AW4 Troubleshooting and electrical gremlins?

My transmission has recently been acting up, and I've been trying to troubleshoot it. It would not shift into overdrive at all, and the later it wouldn't shift all the way down to first gear when trying to get going from a stop.

I've checked everything I can think of short of the replacing the TCU, and with exception of some weird solenoid resistances everything looks fine. Sorry for the lengthy post, but I've done a LOT of testing without solid results!

I've done:
-Check fluid level/color
-Torque converter will lock up in third like it should when the gear selector is in the '3' position.
-Unplugged TCU and shifted fine "manually" (1st, 3rd and OD)
-Check both power feeds and the ground to the TCU.
-Checked speed sensor, TPS, Brake switch and NSS. Verified these right at the TCU connector.


I've tried to test the solenoid resistances and come up with some strange results that defy everything I know about electronics. For all of this the TCU is disconnected. Despite my FSM diagram, the solenoids appear to be grounded through the transmission, and not the 'ground' wire at the TCU or connector found in the engine bay. I've done my testing both at the TCU connector and at the connector in the engine bay right before these wires run into the transmission.
-At the TCU connector the solenoid resistance is always zero ohms between the signal wire to a body ground.
-Disconnecting the plug in the engine bay to check there, they read normal 11-12 ohms.
-With the engine bay plug disconnected the resistance between all three solenoid TCU pins and ground is infinity (open ciruit).
-Continuity between the TCU connector and connector in the engine bay is fine.
-I've back-probed the connector in the engine bay (while it was connected) and come up with 19-20ohms for each solenoid.
-If I measure between two solenoid feed wires at both the connection in the engine bay and at the TCU connector, I get ~23 ohms, which I think is correct since that's reading through two solenoids.
-Verified that I'm not seeing some sort of voltage that would screw up the multimeter on any of these pins.
-I've used a jumper wire to 12V and could hear all three solenoids engage. This works at both the engine bay connector and to the TCU connector.

Now I just don't know what type of situation could cause the resistance at that connector to be good (12ohm), but jump up to 19 when plugged in, and still show zero at the TCU, which is connected together by a wire I've shown to be good! And through all of this, the solenoids will 'click' when power is applied.

Is there anything I've missed testing? Does anyone have a clue why the solenoid resistances would act like they do? About the only thing I can think of is that maybe my multimeter isn't working right, and I still need to borrow one to verify this.

If the solenoids are working correctly, I'm tempted to call my troubles a TCU issue, but the resistances I've seen don't make sense to me!

Thanks in advance.

Were any DTC's stored? If so, which ones? Sounds like it may have been going into "limp" mode. Which happens when a fault is seen and that should throw a DTC. The DTC should get you in some kind of direction. Be it range sensor, solenoid (open/short), speed sensor etc. Attempting to find a transmission issue with just a DVOM can be very painful. I'm assuming you don't have the ability to watch transmission data via scan tool?

Do you have a transmission pressure gauge? Just because a solenoid "clicks" doesn't mean it flows.

You are obviously knowledgeable in electronics. With some simple wiring, you can control the transmission's electronics. Throw in a pressure gauge and you have the ability to see if the solenoids are doing their job.

I wouldn't do anything without first knowing if/what DTC's were stored.

The AW4 is a very resilient transmission. Probably one of, if not the best part of an XJ. I would highly doubt it to be mechanical. Most common is solenoids/wiring. TCM's in my experience are not common failure points but I have seen it. Usually due to a short elsewhere.

I doubt I have been able to provide any help with my ramblings. I hope it is at least comprehensible, it's late and I'm way too lazy to proof read all that crap.

What year is your XJ?

Did you use the "Shift Pointers" document as a guide while you tested the TCM and TCM harness?

You can pull the pan and test the solenoids directly, to be doubly sure that they are 100%.

However, since everything else checks out until you introduce the TCM, it's likely the TCM can be at fault. Granted, a TCM failure is known to be uncommon, it can still happen.

When you unplug the TCM, and you manually shift into '1', will the tranny shift itself into '2', then back into '1'? You'll have to rev the engine ridiculously high for it to shift into '2' when driving, but it's designed to do that.

I have a spare TCM for 97+ and 97- years. Not completely sure if they work, though.

Thanks for the ideas guys.

When you refer to DTCs, are you talking the typical check engine light or one stored in the TCU? I have no check engine codes, and of unfortunately don't have the tool to scan the TCU.

I'm going to rig up switches tonight for the solenoids and take it for a careful test drive to see if I can manually trip them. If I can 'shift' using the switches, can I assume my solenoids are good, or should I run and grab the pressure gauge anyway? A gauge don't seem too expensive if it will really help.

I was using both "Shift Pointers" and my FSM to do the debugging. That link wasn't 100% accurate on everything when compared to my FSM and Jeep, but it was still awfully useful.

I did not even know you could get it into 2nd gear with the TCU unplugged, so hopefully I can try that tonight as well.

