Just got the new motor in and I've run into three issues.

First, code 05 check engine light for the knock sensor. Replaced the sensor and no help. Next I'm going to run a new shielded wire from the ecu to the sensor, but does it need to be a shielded wire and where can I find shielded wire to use?

Second issue, flashing seat belt light. I've found CPU#2 usually causes this which I believe to be the small black box near CPU? I can't believe it would go bad from sitting in my garage. Any other ideas?

Last is my odo was working before, but now won't come on. It shows mileage with key turned on but once motor is running it stays off. I know the clusters are known issues, but now I randomly get a add coolant light and buzzer although I've got coolant. The sensor is also wired properly ( black to black single wire). Weird part is as soon as the light comes on, the odometer starts working. Any ideas here?

Thanks guys, just trying to keep from getting frustrated over this stuff.

Anthony

 

No... CPU #2 is a rather large longish unit on the left side cabin wall. (Did you check the "buckle" switch and its connection on a green-yellow wire to terminal 1Q of CPU #2? If open, that switch will cause a seat belt light.)

Re: Code #05, did you read five short (0.4 second) equally-spaced flashes? Are you sure one or more flashes was not 1.2 seconds long?

 

Page C1a in Section Z of the '94 FWM says the odometer LCD is connected only to the "CPU and Speedometer" block inside the instrument cluster. Also the coolant warning lamp connects to that same block. Your odometer problem and cooling light problem both sound like the causes are internal to that block.

 

Thanks for the responses. I guess the weirdest part for me is that none of these issues existed before the motor died in January ( minus the odo not working ). I'm gonna pull the cluster this weekend and try to fix the knock sensor

 

Normally the fault detection for the knock sensor looks for a short circuit, but Mazda claims it also would fault on an open circuit. That's odd, because I've measured the resistance across a brand new sensor and found only an open. I think the device is a piezo-electric voltage generator, like a spark generator in a propane barbeque. If so, one would expect an open circuit to be seen on a new one.

 

Im about at my wits end on this knock sensor. Ive tried two known good sensors and run new wiring from the ecu to the knock sensor and no help, still check engine light, code 05. Any ideas on a way to trick the ecu into thinking the knock sensor is working?

Also, any insights on a code 76?

 

What is your response to my question in post #2 above? Is it maybe not Code 5?

Regarding Code 76, that applies only to an automatic transmission FD. The '94 manual says "Slip lock up off signal (PCMT)" and a PCMT is not present in a manual tranny car.

 

The flashing is five short, equally timed blinks. Im quite sure it's 05 as it's also running in limp mode. The code 76 is odd as I never had a check engine light before the swap. 76 may not actually trigger a check engine light though. As far as the knock sensor, I may just have to bite the bullet and go PFC earlier than planned.

 

Are you sure the knock sensor wire is not shorted to ground inside the wiring harness? What happens if you disconnect the knock sensor wire, leaving it open, clear the fault codes by disconnecting the battery for 30 seconds or more, and pressing on the brake pedal, and then reconnect battery and retake the fault codes? Does the 05 code disappear? (IIRC, the fault system will not detect an open from the knock sensor, only a short to ground.) If you still see a fault, there is probably a short in the wiring. Check resistance between the wire harness terminal and chassis ground. If you don't, re-connect the knock sensor and check faults again. If code 5 is still gone, the KS is probably OK.

 

I'll do that today. The light went off for about ten minutes today while driving. Came back on but leads me to think the problem is in the harness.

Thanks

 

So after doing what you said, disconnecting knock sensor resetting ecu and rechecking code, it's still there. Does that mean there is a short to ground in the harness somewhere on that wire?

 

At this point I would check the resistance (ignition OFF) from the wire harness terminal to chassis ground. If it is very low, comparable to say the lead-to-lead resistance of your ohmmeter, there's a short in the wiring or in the PCME connector or inside the PCME.

 

Im going to do this today. I think you are on to something as anytime I hit a decent bump in the road, the light goes out then comes back on. Im assuming there is a short in the harness. Do you know which color the knock sensor wire is at the ecu plug? From what I saw on the wiring diagram in the FSM, it shows W, I would guess white? I read on a knock sensor thread that its light green however. Just want to make sure I get the right one.

Thanks

 

The wire is shielded. The jacket is grey, but the inner conductor (inside the shield) is white all the way from the knock sensor connector (B1-46) to the PCME terminal 3M. This is according to the '94 FWM Section Z page B-1c. But the short wire connecting into the knock sensor could be light green, or any other color depending on the knock sensor vendor.

Incidentally, the schematic representation of the knock sensor on that FWM page confirms that it should look like an open circuit to an ohmmeter and is a piezo electric crystal, which generates a voltage spike when shocked. I seriously doubt there is any test equipment that could independently show operation of the sensor, short of an oscilloscope.

 

How would you recommend I test/fix this issue? Is a new wiring harness about the only solution?

 

If the PCME wire harness connector containing the shielded wire is still connected to the PCME, I would disconnect that connector (middle, shortest connector of the PCME) and recheck the wire in the harness, i.e. look at resistance between that white wire (that would connect to PCME terminal 3M) and chassis ground.

If you still see a short, the problem is definitely in the wire... maybe a strand of the shield is touching the bare end of the white wire.

If you seen an open circuit, the wire is probably OK (wiggle it to make sure). In that case, I would next check the resistance between terminal 3M of the PCME connector (on the unit) and the PCME chassis. A very low resistance in this test would mean that the problem is NOT in the wire, but in either the PCME connector or internal to the PCME.

 

I tested resistance between the PCM connector and knock sensor plug. With Ohmeter leads touching I get 0.3 and through the wiring I get only fluctuations between 0.5 and 0.6. I think that narrows out the wiring harness?

 

Sounds like you were testing continuity through the wire, if I understand correctly, and that 0.5 - 0.6 ohms would say the wire is not broken. But that's not the test I was suggesting.

On the other hand, if by the "knock sensor plug" you meant a chassis ground point like the case of the knock sensor, that 0.5 - 0.6 ohms would show the wire is shorted to ground, and the problem is in the wire.

I had intended to point you to checking from the white wire (on either end) to the wire's shield. There should be an open circuit indication if there is no short in the wire itself. If you see a short there, the problem is in the wire. If not, check the resistance from the 3M terminal on the PCME itself to a handy chassis ground point. You should not see a short there either. If you do, your problem is in either the PCME connector itself or inside the PCME.

 

Ok I missed the point initially but I'm getting there haha. I'm new to using my multimeter in this fashion so please bear with me. I'll get back out there an check this.

 

So I started tracing the knock sensor wire from the sensor. Found a break where the single wire leaves the main harness under the alternator. Soldered in some new shielded wire from there to the connector and voila! No more problems.

Thanks for all your help!

 

Congrats on finding and solving the problem. So somehow the fault detector system CAN detect a broken wire... I find that very surprising., since the sensor itself is essentially an open circuit, at least according to the schematic. (Maybe there is a resistor in the sensor of some high value that the ECU needs to "see.")

 

I think it was te grounding shield touching the core wire. The break in wire was pretty gnarly. I learn more everyday with this car.