Skip to content
Shop
Explore

INDEPENDENT BMW ENGINE EXPERTISE

Side effect of TwinScroll

N43 / N53 · NOXEM · DIAGNOSTICS

Select a technical image to view it in detail.

In my first entries on the B58 engine, I mentioned that it feels different from the N53. The N53 evens out the work of its cylinders so well that you cannot feel the engine running at all. The B58 is different: at medium-high and high required torque, a vibration appears. I now have more information on the B58 and can explain it.

The vibration is felt in a fairly low RPM range, up to about 2500 to 3000 RPM, when the required torque is above 250 to 300 Nm. It is more noticeable in cars with a manual transmission, because it radiates into the gear selector. It is probably also stronger if the car has RunFlat tires.

Live data: low versus high torque

As usual, I’ll start with live data. This is cylinder performance at low torque (required torque around 90 Nm):

All cylinders perform very similarly, differing by only a few percent. The accelerator pedal is at 20% (red bold curve, Axis 3).

Legend:

In the next graph the accelerator pedal is pressed to 90% (Axis 3), which corresponds to a torque of 400 Nm:

Has something gone wrong? The performance (mechanical efficiency) of the cylinders is hugely different, and the difference grows with RPM. By 2500 to 3000 RPM it reaches at least 10 to 15%.

At higher RPM the situation gets even worse. Flywheel data are accurate and reliable up to 3000 RPM, and I have no reason to doubt them up to 3100 RPM. Above 3100 RPM the flywheel data are no longer precisely accurate, but what we see is still far harsher than it should be.

The engine has split into two halves:

  • Cylinders 1, 2 and 3 are significantly more active.
  • Cylinders 4, 5 and 6 are lazy, with reduced performance.

I should add two points right away:

  • This picture of cylinder performance is stable. The flywheel is adapted, and the adaptation procedure for the other engine systems has been performed several times.
  • The same behavior has been observed on other B58 engines.

Ruling out the injectors, Valvetronic and the DME

If this happened on only one engine, we could suspect that Bosch has not implemented injector adaptation for the high-load mode, or that it does not work. The injectors are split across two rail halves (cylinders 1 to 3 and cylinders 4 to 6), so they could have noticeably different flow rates. That is relatively unlikely, but it could happen if one mini-group of injectors came from one batch and the other from another. But the same problem on more than enough B58 engines? Then the injectors are not to blame, and neither is Bosch.

Before I name the culprit, here is why neither the DME developer nor a mechanical fault is responsible. In the range of ample required torque:

  • Valvetronic is deactivated, so the HVAs cannot cause problems, and all cylinders receive the same amount of air.
  • If the injectors of some cylinders (the less active ones) had a problem, the mixture there would be lean and we would see detonation in those cylinders. The DME would detect it, the driver and the development engineers would feel it, and the engine would not survive such punishment for long.

The mixture in all cylinders is evidently correct: the wideband probe reports the average mixture as stoichiometric (Lambda around 1.00).

In the image, Lambda is the blue bold curve (Axis 4).

The cause: the TwinScroll turbo

So what happened to the three cylinders that lost their energy? The answer is simple: the TwinScroll turbo technology is to blame.

These engines have one TwinScroll turbo unit shared by all cylinders, but as the name suggests, it is not a regular turbo. Each group of three cylinders drives the turbine rotor through its own exhaust gas channel. The turbine therefore works in pulse mode: its rotor is turned not evenly but with quick beats. Each group of three cylinders is effectively loaded as if it had a different turbo.

  • One “turbo”, with the smaller cross-section, is intended for low RPM. It provides outstanding torque at low RPM (on the B58, 450 Nm at full throttle is available already by 1300 RPM) and quick response.
  • The second “turbo”, with the larger cross-section, is more active at higher RPM and higher required torque.

Each cylinder group works on exhaust channels (turbo inlets) of different cross-sections, and the exhaust gases of each group hit different blades, at a different angle. The fact that both halves are mechanically connected by a common shaft does not remove the problem.