So if I can shift with the switches, it's likely the TCU. If I can't, it's probably a solenoid and I could grab a pressure gauge or drop the pan to check the solenoid resistance right there. That sound like a reasonable way to approach this?

I was also thinking I could hook up two LEDs and watch the TCU to see if it's trying to trip the solenoids or not...

My Jeep is a '96. I've looked around and for some reason TCUs seem to be listed as 92-95, 96, and 97+ but there's a local junkyard that lists a '96 one for a reasonable price if I decide to give that a try.

AFAIK, the 96 computer is compatible with 95 and earlier. I can go out and compare my 95 computer with the 96 I have. If they match, I imagine I can send you mine for cheaper than the j/y wants. Just PM me.

When you shift it manually into '1', you'll just have to drive it and it'll shift on its own. It seems counter-intuitive, though, because it almost redlines the tach.

I would just go ahead and drop the pan and test the solenoids directly. That's probably the best way to be sure the solenoids are working properly.

Well I was able to shift with the switches. It wouldn't shift into second gear when the TCU was unplugged. I don't know what that means, especially since I was able to shift by tripping the solenoids directly.

Depending on how I feel tomorrow night I might drop the pan and test the solenoids directly just to be sure, but I guess right now everything points to the TCU?

When you manually shift into '1' using the gear selector, you are changing the tranny using mechanical linkages. However, there is no '2' linkage, so it shifts using the solenoids. At least, that's how I understand it. If I'm wrong, someone will chime in, call me names and correct me.

I would say to drop the pan and test the solenoids manually before assuming it's the TCM outright.

Sorry, I've been busy lately and haven't had time to get on here. Sounds like the solenoids are working fine. It's true the TCM/TCU "tells" them what to do. However it needs to see certain inputs to do so. Speed sensors and MLPS/NSS. Resistance testing isn't always definitive on the speed sensors, don't ask me why. Any updates?

xjparts said:

Sorry, I've been busy lately and haven't had time to get on here. Sounds like the solenoids are working fine. It's true the TCM/TCU "tells" them what to do. However it needs to see certain inputs to do so. Speed sensors and MLPS/NSS. Resistance testing isn't always definitive on the speed sensors, don't ask me why. Any updates?

Click to expand...

True, but you wouldn't measure these sensor's resistance; you'd test the voltage (according to Shift Pointers).

Also, he didn't mention anything being out of whatck except for the solenoids resistance.

1. Wiring issue
2. TCM issue
3. Solenoid issue

Sorry, not familiar with "shift pointers". I would typically watch speed sensor data with my scan tool or for hard to find intermittent problems, I use my four channel lab scope to catch any glitches.

I would never rule out it being a TCM issue. I have quite a few here if it comes to that. I keep them around for quick testing...

xjparts said:

Sorry, not familiar with "shift pointers".

Click to expand...

bigalpha said:

Did you use the "Shift Pointers" document as a guide while you tested the TCM and TCM harness?

Click to expand...

Yeah, I saw the link you posted. I didn't bother to read it, and still won't. Like I said, I'm not familiar with it, not that I haven't heard of it. Maybe if it were a tech article by BillaVista I'd read it. But in all honesty it's not that difficult with the proper equipment. Not having the proper equipment makes it a little harder to properly/easily diagnose. I'm not bashing anyone for not having tools or equipment by any means, most people shouldn't have the crap I have.

I'd rather use the ATSG manual and figure it out. Which by the way is easily found on the internets.

:flame:

I had to take a night off to relax, but I did a little bit of work last night. Right now I'm still avoiding dropping the trans pan and measuring the solenoids directly, since they seemed OK with the switches.

So for kicks I did jam my LEDs into the TCU connector to watch when it's triggering each solenoid and took it for a spin... and everything looked/worked fine. It shifted fine. So it is either intermittent, or it was an electrical connector issue that I cleared up by poking around.

Right now though, if it's working fine I'm skeptical that I can actually find a problem until it does it again. I'm sure if I had the right tools it'd help... buuut all I've got is my DVOM.

As for the speed sensor, I went back and read the Shift Pointers article and I guess I didn't follow it word-for-word. The FSM says to check the sensor by resistance (switch open vs. closed), which is how I did it. I should probably verify the voltage as well to make sure it's not borderline, like 4V pulses.

Up to 97, the rear output speed sensor was a reed relay, which closed once per revolution. One signal wire is biased high and the closing switch pulls it to ground. So with a voltmeter you should see +12 volts or zero depending on the position of the output shaft. Going fast enough that would look like an ac voltage with a dc offset (depends on the quality of your meter). A missing speed signal would cause it to never upshift.

I agree your measurements seem odd.

Update???

Sorry I haven't updated before now, but there's not much for me to say. I've only driven it a couple times, but it hasn't had a problem since then. I can only guess that between my unusual measurements and the intermittent behavior that it was/is some sort of wiring or connector issue. For now though, I'm still not sure there's much I can do about it until it starts causing trouble again.