Subaru flat 4 port and direct injection, non-turbo | Cylinder misfire

When is a misfire not a misfire?

As ever, vehicle history is all important and so an interview with the customer confirmed starting and driveability to be normal (hot or cold) with no loss of performance, but did reveal his vehicle had been seen by another garage where No. 2 cylinder Spark plug, Ignition coil, and Direct injector had already been replaced! The owner added that the warning light would only illuminate when the engine was at normal operating temperature at low road speed.

A basic visual inspection confirmed no issues with hoses, harness, connections, fuel quality, or fuel level. 

Looking at the freeze frame data accompanying the fault code P0302, it did indeed support the customer’s version of events, recording a hot engine, low road speed, and a misfire count of 44 on number 2 cylinder!

Freeze frame data below

In an attempt to reproduce the fault code whilst looking at live data in our workshop, the vehicle actually demonstrated the misfire count seen above! However, the engine idle quality did not appear to suffer other than the characteristic idle of this engine type at 650rpm. How nice it is to have a symptom that wants to reveal itself on demand!

As this was too good an opportunity to miss a 4 gas analyser test was carried out whilst monitoring the live engine data and ensuring the misfire count was still evident. Our results were most strange as all emission readings were correct with no errors both at idle and 2500 rpm. HC 12PPM, CO 0.005%, CO2 14.44% and O2 0.4% at idle with the live data list recording 20 misfire counts on number 2 cylinder! One point to note here was the misfire count fell to zero on number 2 cylinder as soon as the engine speed was increased just 150rpm above idle (return to idle, the misfire count began to increase again).

Idle emissions with a misfire count of 20+

Live data at idle indicating misfire No2 Cylinder

Vehicle Speed	0	MPH
Engine Speed	652	rpm
Calculate Load	27.8	%
MAF	2.41	gm/sec
Coolant Temp	89	C
Cylinder #2 Misfire Count	20
Throttle Sensor Position	1.9	%

At this stage, the dilemma is a misfire code and count that continually returns for No.2 Cylinder only when warm, no misfire evident, misfire counts falls to zero above idle speed, and GOOD emissions! Looking deeper into the data list for any further evidence of a misfire nothing appeared to stand out, O2 signal values, fuel Trim, high fuel pressure,  MAP, MAF, and pending codes all returned OK which leads me away from a possible fuelling or ignition issue.

Here then we appear to have a misfire only the PCM can detect yet has no effect on the engine running condition. Knowing fault code P0302 indicates cylinder misfire we need to know how the PCM calculates this information to conclude a misfire is present. This is calculated by the PCM monitoring variations in crankshaft rotational speed and linking the variations to the camshaft position sensors in order to identify the offending cylinder.

To throw a spanner in the works we could not be 100% sure which cylinder was in fact No.2! Our Technical data claimed No. 2 Cylinder was at the RH rear whilst our wiring diagram indicated LH front and so how could we prove which cylinder the PCM controlled as No.2? My mind was racing ahead here thinking maybe the previous garage had installed one of the correct parts but to the wrong cylinder!
Using the ‘cylinder cut’ feature of the Scan Tool and placing our Coil-on-Plug (COP) probe on top of both the direct injector and Igniter we could monitor the signals as the PCM switched these components ON-OFF (cylinder cut). Here we discovered conclusive evidence that No 2 Cylinder was in fact the LH front cylinder and the cylinder that had received the new components, DAMN! 

The cylinder cut feature of the Scan tool also proved inconclusive as each “cylinder cut” returned an equal fall in engine speed. PicoDiagnostics was used to carry out a non-intrusive compression test. At last our first breakthrough to prove something was not right! There is a saying that goes K.I.S.S (Keep. It. Simple. Stupid) and how true is that? One might argue why was this test not carried out at the start, and I would have to agree, but we had no misfire symptom other than a code and data list count.

NO2 direct injector non-intrusive test

NO2 igniter non-intrusive test

Here we have confirmation of a potential compression issue on one of the 4 cylinders. Please note the results appear to show a near 50% reduction in compression of one cylinder. This is a relative compression test where the 4 cylinders are compared against each other. My first thoughts were a cylinder that had a 50% reduction in compression would surely demonstrate more obvious symptoms, but we have to remember the engine speed here is only 200RPM (cranking speed). One of the symptoms highlighted a misfire count at low engine speed only, anything above idle and the misfire count fell to zero!

So we now had ammunition to go deeper regarding compression and of course the next step had to be “what cylinder is suffering from the loss of compression?”.

