Speedtoys

I think he covered that in post #7. I think.

Hilton

Good to know on the injector harness, thanks Bill. I'll test each injector plug with a noid light to make sure once I'm done sticking the plug wires back on the car tomorrow.

ptuomov

Sounds like your finding stuff.

I don't have much useful to contribute, but will write some random thoughts:

- The spark plug wire hypothesis is a good one.

- The suggestion to swap the wbo2 sensors between the banks is a good one. If it's a faulty sensor the reading will follow the sensor. I suspect it's not faulty wbo2 since the readings are 14.0 and 15.0 in closed loop.

- If it's cam timing, it would have to be pretty big problem to cause a 7% difference in air flow (14 vs 15.) It would also probably have to be an effect that varies with rpm.

- If you can fit your fuel pressure gauge to the individual rails, it might make sense to verify whether the fuel pressure is the same in each rail. I can't think of a good reason why it would be different on a stockish fuel system, other than maybe one of the front fuel dampers failing and obstructing flow.

- In theory, and I know nothing about practice, misfire should lead to different wbo2 readings before and after the cat. Let's say you hook up the cat but don't have the aux air pump feeding it. Pre cat, the wbo2 gets confused by the amount of unburned air and fuel in the exhaust gas, and shows a lean reading. The cat burns some of the oxygen and fuel, and therefore on a misfiring car the post cat wbo2 should read richer than pre-cat wbo2 sensor. In theory, I don't if this would work in practice.

worf928

Everyone's chasing fuel, timing, etc.

How about false air?

Poor lower injector o-rings or a poor intake gasket can introduce false air to one bank.

If its an early 87 where the fuel rails are mounted to rubber buffers rather than directly to the manifold then I would check the lower o-rings to make sure that they are deeply seated in the bores.

Hilton

Didn't have time yesterday to do car stuff - took the kids and visiting family to the beach and out for dinner. Should get time today to put the plug wires back on with a replacement boot which correctly measures 1k Ohms.

The concern I have at present is that I'm not sure how far different AFR readings are for a correctly-working 928 between each bank I've seen conjecture that the 5-8 side might naturally be richer than 1-4 (due to trumpet proximity to the intake side plates), but no actual data.

Still - if I can get it within a couple of tenths each way, I'll log with the sharktuner, maybe adjust the maps a little and see if I can get it through the emissions. Being able to legally drive it on the road will be a big help for diagnosis (currently I'm only legal to drive to/from the test center or a workshop).

It is an early 87 with the different casting for fuel rail bolts - but the rubber fuel rail mounts have been replaced with aluminium spacers the same thickness as the rubber bond, so its fixed in place.

False air is a good idea tho - I'll check with some brake cleaner once its all back together.

Thanks for the suggestions. I should have some data in a few ours once Sunday morning kids stuff is fiinshed

Hilton

Ok.. new readings, as I've put the ignition wiring back together with a replacement boot. Each complete wire is between 3.95 and 4.05 kOhm now - I took them all completely apart, measured all the bits, and matched up the boots and plug ends to get the wires as consistent as possible across all 8 cylinders.

The Narrow Band O2 sensor is providing signal to the ECU.

Wideband on 1-4 side, Narrow Band on 5-8 - reading 15.05 +/- 0.1 AFR (1.02 +/- 0.01 lambda)
Wideband on 5-8 side, NB on 1-4 - reading 14.3 +/- 0.1 AFR (0.97 +/- 0.01 lambda)

So slight improvement, but still a disparity. I have 3 evenings left to work on the car - after which I have to remove the dual-bung x-pipe, put factory cats back on, and drive it 40km's down to the test center. I'm thinking I'll use the NBsim from my wideband so I can data log with the sharktuner on the way down there and during the test.

