so i'm still checking the wiring... but i had a thought that if the reason for the stall was that the O2 sensor was indicating
not only an artificially lean condition, but also an unrealistic one (volts < 0.1), that i could:

• take a spare old sensor (i happened to have one) and connect it,
• but leave it uninstalled (exposed to garage air) while keeping the suspect one installed in the exhaust (but disconnected)
• and that this would simulate an installed sensor reading fresh air

turns out, the engine died after the sensor got below 0.1, and i noticed that the minute-long delay that the original sensor was taking before the voltage started to increase after the stall was absent (see chart, below). when i removed the original sensor and installed the spare sensor, the engine did not die, and the sensor voltage oscillates around an average of .478 volts with min of 0.1 to 0.8 volts... so i'm thinking the original is at least part of the problem.

the sensor i pulled out was also coated with black soot--i suppose it's been running rich for awhile, and i have no idea how old it is.
i ordered a new NTK sensor to replace it (the re-designed one), and will continue testing the wiring.

i downloaded the Drivetrain Diagnostic Manual from your google drive (thanks!!!) and can immediately see how valuable it is--will be using it to guide some of my harness testing.

does anyone know the details of the mapping used by the PCM to drive the IAC based on the sensor data? wondering if perhaps the artifically low voltage might be sending the map into non-viable realm for engines that aren't fresh from the factory...

 

beaker152... I applaud your tenacious diagnostics. I've gone down this road before when my ECU board completely fried. I did not know it and replaced ignition parts, sensors, and cleaned as many ground points as I could reach before I got out the test meter and started noticing irregularities. Thats when I jumped on her and started getting diagrams and help. It wasn't long before I realized that my ECU was bad. I'm not saying yours is bad.
I have a 1991 so I don't have the luxury of OBDII. I will assume that some things did not change, you can check your 1998 Engine and double check. Your O2 sensor gets it's ground for the heater from the engine block. It does not share that ground for the sensor. The ECU provides a common ground circuit, for all of the sensors, O2 Sensor included (Speed Sensor, Air Temp Sensor, Coolant Temp Sensor, Crank Shaft Sensor, Camshaft Position Sensor, MAP Sensor, & Throttle Position Sensor). If your O2 sensor is bad in such a way that internally it is causing intermittent problems that may or may not be related to outside ambient temperature then it might be a good idea to replace it.
Your O2 sensor gets power from the fuel pump relay wire off of pin 87 which feeds power to the Ballast By-Pass Relay which engages after you first start your engine and it reaches proper running temp. The ECU allows full battery voltage to run to the fuel pump through the Ballast By-Pass Relay during warmup then when the coolant temp sensor tells the ECU that it's up to a certain temp the By-Pass Relay opens and the Fuel Pump Relay power is sent through a Ballast Resistor that lowers the voltage to the fuel pump intentionally to cut pressure to the fuel rail. This could explain the different fuel pressures that you observed. If the Ballast Resistor is bad or completely broken this could explain the various run times from cold start and why you can't immediately restart the engine after stall. The engine is already warm so the bypass relay is sending power directly to the resistor thus will not start. After you let it cool down, the fuel pump relay is sending full power through the bypass relay to the fuel pump and allows it to run until it's warm. Check continuity and ohm out the ballast resistor, I don't know the proper ohms for the resistor but your field manual should have that information. If your Ballast Resistor is bad or broken, that's a cheap and easy fix, if it's been on its way out and possibly allowing too much voltage through then that would explain the amount of soot on your O2 sensor and possibly poor mpg?
I hope your engine has a ballast resistor, otherwise I wrote all of that for nothing.
Good luck, and keep posting, I'm anxious to hear what the true problem is. I've got an electrical gremlin in my 91 Cherokee Sport that I'm chasing down right now, has to do with the auxiliary cooling fan, previous owner did some Frankenstein work and lost the factory wiring and put his own wires into the dash and wired the fan to a manual switch. I'm trying to get that aux fan back on the ECU wiring so it turns on and off automatically like it's supposed to.

 

cspaltro,
thanks for the kind words! it's been a frustrating couple of weeks, but i think i've made some incremental progress. but first, i don't believe the '96 has a ballast by-pass relay, but i can see how the temperature dependent behavior could present in a similar fashion. luckily, i've had consistently high fuel pressures (i keep a frequent eye on the fuel pressure gauge when testing now), but there's still indication from the *new* upstream O2 sensor that it starts running lean before it dies...

i got a test light and a spark indicator to enhance my testing joy, and whipped up a data logger from a raspberry pi and the ADC pins on an esp 32 microcontroller in an attempt to gather some deeper data (it's logging at about 1kHz) than what's available over my OBD II gadget... i felt paranoid about my crank and camshaft sensors (despite having no codes to indicate otherwise) so i looked at those first and saw what appears (to my novice eyes) to be reasonable:



unfortunately, i didn't use properly configured resistors in my voltage divider (had to lower the voltage to not fry the esp32) and the engine would die whenever i plugged in the data connector--i believe it was drawing too much current (a few milliamps) from the CKP signal... the chart above, therefore, was from a crank no-start scenario. i have since upped the resistance to 1 M Ohm and now the engine will run with it plugged in

the spark indicator is in-line between the coil and distributor cap, which makes it handy to tell when the coil is working. as per usual, it started up fine this morning when it was very cold, then died after a few minutes... i cranked again and it tried to start, spark indicator lit up, but it only fired for a few revs, then died. as the test light had been indicating good voltage at the coil from the ASD, i made a probe that would fit snugly inside the coil driver side of the connector so i could test the signal from the PCM--here's what i got:

