 sensors and it does the same thing? Have you opened up the stock ECU? Sometimes they can go bad. I've seen a few posts where people indicated they get old enough to blow out capacitors.

I'm going the same direction as Harvey. How much of this system can you disconnect and still have it run? I.e.: Can you remove/disconnect the EMB and wide-band O2 sensor (remove the o2 sensor and put a plug in it's place, it's not good to let them in the exhaust without the heater on..)? What about swapping back to stock injectors?

I'm going to have to look at it a bit closer to see how hard it is to unplug the EMB. Luckily the previous owner used a Boomslang harness for the install. So it might be plug in play.

A local Miata racer has a '92 ECU for sale (he went MS) for $150. That's a bunch of money for a test part but I guess that's life.

Luckily the wide band is not connected to the EMB, so that can stay in the car to monitor the AFR even if I pull out the EMB.

As for the injectors, I don't have any stock ones though I'm sure I could find some used. Though perhaps I'd be better off sending mine out to be cleaned and flow tested. Any injector shop recommendations, preferably on the east coast?

Where are you on the East Coast? Marren is here in CT and will service most injectors.

If you are in CT I can let you borrow my stuff. I have 1.6 injectors that are flow tested as well as a stock ECU I'm not using at the moment, which I know is good.