A most intriguing cold-start misfire mystery

Question

I'm trying to help a friend out with a CEL on his 2009 Merc GLK 280.

Hooked it up to an OBD2 reader and retrieved a single code:

P0300 - 'Random/Multiple Cylinder Misfire Detected', via the 'OBD Fusion' app on my iPad.

Here is the P0300 freeze frame:

+-----------------------------------------+-------------+
| Fuel System 1 status                    |           1 |
| Fuel System 2 status                    |           1 |
| Fuel System 1 status                    |           1 |
| Fuel System 2 status                    |           1 |
| Calculated load value                   |      23.14% |
| Engine coolant temperature              |        49 C |
| Short term fuel % trim - Bank 1         |          0% |
| Long term fuel % trim - Bank 1          |      11.72% |
| Short term fuel % trim - Bank 2         |          0% |
| Long term fuel % trim - Bank 2          |       7.03% |
| Intake manifold absolute pressure       |      28 kPa |
| Engine RPM                              | 1293.75 RPM |
| Vehicle speed                           |     17 km/h |
| Ignition timing advance for #1 cylinder |    42.5 deg |
| Intake air temperature                  |        46 C |
| Mass air flow rate                      |    5.63 g/s |
| Absolute throttle position              |      13.33% |
| Time since engine start                 |      26 sec |
| Fuel rail pressure                      |     380 kPa |
| Commanded evaporative purge             |          0% |
| Fuel level input                        |      45.49% |
| Barometric pressure                     |      99 kPa |
| Control module voltage                  |     13.29 V |
| Absolute load value                     |      17.25% |
| Fuel/Air commanded equivalence ratio    |        1.54 |
| Relative throttle position              |       1.96% |
| Ambient air temperature                 |        36 C |
| Absolute throttle position B            |      12.94% |
| Accelerator pedal postion D             |       6.27% |
| Accelerator pedal postion E             |       6.27% |
| Commanded throttle actuator control     |       2.75% |
+-----------------------------------------+-------------+

---

I collected further information via the scan tool:

• Lambda sensors

  The front sensors on both banks return 1.00 V. which indicates that they're toast (assuming they're narrowbands, normal output range should be 0.2 - 0.8 V).

  The rear sensors read around 0.63 V.

  Both readings taken with the engine warm and at idle. Blipping the throttle registers a change in the voltage output of the rears only; the output of the fronts doesn't change.

• Fuel trims

  Long-term fuel trims are +7% for one bank, +11% for the other.

  Short-term fuel trims are 0% at hot idle.

• Mass Air Flow, Fuel Flow

  At hot idle:

  Mass air flow = 3.66 g/s Fuel flow = 1.27 l/h, Fuel rail pressure = 380 kPa

  This indicates to me that AFR is around 14.88, close to stoich.

---

Questions

I'd like to say based off my calculation that the front lambdas are to blame, but a couple of things are puzzling me:

• If the front lambda sensor voltage is indicating rich, why is the fuel trim positive (indicating that the injector pulse-width is increased because the AFR is deemed too lean by the sensor)?

• The misfire happens after a cold start with minimal throttle (note the time since engine start in the freeze frame). I'm not sure I completely understand the interaction here with the faulty lambda sensor, since cold start enrichment is playing a role here. If the system is running slightly richer than stoich, it would it not be less of a reason for the falsely-sensed rich condition to cause a misfire?

• The owner is complaining of poor fuel economy as well. If the front lambda is sensing rich, I would expect the AFR to be leaner than reality, so I would expect fuel economy to be slightly better.

Answer 1

Root Cause: probably a couple of things

Sadly, I feel that there isn't enough data here to uncover the smoking gun, but I have realized some things upon reflection that can indicate what is probably wrong with the vehicle:

Question 1

If the front lambda sensor voltage is indicating rich, why is the fuel trim positive (indicating that the injector pulse-width is increased because the AFR is deemed too lean by the sensor)?

A couple of possibilities spring to mind:

• Fuel management has stopped trusting the front lambdas and stays in open-loop mode

  However, this alone will not explain why the engine misfires only when cold, since the engine should be running in cold-start enrichment mode to avoid that scenario.

  The one thing the freeze frame is sorely lacking is fuel flow, which would confirm whether or not cold-start enrichment is taking place.

• The fuel management is defaulting to a failsafe map

  If this was the case though, why would it misfire under light throttle? I highly doubt that Mercedes would release a fuel-mapping which would induce misfires, especially one that is designed to protect the engine.

  This leads me to believe that there is another issue present, something like unmetered air or partially-clogged injectors which the fuel management cannot accommodate for due to the absence of lambda feedback.

---

Question 2

The misfire happens after a cold start with minimal throttle (note the time since engine start in the freeze frame). I'm not sure I completely understand the interaction here with the faulty lambda sensor, since cold start enrichment is playing a role here. If the system is running slightly richer than stoich, it would it not be less of a reason for the falsely-sensed rich condition to cause a misfire?

Unmetered air or insufficient fuel delivery + lack of lambda feedback would totally explain this. Cold-start enrichment is there for a reason; cold engines need the extra fuel injected to compensate for the relative lack of fuel vaporization.

The positive long-term fuel trims have me leaning in this direction as well. It could well be that the values shown by the tool are the values from a simulated signal or simply that last values that closed-loop feedback was operating off of.

---

Question 3

The owner is complaining of poor fuel economy as well. If the front lambda is sensing rich, I would expect the AFR to be leaner than reality, so I would expect fuel economy to be slightly better.

In all probability, the fuel management has disavowed the front lambda sensors, so it doesn't matter what they're reporting.

---

Here are my top suspects

Both:

• Unmetered air or partially-clogged fuel injectors
• Malfunctioning front lambda sensors

I can see how the first problem could lead to the second - if excess fuel is being dumped on the lambda probes via positive fuel trims, it could lead to fouling of the sensor.

---

Recommendations

As a first step, replace the lambda sensors.

This is to let the management go into closed-loop mode so that a clearer picture of AFR correction is obtainable.

This would then help quantify the extent of the AFR correction. If it is present, it may not be long before my friend needs another pair of front lambdas.

Plus, it's helpful to tackle one problem at a time.