of course it doesn't start with the connector unplugged, so i plugged it back in (confident in my coil driver signal) and saw this from my crank/cam position sensors:

that constant CPS voltage caught my attention, so i measured the voltage from the CPS and saw it was 5.16V... which would be fine if it alternated with something low, but as i bumped the engine several times i verified it was remaining constant, took off the distributor cap and pulled the CPS and noticed the voltage change... then started moving the wires around and saw it change again... i had previously wiggled wires around in many different places, including the CPS, but if i put just the right pressure on them, in just the right place, i could watch the voltage go from 5.16 to a few millivolts... seemed to me like it was a fair assumption that there was some strain-related effect, and that that effect might be activated by thermal expansion as the engine went from freezing to warm that ended up confusing the PCM so much that it didn't know when to pulse the fuel injectors (maybe that's what caused the O2 sensor to indicate lean and the engine to die?), so i ordered a new mopar sensor today and hope to have it later next week and in the meantime, will be investigating possible wiring problems (intermittent shorts/open circuits). i'm trying to avoid firing the parts cannon, especially if the last cannonball is going to be a new PCM that still won't fix the problem, so i will continue attempting a data-driven approach until my patience runs out... i have found the suggestions on this thread and other forums very helpful, and hearing other's adventures in jeep wiring have let me know my situation isn't unique, so i hope this helps the next person. looking forward to the new CPS and hoping to close this thread!

btw, to tap into the sensor data, i went to the junkyard and got the cps, cpk, and o2 connectors and spliced the pairs with a data tap, so i can insert it between my existing connectors.

 

Nice work, I gotta say I didn't expect a Raspberry Pi to be employed but kudos on that, I like the granularity of your data. I'm a bit of a computer geek myself, I never thought about using a Pi to log data from my engine... I may do that. Did you download a logging program, or did you write it yourself? I'm up in Idaho and in the summer have problems with overheating, the gauge on my dash is not as accurate as I'd like... I'd also like to monitor individual injector volumes if that's possible as well as some other stuff. with some custom harnesses I could do that.
Let us know what happens with your new CPS and if you have to replace the wiring harness to it . I can't imagine much heat building up on that side of the engine, in the Colorado Winter temps, enough to cause the PCM to get confused. I'm looking forward to your next post. Good Luck, and Merry Christmas!

 

a question about that coil driver voltage...
when i first gathered data, i noticed what appeared to be a lot of noise at 20Hz. when i apply the test light, the noise drops out, remove the test light and it comes back.
this chart is a window from the previous one i posted, but zoomed in. the voltage has a baseline of about 1.2V, then with cranking, i get pulses of 10.V at 10Hz for a second or two before maintaining about 10.8V, then drops back to 1.2 when i stop cranking. when i remove the test light, it resumes oscillating at 20Hz with about a 10V swing...

my expectation is that the coil driver circuit should show 0V until cranking, then show the pulses for the ignition coil to provide spark, and if there's no start and cranking is stopped, it should go back to 0V. to me, the noise looks like there's some ringing in the circuit when it's not grounded by the PCM, which makes me wonder if there is some incidental grounding happening... (i noticed the noise even with PCM 1 unplugged, though at a lower baseline voltage)

shouldn't the coil driver be pulsing the entire time it's cranking? i thought the high voltage for the spark comes from the collapse of the circuit on the driver side...
does anyone have a scope they can throw on their coil driver circuit before/during/after cranking?

 

i resorted to the pi/esp32 combo because i don't have an oscilloscope and (among other reasons) the one i'd like to get if i got one isn't in this month's budget and i happend to have them lying around...

i'd also like to instrument the injector pulses--i saw on another thread that there can be a strong voltage spike (60-90V) when the magnetic field collapses, so that has to be protected against to keep the esp32 safe--i've been reading a bit about the kind of circuits that can protect against such things before sending the signal on to the more delicate components. i'd like to measure the coil driver signal when it was connected, too, but i was afraid it might also have some big voltage spikes from the high side of the coil...

the code to read the voltages, send as serial data to the pi via usb and write to a file (i know there are other ways to do this) was surprisingly straightforward and short--i'd be happy to share what i have if you send a private message. i think there's another microcontroller other than the esp32 that would be good for reading the data, perhaps a separate dac chip, too. would be interesting to collaborate on a data acquisition logger and custom dash gages.

 

i got the new Mopar CPS, installed it, and was able to start the engine in the afternoon when it wasn't freezing and i let it idle for about 30 minutes, everything seemed fine. started it the next morning when it was freezing and it ran fine for another 30 minutes with everything seeming fine... at this point, i'm going to assume that the problem was the previous CPS only working properly when it was very cold, then failing once it warmed up a bit (or even when i tried to start it when the ambient temperature wasn't cold enough). i found the data i gathered helpful in the diagnosis, and i think it has pointed to some other potential electrical issues that i'll be investigating in another thread or two.

thanks!

 

Did you have a code on the dash? I got no codes on dash or OBD reader. I don't throw money at it unless I know it's bad. 95* here so haven't checked ohm's. It doesn't happen warm keeps running. Been through grounds & power. If a sensor it should have a code. My bike software will tell me which pin connector to look at and it + or -. It's a 2000 so I haven't found a ballast relay or that stuff at all. Too hot to work in the sun. Maybe start heating things up w/heat gun to see if it makes a change.
ASD relay has power & also from ign switch, if I ground it on D6 and start truck then it would be no signal(ground) or crank sensor as my issue