The TwinScroll turbo:

These two different “turbos” cause two problems:

  • Different exhaust gas dynamics. As a result, the filling of the cylinders (or of each group) with air, and therefore oxygen, differs. The amount of fuel supplied is accordingly also different, and has to be. The cylinders really do work with different power; the DME knows this and takes it into account.
  • Different average load, meaning the gas resistance that arises as the turbo spins up.

If I overcome my laziness, I will investigate how significant the air-filling problem is and how exactly the mechanical load affects the situation.

Why the engine vibrates

The firing order is 1/5/3/4/2/6, so by group the sequence is 1/2/1/2/1/2. The engine therefore runs in an “active / less active / active / less active” pattern. As you would expect, this produces a vibration whose frequency is exactly half the cylinder firing frequency. At 2000 RPM, for example, the vibration corresponds to 1000 RPM. The lower the frequency, the worse it is suppressed by the flywheel and the engine itself. You can read more about this here.

Unfortunately, this vibration is a side effect of TwinScroll turbo technology, and nothing can be done about it. BMW managed to widen the RPM range of maximum torque and shorten the engine’s response time, but the vibration comes with the package.

Is it critical? Many owners may never have noticed it or consider it normal. Someone may have gone to the dealer and asked them to find the problem. Some may even enjoy it, because it gives the feeling that the engine under the hood is alive. For me, when I first noticed it and the cause was unclear (one theory was incorrect DME behavior), it was uncomfortable. Now that I know the engine is performing correctly, the vibration reminds me that there is a 450 Nm engine under the hood. That is nothing terrible.

Consequences for tuning

This small nuance does change everything about “upgrading” these engines. Any kind of chip tuning that raises power, whether by changing the maps in the DME itself or by connecting other or external units such as the JB4, upsets the differences in mixture between the cylinder groups. Even if the DME keeps the average mixture correct, the mixture of each individual cylinder will be wrong. The consequences:

  • Wrong mixture means detonation and excessive temperature, which damage the piston group.
  • Wrong mixture also puts a large load on the catalytic converter, leading to overheating and destruction.
  • Cylinder imbalance means increased vibration, and therefore more load on the flywheel, gearbox and transmission, as well as the output shafts, differentials, and the engine and transmission mounts.

Several further unpleasant points should be kept in mind:

  • The difference between the cylinder groups (and possibly between individual cylinders) is accounted for in various multidimensional control maps. Chip tuners do not have access to these maps and cannot easily modify them. Creating and modifying them requires highly complex equipment, deep knowledge and considerable other resources, which chip tuners simply do not have.
  • These correction maps, which describe the differences between cylinder groups, cannot be made adaptive. In theory the DME could run tests of the chemical performance of the cylinders and so determine each cylinder’s mixture (ignoring the fact that the engine has a single Lambda probe shared by all cylinders). But such tests need a static driving mode lasting several tens of seconds, which is impossible in real driving. The only option is exact, fixed control maps.
  • Any interference in the control of these engines has critical consequences in steady-state (slowly changing) modes, and especially catastrophic ones in transient (rapidly changing) modes. During transients, the situation after chip tuning is uncontrollable. A few seconds of incorrect mixture will probably not melt the catalytic converter, but it can damage it. One such overload cycle probably has no consequences, but what about 100? Or 1000? The same applies to a lean mixture in any cylinder: one spurt, no consequences; two, no consequences; but 100?

Of course, it is very tempting to cheat the MAF and MAP sensor readings a little, indicating slightly lower airflow and boost pressure in the high-torque mode, and gain some power that way. I believe a slight torque increase (up to +10%) is possible without error entries in the DME fault memory: the DME will re-adapt its mixture adaptation maps, and the formulas of the energy module will more or less match. Unfortunately, the cylinder imbalance described above is neither compensated nor controlled in any way. The consequences are what I already described: overloaded catalytic converters and damaged or even destroyed pistons. In the best case, the real torque gain shrinks because of retarded ignition, since the DME has noticed detonation. Thank you, but I don’t need such an upgrade, even for free!

BEYOND THE FAULT CODE

A clearer picture.
A better next step.

Get in touch
Enlarged technical image
Open original image