EFI question

[jetjock]

Thought I'd throw this out on the off chance someone might be here who does in depth EFI and emissions diag.

Check a friend's Supra today prior to going in for smog. The car is stock. Engine is running an A/F of 16.5 at idle and about 15.5 at all other times. Needless to say NOx is high. What has me baffled is the O2 sensor is cross counting and injection duration is not maxed. Dual trace of the sensor and injectors shows the sensor is in command and can be driven full lean or rich via induction leak or propane. Timing and advance are OK and my DIS scope shows IGN is good without misfires.

Tried two different O2 sensors, one of which is new. Tried two different gas analyzers after calibrating them both and then checking them against each other and a car known to be running dead on. Any ideas why the loop is shifted lean?

 

[shaeff]

perhaps the TPS is malfunctioning. it wouldnt hurt to check it. the TPS lets the ECU know about throttle input, or lack thereof. here's how to check it:

http://www.cygnusx1.net/supra/library/TSRM/fi/FI_100.html

-shaeff

 

[jetjock]

Thanks Shaeff but it's leanest at idle and since the TCCS ignores thotttle input whenever the IDL contacs are made and an open contact would result in MIL indication it has to be something else.

Fwiw, in an EFI system (or any control loop for that matter) an offset in setpoint while the loop is in control is almost always caused by either a bad process variable signal (in this case the O2 sensor) or an error in the device used to confirm the offset (in this case the gas analyers). Neither seem to be the cause here so I'll have to dig deeper. And while it's true the O2 sensor doesn't always have authority and, in fact, is subservient to other engine sensors, the waveform shows it to be in command.

 

[IJ.]

JJ: Have you tried a resistor in place of the 02 to rule it out as a variable?

 

[3p141592654]

Some thoughts. If the O2 sensor is cross counting then it must be (on average) stochiometric at the sensor. Your gas analyzer says it is not, so something is fundamentally out of whack here. You've calibrated the analyzer, so what's left? I would look for a miss on a cylinder. That would make one cylinder very rich and the ECU would compensate and make the rest very lean to give an average that is okay, but the 5 good cylinders would run lean and have high nox. You are sampling downstream of the cat with your analyzer (probably?), so you would not see the same mixture that the O2 sensor sees since the cat is oxidizing the unburnt fuel? I would start by checking the Vf signal to see if the ECU fuel trim is lean. If so, due a cylinder balance test and see if one is lame. Possibly a gummed up injector or such.

 

[jetjock]

IJ: Can't do that, a resistor won't simulate an O2 sensor as it's not a resitive input. The ECU needs to see a cross count above and below stoichio. I could hook up a frequency generator to it but I've already tried two different sensors and drove them full lean and rich while in the car. I also bench tested them and both are good. Plus they cross count more than 8 times/sec (Toyo spec) when looked at using the diag connector. It can't be an O2 sensor problem.

3p141592654:

Yeah, something *is* fundamentally wrong and that's what's bugging me. How could the loop be cross counting and yet be shifted lean? Just shouldn't be so it has to be something like you suggested. I agree it could be a miss but secondary patterns show none and the spark lines seem to indicate the same mixture in all cylinders. Rpm drop also seems good during power balance and the idle is smooth. Plugs and wires are new. Btw, the 7MGTE isn't sequentially injected so keep that in mind..

VF shows 5 volts at idle so it does want to richen up. VF is normal (midrange) under load but the A/F is still lean. Not as lean as at idle but still lean. Injection is about 1.5 ms across the board at idle so there is plenty of room to spare. They're only a few months old but spray pattern hasn't been checked. Yeah, I'm sampling post cat and know I should sample engine out but haven't done it yet. It's my next step. I'm beginning to think it's a mech problem too as everything EFI signal wise checks out. I was just hoping to avoid pulling injectors, plugs, and installing test pipes ect but it may come to that.

I'll go over everything again more carefully. Thanks for your suggestions, it's nice to find someone here with a good grasp of this stuff. Thought I was alone there for a while

 

[IJ.]

Learning all the time !

 

[jetjock]

Learning all the time !

I may know the minutia.of EFI but I'love to sit and pick your brain on other stuff. I've learned a lot about mechanicals from your posts and the next time I make it down there we should hook up.

 

[IJ.]

JJ: LOL I'm an electronics Neanderthal but enjoy learning!

I had a friend explain the basics in mechanical terms and that gave me a small grasp as with no moving parts it just wasn't making sense to me.

If ever you're down here for sure let me know!

 

[3p141592654]

I was thinking about this one on the way home from work. I understand your comment about the non-sequential injector control, but even if it was sequential there is only the aggregate O2 sensor to go by, so the ECU cannot work at the cyinder level but rather makes its decisions based on the average of the 6 cylinders. If you have a drippy injector, then one cylinder runs rich. The ECU backs off all the injectors to compensate, the drippy one stays rich, and the rest go lean. This will be worse at idle, where the drip volume is a greater fraction of total fuel consumed. Have you checked the plugs? Yeah I know, that's a two hour job on the 7MGTE! LOL

 

[Nick M]

You have an exhaust leak downstream of the oxygen sensor?

Where is the CO? If the car can accelerate no problem, you arent as lean as your gas analyzer says.

How do you know the analyzer is calibrated correctly? Is the EGR functioning correctly? Do you know how to test it and pull it open with a vaccum pump?

 

[jetjock]

Nick: Thanks for your input. No leaks downstream that I can see but that's one reason I need to sample engine out rather than tailpipe. CO is just a bit low (.3%) but I'd expect that with a lean mix. HC is slightly elevated so I'm inclined to think it's a misfire in one cylinder. I have two analyzers, both calibrated using BAR 97 certified gases. The car runs fine, as it should at an A/F of 15-16. I'm trying to solve an emissions problem and not a drivability one as NOx will alway go up when slightly lean.

3p141592654: I see your point. Even though the injectors are new I may have to do a flow check/balance. Fwiw rest pressure is fine and holds for a long time. Plugs and wires are new and the scope shows no ignition problems. Not that I'm an expert at secondary pattern analysis but I've a fair amount of experience. I may pull them anyway but I tend to trust the scope. Before I do anything I'm going to sample pre-cat I'm betting the elevated HC will really show up when I do that so I'll work from there. Thanks again for your thoughts.

 

[Nick M]

Doing a visual on the front cat wouldnt hurt either.