Here we opted for a cranking current test using PicoScope with the “Coil on Plug” (COP) probe placed on what we know confirmed as No. 2 Cylinder igniter assembly (no surprises here, No2 cylinder has loss of compression). Again another non-intrusive and simple test to provide evidence that will allow us to proceed with confidence and form the basis for a portfolio to present to the customer as this was starting to look very expensive!

Compression Test

The above waveforms provide the conclusive evidence that not only is there a loss of compression whilst cranking but we have now discovered No.2 Cylinder is the offender.

Knowing we have a loss of compression we had to discover where before we dive into compression testing and dismantling the engine. Here two WPS500 pressure transducers were used, 1 installed into the exhaust and 1 into the inlet manifold with the COP probe placed over No.2 igniter in order to relate the waveforms to a firing order. With the engine idling this revealed an area of real concern and eliminated another!  The pulsations measured via the exhaust confirmed no excessive peaks so for now at least could be overlooked.  However the intake pulsations via the intake manifold revealed an area of concern for inspection when dismantling the engine (see below).

Subaru Waveform 1

Every 4th intake manifold pulsation peak to peak proved to be low, which again linked to No.2 cylinder. By aligning the No. 2 cylinder firing event we are able to locate the offending intake pulsation as the suspect intake pulsation takes place before the ignition event on No.2 cylinder. A further test was carried out again using two WPS500 pressure transducers this time via the intake manifold and the dipstick tube. The dipstick tube pulsations were all even so confirming no issues with excessive crankcase pressure hence piston and ring efficiency appears fine.

Subaru Waveform 2

To recap, it is worth mentioning here that all the above tests have been non-intrusive other than removing an intake manifold hose, yet we have been able to deduce we have a compression issue on No.2 cylinder that appears to be related to the intake stroke. Equally we have also been able to confirm areas that are not of concern, for example Fuelling, H.T, piston to cylinder sealing, and exhaust valve sealing efficiencies.

Now we have the confidence and more importantly the evidence to remove the LH Spark plugs which on this vehicle is no mean feat at over an hour labour, highlighting the value of non-intrusive testing.

The WPS500 pressure transducer was then installed into the LH rear (NSR) cylinder and the waveform acquired by cranking and running the engine, saved and kept as a reference to compare against the LH front No.2 cylinder (NSF) under the same conditions.

Subaru Waveform 3

Here we can see the reduced peak compression pressure of the NSF cylinder (No.2 Dark blue) in comparison to the under laid reference waveform of the NSR cylinder (Light blue). No.2 Cylinder peak compression achieved 9.77 bar compared to the NSR at 12.75 bar, a difference of 2.98 bar as indicated by our relative compression test earlier. Using the phase markers of the PicoScope Beta software we were able to mark out the Zero (TDC) and 720 degree (TDC) points of crankshaft rotation and then divide the distance between the phase markers equally using the phase marker partitions, enabling us to plot the events of the 4 stroke cycle.

What became very apparent was the lack of “Ripple/Turbulence” during the intake stroke of No.2 cylinder in comparison to the reference waveform.

Subaru Waveform 4

We always have to look for differences or anomalies between known good signals and here we can see good turbulence from both cylinders during the exhaust stroke indicating good airflow dynamics but a definite reduction in turbulence of the air flow during No.2 cylinder intake stroke. We can also identify the early and prolonged opening of No.2 cylinder inlet valve in comparison to the reference waveform and the reduction in peak vacuum, all linked to the condition revealed below.

The above engine varies the valve timing according to engine load conditions. During cranking and idling, valve overlap is eliminated to improve idle quality.

At this stage enough is enough as no further testing can be carried out with the engine installed. Sufficient evidence had been gathered and presented to the customer confirming we were now looking at a fault of a mechanical nature. Permission was then given to remove and dismantle the engine that finally revealed near zero valve clearance with No. 2 Cylinder Inlet valves at 0.03 mm and 0.00 mm. This confirmed the misfire that proved evident only when warm and why the misfire would be more pronounced at low engine speeds as the valve would have been seeping. As engine speed increased both the momentum and increase in pressure would disguise the misfire (no misfire count at 150rpm above idle but near 50% misfire during relative compression test at 200 rpm). Looking back at the testing carried out the inlet valve could only have been marginally miss-seated as no misfire could be detected using cylinder cut or emission methods, yet the PCM had identified an anomaly with crankshaft rotational speed as did the relative compression test. The engine has since been reassembled with an inlet valve clearance of 0.12 mm and all is well with no misfire count or code apparent.

Subaru Waveform 5