• Welcome to the new NAXJA Forum! If your password does not work, please use "Forgot your password?" link on the log-in page. Please feel free to reach out to [email protected] if we can provide any assistance.

Intermittent starting problem - fuel? or something else?

br1anstorm

NAXJA Forum User
Location
United Kingdom
My 1993 XJ 4.0 (auto, injection, US-spec) - with 106,000 mostly on-road miles - has developed an occasional starting problem. I've searched the forum (and FAQs) and there seem to be so many possible causes I need help in narrowing it down.

First off, I haven't done any recent work on the engine except to fit a new Optima battery when my 5 year old AC Delco finally died. But the starting problem had showed up before that, and it has now recurred.

Symptom is that the starter motor cranks, the engine turns but doesn't fire. After 2 or 3 attempts, if you floor the gas pedal and try again, it starts. Once running, everything is fine (no rough idle, no stalling). The starting problem first showed a couple of months ago, just after filling up a near-empty fuel tank. Sometimes starting was normal; sometimes it wasn't. Battery and spark were OK - it simply felt as if fuel wasn't getting through - thouh I guess it might also mean it was flooded? Typically, when I took it to a workshop, I couldn't show them the problem - it started faultlessly every time!

Then the problem seemed to go away. I assumed the cause might simply have been bad or dirty fuel (crud stirred up when filling the empty tank) or a temporary fuel blockage. After several more fillups and a squirt of injector-cleaner all has been well for a while.

Now the problem has reappeared. Just occasionally, it cranks but won't fire. Flooring the pedal (which I would normally never do...) when turning the key then seems to get it going.

I'm not keen to begin randomly dismantling/replacing parts on a guesswork basis. What is the most likely problem - is it fuel-supply related (pump, filter) or is it settings/sensors/controls (I'm not sure I understand all this TPS, IAC, CPS stuff)? What checks/tests should I do, and most important, in what order, to pin down the problem? Would there be any diagnostic codes, and how can I check them? (I'm in the UK, and local dealers don't have much expertise on a 14 year old US-spec XJ, so it's down to just me and the Haynes manual!)

Any advice would be welcome...

br1anstorm
 
Well you could start off by changing your fuel filter, pretty cheap part at any auto store, see if that helps, if not, then check your CKS crank position sensor, those are pretty common on XJs, if it is your CKS then its intermittent which means that it will work fine and then out of no where, your jeep wont start, I had the same problem in my XJ last summer, when I took it to the shop they could never get it to reoccur and finally the sensor stopped working. I think the CKS is about $87??? cant really remember, as far as testing your sensor goes, you'll want to check the resistance, there should be no resistance, your digital multimeter should say OL, for over limit i.e. 0. Make sure to set your dmm to the 1k to 10k scale when you do this. If any resistance shows up, replace your sensor. I think the haynes manual has a diagram of the sensor.

-Adam
 
Thanks. New filter will take a few days/weeks to find: I live in the Scottish Highlands, so it's a matter of mail order rather than local store. Which also explains why I can't simply try fitting new parts in a random way to see what effect they have - especially as the starting problem is intermittent. Also, I don't want to fix what ain't broke...

But I did find the diagnostic codes trick - ignition key off/on to get the 'check engine' light to flash. This is what I got:

12 - battery has been disconnected. Fair enough, I've recently changed battery;

17 - engine running too cold (thermostat problem?). Haven't noticed any other symptoms: temp runs at solid 210, a bit hotter in summer. How would this affect starting?

14 - MAP sensor voltage too high or low. Tried to check the A-B and A-C voltage at the pins with my old-fashioned non-digital meter, and got no readings at all! What next? Try another meter? Or does zero readings mean the MAP sensor is totally shot? If so, how does its failure affect the starting and running of the engine?

BTW, I didn't get a 55 code at the end of the sequence. But does the fact that I got no other fault codes mean that my CKS, TPS, and IAC etc are all OK?

I'm beginning to feel like Sherlock Holmes already - hunting for clues...

br1anstorm
 
My 91 did exactly what you are describing a few months ago. Got the code 17 on mine as well.

