Rod bearing wear: what you can hear, and what you cannot

Some of the best engines of the Analogue era carry a quiet worry. Owners of the BMW S54, S65 and S85 talk about rod bearings the way Porsche 986 and 996 owners talk about the intermediate shaft bearing: a part you cannot see, that may be fine for another 100,000 miles, or may take the whole engine with it next weekend. I want to set out what is actually known, what you can hear, what you cannot, and where an oil sample and a phone fit in.

Why these engines have the reputation

The BMW S engines. The S54 went into the E46 M3 and the E85/E86 Z4 M, the S65 V8 into the E90/E92/E93 M3, and the S85 V10 into the E60/E61 M5 and E63/E64 M6 (Wikipedia lists the applications for each). Wikipedia's S85 article calls premature rod bearing wear a well-known problem and puts it at around 80,000 km, roughly 50,000 miles, linking it to high revs and tight clearances. BIMMERtips names all three engines as the ones most prone to it and describes a bearing clearance of roughly a thousandth of an inch.

James Clay of BimmerWorld, interviewed by Grassroots Motorsports in December 2024, puts it plainly: the S54, S65 and S85 all need rod bearing attention as routine maintenance, recall or not, since early S54 cars were split by a mid-run BMW recall. In his view most of the wear happens at startup and warm-up, and engines that routinely rev past 7,000 rpm are the ones most exposed. That is one experienced shop's view, not a factory statement, but it matches what the other sources describe.

The Porsche M96 and M97. Here the worry is a different bearing. The IMS Solution's IMS 101 page explains that these engines use an intermediate shaft running on a ball bearing, and calls failure of that bearing the most common deficiency of the M96 and its revisions through 2008. It gives the generations: a dual row bearing from 1997 to 1999, a smaller single row from 2000 (in every engine by 2002), and a larger single row from 2006 to 2008 that cannot be serviced without taking the engine apart. It reports a failure rate of about 8 percent for the small single row, taken from the Eisen class action settlement documents, against under 1 percent for the dual row. When it fails, the debris goes wherever the oil goes and the engine usually needs a full rebuild.

What a worn rod bearing sounds like, and why it is quiet early

A rod bearing is a plain bearing. It does not roll; the crank journal rides on a thin film of pressurised oil. The background section of a Hyundai and Kia patent on detecting bearing damage describes it well: while that oil film holds, noise and vibration stay small. The knock appears when wear opens the clearance far enough that the rod actually strikes the crank pin.

That is why early wear is quiet. A bearing that has lost some of its overlay but still holds a film makes no new sound. By the time there is a knock, metal is touching metal. James Clay says he has heard street engines make rod noise before they damage the crank, but calls it a small window, and says that on track you likely get no warning at all.

What does it sound like when it does arrive? The pattern my app listens for is a heavy, impulsive knock whose rhythm is locked to the crankshaft, one beat every two revolutions, so it speeds up as the revs rise. In practice other noises can pass for it. My own E39 M54 has a tick I can hear, and by ear I traced it to the valve cover. It was worse warm and went away off idle, which did not fit a bearing, so I read it as a top-end tick.

The IMS bearing is a ball bearing, so in theory it makes its own frequencies. In practice the IMS 101 page says the small single row can fail with little warning, and it does not list noise as a reliable symptom. Its advice is to cut open the oil filter and check the drain plug magnet for ferrous debris, which it likens to silver glitter.

What oil analysis shows first

For rod bearings, oil analysis is the earliest practical warning most owners can get. Blackstone Laboratories lists lead, copper and tin as metals that come from bearings. Blackstone has published a used oil report from a 2008 M3 with the S65 and about 55,500 miles on the engine: lead read 78 ppm against a universal average of 7 for that engine, with iron, aluminum and copper also above average. The lab's comment was careful even then, saying the engine may have a bearing problem but that it is hard to be sure from a first sample.

Two cautions come straight from Blackstone's own guidance:

• One number is not a diagnosis. It takes about three samples to get a good trend going, and a sudden change against your own history matters more than a single high value.
• Lead has other sources. Blackstone says the highest lead it has seen in any S65 was 1,055 ppm, and the rest of the metals looked fine; the owner had used an additive. Leaded fuel and some octane boosters also raise lead.

Oil analysis has limits too. Noria's guide to iron in oil notes that the spectrometer used for wear metals typically cannot see particles larger than about 5 to 8 microns. Large debris from a failing part can be missed, which is one reason the IMS crowd inspect the filter as well.

What an acoustic scan on a phone can and cannot do

I built Sonic Pulse, the listening part of The Analogue, so I will be exact about where it stands.

What is real today. The recording is real microphone audio, and the frequency analysis is real. For the S54, S65 and S85 there is a rod bearing check, and for the M96/M97 an IMS check. Both need a held rev, which the guided test asks for at 3,000 rpm. At idle the app reports the bearing as not assessable rather than giving you a reassuring zero, because the patterns it needs are too low to resolve at idle speed. The IMS check will not assess anything below 1,500 rpm.

What is not proven. The scoring is rule-based physics I wrote, not a model trained on engines with known outcomes. As I write this, owners have shared six captures with me for training, none of them labelled and none from an engine with a confirmed worn bearing. The thresholds have never been checked against a car that was opened up afterwards. The IMS check is marked early testing in the code and is not allowed to report anything stronger than an observation.

What I have already got wrong. Twice the matcher could have mistaken a healthy engine's normal firing rhythm for a bearing fault: once on the BMW six, and once on the Porsche flat six, where the search window sat right on the firing order of every healthy engine. I found and fixed both, and the Porsche fix now has a test built around a healthy flat six that fails if the old mistake comes back. And on my own E39, the detector scored a tick I can plainly hear as low. Right now it is less sensitive than my ear, not more.

So an acoustic scan is a screening aid that is still being trained. It can tell you a recording looks unusual and is worth following up. It cannot tell you a bearing is worn, and it certainly cannot tell you one is fine.

What I would actually do

1. Start oil analysis now and sample every change, so you build your own trend before you need it.
2. Watch lead and copper against your own history and the lab's averages, and read the lab's comment, not just the numbers.
3. Find out whether the bearings have ever been done. If not, and the miles are high, talk to a specialist about inspection.
4. Use a recording as a nudge to look closer, never as the reason to relax.