2008 mazda cx-7 2.3L TC; Charging system, Engine performance

Got a wonderful car lot "auction special" in the other day. Came in for vehicle dying, rough idle, various engine codes. Measured battery voltage after it needed a few jumpstarts, and would stall once jump pack removed.

Battery tested bad, and the alternator as well. Pcm controls the alternator with a duty cycle to a transitor base inside the alternator, and watches the voltage output on another wire. (The pcm wiring checked out later on.)

Long story short, I replaced the alternator and the battery, started er' up. I watched battery voltage sky-rocket to 18 volts, and started smelling the rotten eggs from the battery.

Looked at scan data again, saw the the pcm was still commanding 100% field strength for the alternator. (Which is probably what fried the old alternator and battery.) Also found this in scan data, module voltage data PID showed 7-8 volts.


Ended up being this ecm 10a fuse, bottom row right. replaced and back to normal. 14 volts. No more weird electrical anomolies.



Now we're getting started. After the repair, the car is noticeably misfiring, and pulling a p0203 injector 3 circuit fault. (Never there before) I, of couse, ignored it, and went for a test drive. Really low power, sluggish, barely even accelerates. Never stalled, different symptoms than before. Parking lot test, when it actually would enter closed loop after start up, showed average 25% st, and no long term correction. (Ignored it, mistake.)
Then it would enter open loop. And run worse.

Suspected plugged exhaust, performed back pressure test at the upstream oxygen sensor. 2psi max under snap throttle, and no pressure at idle. Checked out. Moved on to engine timing, Engine mechanical. Potentially wasted a bunch of time on manifold vacuum readings that seemed bad. Checked engine cranking compression, and running compression. When the compression results "checked out". I started back at the beginning. The lean condition.

In the process of checking the alternator I noticed pcm had a business card and an eBay warranty taped to it. So this thing has been in the back of my mind ever since. I went back to scan data and and payed more attention to the open loop, closed loop situation. Finally narrowed it down to almost clock work.

Approximately 40 -45 seconds after start up, the vehicle reverts back to open loop, and the idle quality suffers greatly. Fuel data PIDs are normal during closedloop. Stft is stuck around 25%, but the injector pulse width is normal, the injected fuel amount is steady, and the a/f o2 sensors appear to be switching normally. Once this thing goes into open loop, all the data PIDs appear to be duty cycled, showing 0ms injector pulse width, 0mg/s injection amount, and the o2 sensors volts/amps are pegged in the lean range. Desired equivalence ratio becomes 2.0

Is this some kind of limp mode for the misfire, or is this a cheap ebay pcm fuel strategy failure that has me pulling my hair out. Hopefully I'll find out more soon to post. Had to rant about this one. Here's some captures of the cld/open loop s***storm.

maybe try unplugging the alternator to eliminate an ac ripple issue. since it didnt run like that before. also sure seems like an ebay special ecm issue to me. id check the powers/grounds and make sure nothing funny is going on with them. thats where id start....

What a trainwreck! :lol: Not your fault, just typical auction BS. For my information, at this point with the fuel control loop shenanigans, are you still getting the P0203? Or any other codes?

I don't doubt the PCM is probably garbage, but this initially strikes me as a default strategy in response to a perceived misfire. The 'Target AFR' PID is asking for 2.00 (stupid lean), and it's getting it. I think it's a default because I've seen many Honda engines react the same way in response to a misfire. Once it's decided there's a Type A miss at work, they'll take all the fuel away from that bank, and leave it that way for the rest of the key cycle.

no other codes, besides p0300. I was thinking all day about how to attack this thing, and decided to stay late after work and go for the injector fault code. I'm with you tyler, thought that it could be a strategy to combat the misfire.

No control from pcm, and on this system it seems both injector wires are pcm controlled. measured all four injector's resistance, injector waveforms, continuity between pcm and injector harness, short to ground or power on affected wires; all checked out. Here's a couple of snapshots and waveforms.

