[FIXED] Mazda 6 (06) 2L (UK) - -ve fuel trims and rough running

Hi everybody,

I was hoping that more experienced heads would be able to give me pointers on what to check first with my Mazda 6.

Complaint is, the engine is running a little bit rough. Not really rough, just enough to be noticeable. Idle is steady, but doesn't sound as smooth as it used to. On gentle acceleration it is not very smooth. Not really jerky, but not smooth.

[ETA] There are no DTCs.

So I checked it with a simple scan tool the other day, and found that the total fuel trim at idle was hovering around -9 / -10, depending how warm the engine is.

Having done some more investigations today, and graphing out Engine RPM, Total fuel Trim and O2 Voltage, I've found that the O2 sensor is pegged Rich (Voltage reported by the scan tool ~ 0.8v) at Idle. It does go lean on deceleration (95% sure the computer cuts off fuel on deceleration with manual gearbox, as the "Current fuel consumption" display in the car also goes to 0), so it's not completely stuck.

The fuel trims do go positive with sustained acceleration, but only just, they're soon back in the negative again.



So my question is, where do I start? I know that it's unlikely to be a vacuum leak, or a dirty MAF as that should result in a Lean O2 reading.

So do I start with fuel supply? Leaking injector? Could it be poor spark not burning all the fuel?

What's best to check first?

Many thanks.

Hey real_hulver! Thank you so much for adding the scan data capture. We love stuff like that.

So the Mazda 6 here in States never came with a 2.0L, but I figure they can't be THAT different from our models. Looking at the upstream O2 signal, I'm gonna guess that this doesn't have a normal zirconia O2, but an air/fuel ratio (wideband) sensor. You could confirm this by doing a quick visual on the upstream sensor wiring. If there's six wires at the sensor connector, then it's a wideband. If there's four, then it's a typical O2.

The other possibility is that the scan data is showing you the downstream as O2 S1 V, instead of the upstream. I mention this because it's not unusual for warmed up catalysts to produce a downstream signal like the one you posted. If you wanted to make sure, you could try unplugging the upstream O2 and watching the data. Should zero out immediately upon unplugging.

Some scanners don't always translate wideband sensor behavior well, and you can end up with misleading data. The fuel trims themselves support this, between frames 386 and 470. If the upstream sensor were truly fixed rich, the PCM would be working the fuel trims a lot harder to get the situation under control. I'd expect trims around -30% or worse.

Hi Tyler,

Thanks for your response. I'd not considered looking for wideband stuff, as I've not had any exposure to it.

Have a quick look at some more scan data this morning, there are pids around "O2 Sensor lambda", and "O2 sensor current wide range" as well as some enhanced mazda pids for Equivalence Ratios which I'll take a look at this weekend.

I'll have a good read of your fantastic thread on Air/fuel ratio sensor testing as well. Thank you.

Ah yeah, that lambda PID is another good sign of a wide band at work. :lol: Those equivalence ratio PIDs are also excellent for fuel control issues. For reference, your Mazda is likely to use a sensor electrically similar to the Nissan five wire.

Just curious, what scanner are you using? Always on the lookout for affordable scanners.

As for the roughness and misfire, it might be good to start with a power balance test, just to see if one cylinder is contributing less than the others?

Thanks Tyler.

I'm using my laptop, with a bit of software called TouchScan, with the "professional" and "Mazda" addons. It's ok for a pretty basic tool. The Mazda addon gives more information, but some stuff is lacking. It's pretty easy to log information to look at later, but just looking at data right now while you're using it can be awkward. There's no real way to just pull up the data for a bunch of pids, not that I've found anyway.

The Wideband and O2 sensors seem to be working ok. I created a vacuum leak, and the Wideband picked it up straight away, so I don't think there's a problem with that (unless it's just reporting the wrong values, but I'm not sure how to test for that)

I tried doing a cylinder balance test, but I don't have a vacuum gauge so I was just going off engine RPM, which the computer adjusts anyway.

This graph show 5 times me unplugging the injectors (cyl 4-1, then cyl 4 again for some reason) then unplugging the coils (again cyl 4 through to 1)

The Wideband picks up the different changes ok I think, no one cylinder appears to be contributing less than any others.

So I'm not sure where to go now. I appear to be chasing -ve fuel trims that are pretty -ve but not enough to worry some people, but the lack of smoothness is still there. It's not really rough, and most people don't even notice it, but I do.

I like the graphs! :ohmy: :woohoo: Great results from the power balance test. Particularly cool because it shows a clear trim difference between a no fuel misfire and a no ignition miss. Goes along with one of Paul's earlier videos:



Another thing that occurs to me is engine mounts. If they're worn/torn/leaking, then you can end up with something that feels like a roughness at idle. You can try gently using a floor jack and a piece of wood on the oil pan to take load off the mounts, and see if the roughness improves.

Last idea I have is a timing problem. If your 2.0L is anything like the one we have Stateside, then it's a chain system that has no keyways for the chain sprockets. This means you can end up with mechanical timing issues that are difficult to diagnose without a physical inspection. Has the engine been apart recently?

Ok. So this is a bizarre random fix.

The problem of rough running turned out to be ....

(Drum roll please)

A warped wheel hub.

Yep. The roughness was getting worse, but started getting other things as well, such as pulsing on the brakes. Eventually took a dial indicator to the wheel hub (after checking the disc) and found about 1mm of movement when spinning the wheel. Bearing felt fine, no roughness or play.

So all the "roughness" that could be felt while driving the car was due to a wobbly wheel.

Changed the hub (and bearing) at the weekend, and it's a lovely drive again.

Hey, nice to see an update with a fix! Even if it wasn't an engine performance issue. :lol:

Good eye catching the warped hub. I've only ever seen that once, and it was after an ape I used to work with used a torch to cut the outer race off a hub. Customer came right back, and I used a dial indicator the same way you did to find it.

Anyway, glad to hear it's all good.