• Fuel pressure holds fine for 30 mins, so no leaky injectors.
• All injectors working (can listen to each injector's body with a stethoscope, and noid light flashes in each connector)
• No intake air leaks (brake cleaner sprayed liberally around) - the car has had a complete intake refresh (including replacing throttle/flappy bearings, cam covers etc.)
• Ignition wires all measure ok

So at this stage, I'm wondering three things:

1. Am I chasing ghosts? What is "normal" disparity between each exhaust bank on a stock S4?

2. Maybe just having the car running (idle + intermittent prods of throttle) for 5 minutes on lift-bars isn't enough to "normalise" the O2 sensor mixture compensation and my data is invalid? (I doubt this as I'm not seeing much variation after about 2 minutes - car is already warmed up and MAF is new). Do I need to be driving the car around before noting the reading?

3. Could it be one or more dirty injectors not flowing enough?

Item 3 above is the only thing which could reasonably be causing the problem at this stage, other than inherent design issues with the S4 intake manifold.

Does anyone have data on what AFR readings a stock S4 will have on either bank? I know at least a few people here are running a wideband on the second port of their Motorsport slc x-pipe, so hopefully someone without boost has a number, even if only from one side

Hilton

After a beer and a quiet Sunday night, I think my plan is to put the coding plug back to non-cat, adjust the idle mixture to stoich, and take new readings. This will take the O2 adjust out of the equation, which cures concern number 2. in my post above.

If AFR are still uneven, then I'll swap injectors out, which is a little fiddly, but can be done in an hour, and take new measurements, and cure concern number 3, at which point I can be confident in what I'm measuring.

Hopefully out of all this, I'll at least have some solid data on how much variation there is between each bank of the S4 engine, and a clear answer on whether using either of the 2 bungs in a motorsport X is acceptable, or whether we should all be modifying the X-pipe to add a third bung in the merge.

ptuomov

Once you get it to 3% or so, it could be any number of small things. My instinct is that the number of potential causes grows a lot once the disparity gets smaller.

Todd Tremel should have individual cylinder AFR data logged. Hacker might help you get in touch with him.

My logic tells me that variations of the order of 3% can plausibly come from a large number of sources, whether or not they show up in Todd's data.

As far as passing a test goes, logging data in conditions most similar to the test makes sense.

Here's one thing that might change with a half an hour of hard driving: The camshafts are timed slightly differently cold per factory instructions, because the factor wants to match the timing between banks at certain block temperature. The block cold you get 1 degree or so difference between the banks, but then when the engine heats up the aluminum block expands thermally and the cam timing gets equalized between the banks. Once your % air flow differences are small enough that they could plausibly be explained by small variations in cam timing, then testing the engine with the block hot will give more accurate results. With 7% differences I don't think it's worth it as that's such a large difference that it plausibly can't come from 1 degree of cam timing.

Of course it could, right?

John Speake

Hilton, what was the emissions test the car failed, and what are the pass limits ?

jcorenman

Hilton,

At this point you might well be chasing ghosts. A difference of a couple percent in fueling is pretty small-- and well within the capability of the cats to deal with.

There are differences in the intake geometry from cylinder-to-cylinder, but it is hard to believe that those are significant at idle and light load. The airflow is a tiny fraction of what it is at WOT. Idle air is provided through a tiny slot in the ISV, how can trumpet clearance be a factor?

I think injectors are the ghost here-- not leaky injectors necessarily, but simple variation from one to the next. I don't know what's typical, but a casual search for injector flow-rate variation comes up with numbers like 5 to 10% for injectors that have not been flow-matched. And that's static flow-rate, but there are variations in opening-time (latency, dead-time) which effects pulsed-flow.

Ideally you would want the rich and weak injectors equally distributed to both banks, but that's not how "random" works. If you've had them cleaned and tested, and have the flow-data, then certainly take a look at rearranging them. Otherwise ignore it, for now-- I don't think +/- 3% is going to mess up the emissions test.

The objective is to get through the test in a few days. You found the bad ignition connector which could be the problem. Did you ever post the previous test results? Are you looking for high HC or something else?

Cheers, Jim

jcorenman

I don't think this is an issue, but some details of what you are seeing for O2-adjust and O2-adaptation in Sharktuner would be helpful.