This is on one injector on both sides, in hindsight i forgot that two injectors share a ground. im assuming thats why im seeing two patterns? I dont know..


diagram, i was on injector 1 in the above waveform


this is on inj 1 and inj 3, on the wires that are not shared.


probing both sides of inj 3, no activity on either. I wonder if inj 2 was being affected too, since it shares a wire with 3. who knows.



Will advise customer that he definitely needs a pcm and it should fix some other issues with clsd loop operation too. He's been dying for this car since he dropped it off 3 days ago, but hopefully ill have time to post the good waveforms when and if he decides to fix it.

Man, I'm really not as familiar with direct injected electronics as I should be. :blush: I also don't know how valid resistance measurements are on these injectors, but with the checks you've done, I'm inclined to suspect a bad injector.

Also, not nitpicking, but I thought these systems shared a power? Instead of a ground? I'd always heard that there are two high voltage capacitors in the PCM, and those are shared. But like I said, I'm pretty weak at DI. :silly:

This Premium video comes to mind:

I'm not too familiar with the circuit designs on gdi injectors either. I've watched this video before but I will review the gdi part and rethink all of this. I'm pretty confident the pcm is bad, but I do lack some knowledge here. This one has me scratching my head a bit, there's a little more too this vehicle that I don't post just for ease of explanation.

But after I did all those checks the other night, I took my diagnosis and put the car back together until the pcm arrives on monday. Started it up the next morning, waited for it to enter it's fault condition. It never did, no misfire, stayed in clsd loop, fuel trims were perfect. No lean correction on st, 0% on lt. Took it around the block, accelerated fine. Might take it on a test drive today, hopefully won't get stranded ha.

I'm 99% positive I checked with a multimeter for a short to ground on the injector harness before I unplugged the pcm, I don't understand how me just simply unplugging and plugging back in the pcm connectors how it could have fixed itself.

Random little tidbits I've left out

Bent pin on the other pcm connector port, brake switch 2. Ignored it.
Broken connector guide on the pcm connector port the injectors are on.
Trouble commuicating with the vehicle unless I hold the dlc connector in a certain position.
Never checked every power and ground on pcm, just a few until found the ecm fuse blown.
Pcm does make a good 5v ref.
Originally had a intake cam solenoid code that has since disappeared.

I'm going back up to the shop later to grab some "known good" injector waveforms since it seems everything is normal now. Maybe do some harness wiggle tests around the inj and pcm plugins. We'll see.

Gjauto wrote: But after I did all those checks the other night, I took my diagnosis and put the car back together until the pcm arrives on monday. Started it up the next morning, waited for it to enter it's fault condition. It never did, no misfire, stayed in clsd loop, fuel trims were perfect. No lean correction on st, 0% on lt. Took it around the block, accelerated fine. Might take it on a test drive today, hopefully won't get stranded ha.

I'm 99% positive I checked with a multimeter for a short to ground on the injector harness before I unplugged the pcm, I don't understand how me just simply unplugging and plugging back in the pcm connectors how it could have fixed itself.

Random little tidbits I've left out

Bent pin on the other pcm connector port, brake switch 2. Ignored it.
Broken connector guide on the pcm connector port the injectors are on.
Trouble commuicating with the vehicle unless I hold the dlc connector in a certain position.
Never checked every power and ground on pcm, just a few until found the ecm fuse blown.
Pcm does make a good 5v ref.
Originally had a intake cam solenoid code that has since disappeared.

:lol: This Mazda is such a basket case. :silly: Well, I guess that means the bad injector is out, then. I suppose there could have been a pin fretting issue, but that usually takes time to develop. Maybe it was all down to someone hacking up the PCM and connectors?

I'm going back up to the shop later to grab some "known good" injector waveforms since it seems everything is normal now. Maybe do some harness wiggle tests around the inj and pcm plugins. We'll see.

I'd absolutely be interested in seeing these known good waveforms, if possible! Perhaps those, combined with the waveforms you took earlier, might explain why the PCM connector thing worked out?