Another Porsche Engine Failure

I finally got to Road America after years of anticipation, but I came home with a blown engine. It was the annual TRAC event hosted by Chicago Region, where I met a lot of nice people who were very helpful and welcoming to this visiting Canadian. I had six good stints on track over the first two days until the trouble began.

My engine is a built 3.2L with PMO carburetors, based on a 1979 case, with 120/104 cams, MSD ignition and a 9.5:1 compression ratio (CR). Oil appeared on the fibreglass engine cover on the left side and I called my mechanic to discuss the possible reasons. The engine seemed to be running fine otherwise. Oil pressure was strong; temperature never exceeded 120 C; the Innovate Air Fuel Ratio (AFR) gauge which monitors both headers never exceeded 13.5. We concluded that it might be the low oil pressure switch that was leaking up the centre. So I made a diaper out of paper towels and duct tape and did another stint. There was much less oil this time, so I felt that we had correctly identified the cause. In my last session, I lost power dramatically going into Turn 5 and pulled off the track. The only smoke was coming from a small amount of oil dripping onto the left header.

After the 1500 km drive home with truck and trailer, I took the car to the mechanic for diagnosis. We got the engine to fire, but it ran really roughly. Here’s what we found:

The #1 spark plug had a broken ceramic core and the electrode was half burned off. The #2 plug looked normal. The #3 plug had a broken ceramic core. The plugs on the right side were fine.

Compression in #1 was 25 psi, in #2 we had 105 psi and in #3 it was zero.

This engine was rebuilt twice last year after melting pistons and breaking rings, which we attributed to a worn/dirty/defective CIS system that caused an excessive lean condition. Hence the switch to carburetors and the installation of the AFR gauge. When it failed this weekend, it had 35 hours of track use since going back into the car in April. I have never missed a shift or exceeded 6500 rpm.

A few days later, the mechanic had removed the engine from the car and proceeded to dismantle the cam towers, remove the cam shafts and remove the cylinder heads. Here’s what we found.

  1. Cylinders 4, 5 and 6
    1. All spark plugs look normal
    2. All valves are very light grey
    3. All combustion chambers and piston tops are lightly coated with carbon deposits, which is very dry and baked on
    4. All rings are intact
    5. Deck height is between 1.27-1.35 mm
    6. Cylinders 1, 2 and 3
      1. All valves are dark grey or chocolate, but the colouring is a bit oily
      2. All combustion chambers and piston tops are lightly coated with carbon deposits, which is a bit shiny
      3. Deck height is between 1.23-1.48 mm
      4. Cylinder #1
        1. Spark plug cathode burned off by 50%
        2. Spark plug anode ceramic missing to a depth of 7-8 mm
        3. Core of spark plug (metal) is hollowed out to a further depth of 2-3 mm
        4. Remains of core ceramic is heavily crystallized and shows a blue colour
        5. All rings are intact
        6. No apparent damage to piston or cylinder wall
        7. Cylinder #3
          1. Spark plug anode ceramic is broken about 4 mm from the end but still held in place by the central wire
          2. Spark plug cathode is intact but shows signs of burning at the tip
          3. Hole in cylinder wall adjacent to cylinder #2. Hole is teardrop shaped; round end (about ¼” diameter) is centred on the edge of the deposit which marks the end of the travel of the topmost ring. Teardrop extends towards crankshaft.
          4. Erosion of circumference of combustion chamber adjacent to hole in cylinder
          5. Rings are fused to piston adjacent to hole in cylinder; gouging of piston top
          6. Ring end gaps are not close to the hole
          7. Spark plug cathode tip is aimed towards hole in cylinder when fully screwed in
          8. There are tiny metal deposits on the top of the piston, likely coming from the melted cylinder wall
          9. Cylinder #2
            1. Spark plug looks normal
            2. Top ring is broken in one place
            3. Damage (melting) to outside of cylinder adjacent to hole in cylinder #1
            4. Ignition: No damage to inside of distributor cap; plug wires (Magnecor) to cylinders 1-3 have similar resistance levels.

While the carburetor jetting is undoubtedly too lean, based on all available advice and literature, it is unlikely to have caused the above damage.

I have subsequently spoken to numerous acknowledged experts in air-cooled Porsche engine construction and racing, namely:

  • A senior member of The Racers Group, who “grew up” with air-cooled Porsches
  • A former senior mechanic with the Farnbacher-Loles race team, who now has his own Porsche shop and has 30 years experience with air-cooled engines
  • The top air-cooled motorcycle and Porsche engine builder in the Chicago area, whom I had met at Road America
  • An experienced Porsche specialist from the West coast who has a very successful Porsche specialty business
  • A representative of a Porsche racing shop in Florida which specializes in air-cooled racing engines
  • The owner of a Porsche performance shop in Canada with over twenty years experience in building air-cooled engines
  • The Canadian technical support people for NGK spark plugs
  • The Porsche Club of America air-cooled technical advisor
  • The widely acknowledged global expert in air-cooled Porsche engines, located in California, who has written a book on the subject and is revered in Porsche circles

The opinions of all of these people are very similar, as outlined below, although none of them can explain why one bank of cylinders failed and not the other. And each of them has his own opinion as to what triggered the failure, with very little congruence among them.

  1. Having set the carburetor jetting to achieve an AFR of 13.2 at 6000 rpm, we were asking for trouble with an air-cooled engine that would be used almost exclusively on the track. It should have been in the range of 12.5 to 12.9 at wide open throttle and 6000 rpm.
  2. I should have been using racing fuel to minimize the probability of detonation – at least 96 octane if not 100.
  3. I should have been using racing spark plugs with a colder heat range for the same reason – at least heat range 7 if not 9.
  4. The timing was too far advanced at 35 degrees BTDC for exclusive track use – more like 29-30 degrees, to be optimized on a dynamometer.
  5. Perhaps the deck height was too large, which at least one specialist believes can stimulate detonation.
  6. We should carefully examine the cooling vanes to ensure unrestricted air flow behind the alternator.
  7. We should test and possibly upgrade the fuelling system to ensure clear lines – both fuel and tank venting – and sufficient flow and pressure.

There doesn’t appear to be anything wrong with the way the engine was built; rather, the problem lies in variable tuning parameters as outlined above. The last expert listed above wants to measure the cam shafts to see whether the back side of the lobes is shaped properly to avoid improper opening and/or closing of the valves. We have sent the cams to him for this purpose.

Since the AFR sensors measure everything that’s “seen” in the two headers, there is a chance that the observed readings were misleading. Once a plug had failed, that cylinder would be rich. But if the other cylinder(s) were lean, the sum of the three might still be a normal reading. The left bank of cylinders was probably running too lean until the plugs failed, so there should be a very light grey deposit on those valves. But it might have been covered by a sooty coloured deposit once the plugs failed.

The fundamental mystery is

a)      Why did the left side fail and not the right side?

b)      What was the primary cause of the failure?

No one has been able to answer these questions unequivocally. My current theory is that all tuning parameters were at the limit and Road America put such a strain on the system that something had to give. In this case, I suspect it was the two spark plugs, with others soon to follow, as evidenced by the beginning of crystallization on their ceramic elements. I wouldn’t be surprised to find that plug #2 is failing and would misfire if tested.

Leave a Reply

You must be logged in to post a comment.