The LH has two methods of adjusting fuel: O2-adjust, and O2-adaptation. O2-adjust is the primary control, working in real-time and adjusting fuel up and down a few percent in response to the NBO2 sensor reading-- alternating between too-rich and too-lean. There is no "learning time", O2-adjust becomes active once the engine warms up and has no memory.

The "adaptation" is separate, called "O2-adaptation" in Sharktuner, and does get saved. This is a long-term average of the O2-adjust, and can be disabled and reset by Sharktuner. (This is important for tuning, if you are adjusting fuel yourself you don't want help from the LH).

So for example if O2-adjust is cycling between +4 and +10% instead of -3% to +3%, then over time the LH will adjust O2-adaptation to +7%, and O2-adjust will then be running -3% to +3% or thereabouts.

As long as O2-adjust isn't hitting the rails (+/- 20%) then adaptation shouldn't matter, O2-adjust will simply compensate.
Where O2-adaptation does matter is when the LH is running without the NBO2-sensor during warmup and at WOT and high-load. O2-adjust will be zero but the (saved) O2-adaptation is active. But that's not a factor for emissions testing, the LH should always be running closed-loop with the NBO2 sensor active.

Hilton

Its an IM240 emissions test. The test operator has a speed target to follow as the graph progresses, and it tells him when to change gear too. If he doesn't stay close enough to the line during the test it gives an invalid result and has to be repeated. In a 928, pretty much the whole test is under 3.5k rpms.

The test part itself consists of a 4 minute driving cycle of 3.1km (2 miles) on a simulated road including inclines, and a 55mph cruise. The graph below shows the profile of the test - its a standrdised test used in some parts of the USA too.

Our pass values for an '86-99 car are:

0.93g/km Total HydroCarbon
9.3g/km CO
1.93g/km NOx

The test machine is basically a dyno in a sealed room. They sniff the tailpipe directly for CO/NOx, and monitor the whole chamber for THC (so it detects stuff like fuel leaks, oil leaks, venting crankcase to atmo etc).

The test is only needed for imported or modified vehicles, or ones which have been issued an EPA ticket for smoke/smells. My car is imported, so I have to prove it meets Australian standards.

Hilton

O2 adaptation's extremes logged with the ST are +/- 3.6%, but those are the outliers. Mostly its around 2.5%.

I've swung too high on both Nox and CO in previous tests; i.e. one test too high NOx, everything else passed, and another test good NOx and too high CO/THC.

The test with (barely) high NOx was almost a pass, and I put it down to a failed hall sender causing 6 degree retard (raising combustion temps slightly), and kicked myself for not doing it as part of the intake refresh. So I replaced the Hall sender and swapped in a newer rebuilt MAF from my '89, and it ran rich (yeah I know.. a two variable change, but the AFR's were good except in the 55mph cruise part of the test, and taking days off to get tested is getting expensive)

Current results are:

THC 0.54g/km
CO 12.1g/km (fail)
NOx 0.3g/km

You can see the graphs from the test below against time/speed.

At this point the car is stock except for the air pump disabled, an O2 sensor and cats added along with the correct cat coding plug, and the CO pot disconnected.

The car is a factory non-cat car being ex-UK - so I can't run the airpump as it puts air in before the O2 sensor. That shouldn't be an issue, my old white 87 passed in the same config.

PorKen

What are you using to check the spark plug gap? A wire or feeler is the only way to go, IMO.

Small differences in the gap really make a difference.


FWIW, non-resistor, with a 1.0 gap is the Dinky-Di, mate.

HeX-Pipe™

Hilton

Feeler gauge of course I make sure I can get the 0.75mm gauge in there, which given its a flat gauge puts the gap between 0.75 and 0.8.

I chickened out of the non-resistor plugs (I have the NGK BP7ES here) - I got as far as installing the non-resistor plugs gapped to 1mm, but decided introducing more non-stock variables was a bad idea.. so the car has 8 brand-new Bosch WR7DC+ in it.

I'll install the non-resistor plugs once I'm done sorting the '87 through its emissions tests, but for now just need to get the 87 registered. At one test per month due to lead-time on the testing center, its been a few months since it should really have been driving around