O2 Transport Delay
There is a short period of time between when combustion occurs in the chamber, and when the O2 sensor measures the oxygen content in the exhaust. This short delay is called the O2 transport delay. There is a table in the tune that describes how long this delay is, and it varies based mainly on RPM. The faster the engine is spinning, the faster these exhaust pulses go down the exhaust pipe and the shorter the O2 transport delay is.
As the O2 sensor wears out naturally over time, it is less sensitive and takes longer to react to the combustion event. And so the O2 transport delay increases with O2 sensor age. Because of this, the PCM has to learn what the new transport delay is. So the table in the PCM describes the original transport delay, and the PCM starts with that. But over time, it learns what the delay is based on how well the O2 feedback controls the air/fuel ratio. And so there is a table in the Keep Alive Memory that's identical to the table in the calibration. And that KAM table starts out identical to the calibration table, but over time that table increases to match the wear of the O2 sensor, and it now reflects the real amount of time the O2 sensor delay is.
When you relocate the primary O2 sensors further back due to long tube headers, this O2 transport delay must increase. How much? We don't really know exactly. But the further back it is, the bigger our first guess will be. Perhaps 30% at idle speeds. Anyway, if the PCM learns this over time anyway, why do we need to tune for it?
Because the PCM learns slowly. And in the mean time, we want it to be closer to accurate. And we don't want to set a code for slow response. And so we jack up that table using our best guess. That'll do, for now. The PCM learns the rest anyway.
In a perfect world, the custom would drive that car for a few months, without ever clearing KAM or having a dead or disconnected battery, so that table can learn real good. And then the tuner could read the leaned values from this KAM table, and copy and paste them into the actual calibration table. That way, the car immediately knows the right answer, even if you disconnect the battery. This isn't really necessary, but it is ideal.
That never happens in the real world. But a sharp tuner might eyeball a few of these types of tables over time, and he will get a better feel for what those numbers look like with various types of headers. And his best guess will be that much better next time.
There are plenty of other "trims" in the computer which could also be learned from over time. Returnless fuel system cars in particular would benefit from learning these real world values and then punching them into the tune.
I don't believe the car would really have a major air/fuel ratio problem or run lean because of this table. It might be a little off, but it learns quick enough to not really be an issue. Things like this, and fuel trims, are why a car will end up with better gas mileage after you put some miles on it (after tuning or battery disconnect).