majbthrd

I faced a B2290 fault code on a 2007 (GT package), and this documents what I had to do to make it go away (in case it helps others).

I believe this description applies to RX8 model year 2007 onwards (or possibly 2006 onwards... these seat weight sensors definitely didn't exist in the 2005 and earlier models).

B2290 does not cause a check engine light; rather, it causes a "16" error code (blink once, pause, then blink six times, pause, then always on) for the airbag light.

The fault is associated with what is interchangeably called the "Seat Weight Sensor" or the OCS ("Occupant Classification Sensor/System"). In theory, the system is supposed to gauge the passenger's weight in order to tailor the force of the airbag when it deploys.

RX8 vehicles with this system have two sensors (a left and a right) and a "seat weight sensor module". The sensors are inseparable from the seat rails and measure the minute deflection of the seat rail under the passenger's weight; the module is removable and is attached beneath the right seat rail.

In my vehicle, the left sensor went bad, but neither the system nor the factory service manual provides information on how to make this determination. My homegrown diagnostic procedure was to disconnect the car battery (waiting the requisite 10 minutes), and then disconnect each of the sensors from the module. (The connectors are polarized, so it should not be possible to plug them back in incorrectly.) Carefully inserting thin solid wire into the sensor's female connector to access the connector contacts, I measured the impedance with a multimeter. An impedance of roughly 190-200 kohm is "normal" in my experience. Lack of continuity, impedance in the Mohm range, or impedance much less than "normal" would all appear to be a sensor (or wiring) fault.

The factory manual suggests replacing the module when all else fails, but here's why this can be disastrous (at least for the DIY mechanic):

In my experience, when a module detects a bad (or disconnected) sensor, it logs an error condition in its non-volatile internal memory. FORscan (v2.3.55 at the time of writing) does NOT seem to have the ability to reverse this; it just craps out with a cryptic error message saying that it was interrupted. (I've tried contacting their tech support, and they insist that their software is not at issue and to suggest otherwise indicates an "attitude".)

I bricked (at least in a FORscan sense) a couple of modules before understanding what was going on.

Only when I replaced seat rails to have two sensors that I had already verified good with the above the multimeter test *AND* installed a module that had NEVER detected bad sensors did the FORscan "OCS" reset/calibration procedure work. Note that it requires access to one or two 44lb / 20kg weights to place in the passenger seat (which it only tells you about midway through the calibration); it did give the option to only calibrate with one 44lb / 20kg weight (but did not elaborate on the pros and cons of this).

(Aside: I suppose you could luck out that after doing a complete swap out of known-good components it might work without needing the OCS reset/calibration, but we're talking such sensitive sensors that anything can upset it.)

To make life even more complicated, it seems that Mazda had several different wiring harnesses for the passenger seat, even within the same year. I found a 2007 Sport and 2009 Sport in assorted scrapyards, but these did not have the same wiring harnesses as my 2007 GT. (The GT has an extra set of wires that go the seat back.) I resorted to creating a hybrid of multiple units to make a viable whole.

FYI: the part numbers for the seat rails (sensors are integrated) are:

right rail: (2007-2008?) FE87-67-WR0 or (2009+?) FE87-67-WR0 A
left rail: (2007-2008?) FE87-67-WL0 or (2009+?) FE87-67-WL0 A

These part numbers do not seem to be listed in any parts database that I can find, but they are clearly marked with factory stickers indicating this designation.