This problem of intake valve seats dropping has struck quite a few owners over the past few years. We first became aware of it with track cars back around 2014. Shortly after, we started seeing seats actually fall out in our ultrasonic parts washer (heated to just 165*) and our casting oven (heated to 210*) during valve guide installs. Lucky for all involved it happened in the shop and not on the road or track. As you can imagine the results can be catastrophic to the engine and the wallet.
As a result of these issues, three years ago we began compulsory replacement of all 3.4l 987 (not to be confused with the early 3.4 996) and 3.6l 996/997 intake valve seats when we service these heads, which are virtually identical. We have not heard of widespread issues with exhaust seats dropping, but we have seen a few drop in our casting oven, too. For this reason we recommend replacement of exhaust seats, but since failures in service are extremely rare we have not made that compulsory, yet.
We recently had several intake seats drop in the shop on a pair of 3.4 987 heads. We’ve always taken measurements when this happened in the past, but we decided this time to snap some pics and share the results. Here’s what we found: Interference fits of a mere .003″. We’re not willing to share our specs for this operation, but I will say that it’s substantially greater than .003″.
In the past we have measured interference fits as small as .0015″, making that head a ticking time-bomb.
We’re pretty sure these heads did not leave the factory with seats this loose. They wouldn’t have made it this long. It’s my opinion that these seats are loosening over time as a result of a perfect storm of valve-train conditions that leads to major valve bounce on the seats. This bounce issue is compounded by the severe valve train operation of the aggressive intake lobe profile of the VarioCamPlus system which, by itself, is quite capable of inducing valve bounce. The compounding comes from the aggressive valve action inducing strong wave actions in the dual non-interference parallel valve springs. This wave action transfers through the retainer into the valve leading to lateral and spiralling valve motion during valve events, creating the severe intake guide wear so common with these heads. This guide wear also promotes valve bounce as the valve lands out of square with the seat and settles spirally.
Valve bounce is a serious problem, but accompanied with worn valve guides it can be quite destructive and is normally very hard on valve seats, leading to valve seat wear in older engines. In the case of modern powderd metal valve seats, as in these heads, wear is not an issue. These seats flat do not wear. They may break, like the seat in the above picture, but they do not wear. All the shock of harsh landings related to the aggressive cam profile and valve bounce and spiral settling is transferred through the extremely hard valve seats into the aluminum. This shock relaxes the bore, leading to the loss of interference fit. The rest of the story is told in failed parts.
We machine new seat counterbores for a more shock proof seat with excellent wear qualities. Here’s a pic of a finished counterbore.
We have never seen this issue of dropped valve seats with any other model of M96/97 cylinderheads, though we also insist on replacing the intake seats in 3.8l heads due to their similarity to the 3.6l heads.