It was the engine coolant sensor for the computer. My thermostat was fine and the engine running at 200 deg. Scanner confirmed the computer was only getting at temp of 120 deg.

Mine would also flood on startup and ran rich. New $25 sensor and 2 minutes to swap it and all is well.

The sensor is in the thermostat housing with two wires going to it.

Andy
 
br1anstorm said:
Thanks. New filter will take a few days/weeks to find: I live in the Scottish Highlands, so it's a matter of mail order rather than local store. Which also explains why I can't simply try fitting new parts in a random way to see what effect they have - especially as the starting problem is intermittent. Also, I don't want to fix what ain't broke...

But I did find the diagnostic codes trick - ignition key off/on to get the 'check engine' light to flash. This is what I got:

12 - battery has been disconnected. Fair enough, I've recently changed battery;

17 - engine running too cold (thermostat problem?). Haven't noticed any other symptoms: temp runs at solid 210, a bit hotter in summer. How would this affect starting?

14 - MAP sensor voltage too high or low. Tried to check the A-B and A-C voltage at the pins with my old-fashioned non-digital meter, and got no readings at all! What next? Try another meter? Or does zero readings mean the MAP sensor is totally shot? If so, how does its failure affect the starting and running of the engine?

BTW, I didn't get a 55 code at the end of the sequence. But does the fact that I got no other fault codes mean that my CKS, TPS, and IAC etc are all OK?

I'm beginning to feel like Sherlock Holmes already - hunting for clues...

br1anstorm

Sounds like 2 problems. Sounds like the MAP sensor may be dead (which is very rare(1)). The MAP sensor tells the computer what the absolute air pressure is in the intake manifold. It affects the air/fuel ratio calculations at start up and while running. Your other problem is either an open loop colder running operation and or a bad coolant temperature sensor (or electrical connection to the sensor) which can also affect starting if it is lying to the computer, telling it it is cold when it actually is hot, which would tend to flood it during a a warm, or hot engine restart.

Before you replace the MAP sensor (1), exactly how, where did you test it? Was the wiring harness connected to the MAP during testing (If not it needs to be connected)? Was the steering column Key turned on, power on, engine off during testing? If not it needs to be on to get 5 volts to pin C!!!! You should have 5 volts from Pin C in the harness connector to the battery negative post (or any other known good ground). If not then it may not be the MAP, it may be the wiring harness has a loose connection somewhere else.

Nothing wrong with using an older analog meter. Verify the meter by testing the 12 volts at the battery on the volt scale.
 
Thanks a lot, Ecomike. What you say makes a lot of sense. Just a few instant reactions...First, on the MAP sensor.

- like you, I reckoned an old style analog multimeter should show a reading (and my meter works OK - I check-tested it across the battery).

- it may be that I simply didn't get good connections on to (or into) the MAP connections [but see next comment]

- i know the Haynes book isn't infallible. But it clearly says to disconnect the wires from the sensor and, with ignition on and engine off, take the readings across A-B and A-C (which are actually little female 'sockets' in the MAP connector - the wiring harness end has the three little male pins that fit into them). If the wiring is connected you can't get the meter to reach into the connector sockets to take a reading. But having said all that, common sense does suggest that if you disconnect the wiring harness, then you're not going to get a voltage from the (disconnected) MAP sensor!

- if the MAP sensor is dead (however unlikely) would the ECU/engine still start and run on some sort of default setting?

Second, on the coolant temp sensor issue, I did get a fault code 17, and I can see how an incorrect (cold) temp reading might fool the system and flood it on a hot engine restart. The temp gauge seems to be reading accurately, so that sender must be OK. The post by andyr354 seems very relevant. But how can I check the coolant temp sensor? I don't have any kind of scanner or diagnostic tools. Do I simply have to find a replacement sensor, swap it out and see if a new one makes any difference. Trouble is, it is really hard to check because the troublesome starting only happens very rarely!

Still, gotta keep trying...

br1anstorm
 
The computer and guage use separate sensors, so the guage one is fine and the one for the computer is bad if you are getting the code 17.

The sender for the computer is in the thermostat housing, the one for the guage is in the block down below the intake manifold.

