Since I picked up our long-term X3 from the South Carolina factory last April, I've noticed an annoying delay in its acceleration. Everyone on staff who's driven it has too, echoing complaints that the X3 accelerates either one of two ways: too slowly or with your head jerking back. We've also seen comments on numerous BMW forums saying the same thing, all stemming from new BMWs equipped with the new eight-speed automatic transmission.
I believe it's a result of programming, not the transmission itself. Even so, a trip to the dealership resulted in an unsatisfactory "operating as designed" prognosis and no change in its behavior. This month, we strapped our test gear to the X3, and took a closer look.
Findings: During low-speed acceleration from rest, like on a stop-sign punctuated residential street, the X3 exhibits repeatable throttle delay and very non-linear acceleration. Applying a normal level of throttle results in essentially zero forward movement for over half a second. Then, it feels like a gear engages, and then normal acceleration begins. While we expect some delay in throttle application (and miss throttle cables dearly), the X3's is noticeably longer and it makes driving smoothly difficult.
How we tested: At our test track, we strapped in a Racepak data recorder that uses accelerometers and GPS information. It is a system primarily designed for drag racing classes like Pro Mod, Pro Stock, and Super Stock; essentially overkill for this test. We then restricted the movement of the gas pedal with tape so the pedal could be treated like an on/off switch, but would be limited to the low-speed acceleration one would experience on a residential street. Doing so eliminated any wavering a foot might do on the throttle; when it's on, it's on completely. We then wired a trigger that would mark the data whenever the throttle was applied. We then simulated stop-and-go residential street driving, doing both complete stops and "California" stops.
Results: The graph above shows one near stop and start. Time is on the X-axis, mph (orange) is on the left Y-axis, and longitudinal (acceleration) g is on the right Y-axis, in green. This shows a "California" stop where the X3 decelerates to 0.6 mph (longitudinal g jumps to 0 when the brake pedal is lifted). At this point, the throttle is applied to its set point -- seen by the white spike -- and the X3 begins to slowly accelerate. After 0.6 second, the acceleration rate increases substantially, evident by the change in mph and longitudinal g, and then normal acceleration begins. Keep in mind that throughout this change in acceleration, the gas pedal input hasn't changed.
This half-second delay phenomenon was evident throughout our testing. In the five-run sample above, there is a noticeable change in acceleration behavior despite an unchanging pedal input. In 10 tests, we saw an average 0.5-second delay from the time that the throttle was applied to the time normal acceleration began, counting the time from the trigger mark to the jump in longitudinal g.
We believe this half-second delay is the culprit of all the complaints. But now that we've found it, we have more questions: Why is it there? One theory suggests that BMW might be trying to curb quick, "jackrabbit" starts to save
fuel economy, but that's just a theory. We also want to know what's causing it. But, most crucially, can we get rid of it?