You can test the sensor with a meeter by checking its resistance at a certain temp. I don't have a FSM that would have those values in it but if I could remember I could check the sensor resistance on mine once it is warmed up and you could cross reference. Or the values may be on the site somewhere.

Andy
 
i had a recent problem with an intermittant start which sounds very similar to yours and mine turned out to be the fuel injectors. you might want to check them and make sure they are in good shape.
 
deesiexj said:
i had a recent problem with an intermittant start which sounds very similar to yours and mine turned out to be the fuel injectors. you might want to check them and make sure they are in good shape.

Check and clean the eng ground...........if the eng think it at -40 (vary cold dump lots of fuel!!!!!!!!) at normal eng temp.............theses are the fist to placed to start.

unplugging it will not help as it will just tell the the computer that it is at -40 all the time and my be what is happening right now......Make sure its plugged in.


Flash.
 
br1anstorm said:
Thanks a lot, Ecomike. What you say makes a lot of sense. Just a few instant reactions...First, on the MAP sensor.

- like you, I reckoned an old style analog multimeter should show a reading (and my meter works OK - I check-tested it across the battery).

- it may be that I simply didn't get good connections on to (or into) the MAP connections [but see next comment]

- i know the Haynes book isn't infallible. But it clearly says to disconnect the wires from the sensor and, with ignition on and engine off, take the readings across A-B and A-C (which are actually little female 'sockets' in the MAP connector - the wiring harness end has the three little male pins that fit into them). If the wiring is connected you can't get the meter to reach into the connector sockets to take a reading. But having said all that, common sense does suggest that if you disconnect the wiring harness, then you're not going to get a voltage from the (disconnected) MAP sensor!

- if the MAP sensor is dead (however unlikely) would the ECU/engine still start and run on some sort of default setting?

Second, on the coolant temp sensor issue, I did get a fault code 17, and I can see how an incorrect (cold) temp reading might fool the system and flood it on a hot engine restart. The temp gauge seems to be reading accurately, so that sender must be OK. The post by andyr354 seems very relevant. But how can I check the coolant temp sensor? I don't have any kind of scanner or diagnostic tools. Do I simply have to find a replacement sensor, swap it out and see if a new one makes any difference. Trouble is, it is really hard to check because the troublesome starting only happens very rarely!

Still, gotta keep trying...

br1anstorm

If you want to really know for sure start the jeep, then disconnect the MAP sensor (electrical connection only) for a few seconds. As I recall it barely runs if at all and probably won't start at all with out the MAP sensor.

The real issue probably is whether the map sensor is accurate. That would require a vacuum gauge and vacuum tool along with your MM. I don't know anyone who has fixed a jeep yet by replacing the MAP sensor.

andyr354 seems to have answered your second question.
 
Flash said:
Check and clean the eng ground...........if the eng think it at -40 (vary cold dump lots of fuel!!!!!!!!) at normal eng temp.............theses are the fist to placed to start.

unplugging it will not help as it will just tell the the computer that it is at -40 all the time and my be what is happening right now......Make sure its plugged in.


Flash.

Flash,

Actually the computer is a bit smarter than that. If you disconnect the temperature sensor it falls back to a default table of start up values and should run fine like mine did for 2 years, all be it at the expense of fuel economy. The problem is that if it is attached to the temp sensor and if the temp sensor is lying to the computer about the actual temperature, then the computer gets stupid and believes it for a while. Sometimes the computers can wise up, figure out whats up, and then ignor a sensor that is giving it data that is causing operating problems. Or if the sensor output is out of bounds of reality then the computer simply ignors it.

But if the engine is at 210, and the temp sensor says it is 10 F, then the computer acts on that 10 F data and the engine runs poorly if at all, or it fails to start. In that case disconnecting the sensor would be a good diagnostic tool, and might get him running in a pinch while he is waiting on a new sensor.
 
Ecomike said:
Flash,

Actually the computer is a bit smarter than that. If you disconnect the temperature sensor it falls back to a default table of start up values and should run fine like mine did for 2 years, all be it at the expense of fuel economy. The problem is that if it is attached to the temp sensor and if the temp sensor is lying to the computer about the actual temperature, then the computer gets stupid and believes it for a while. Sometimes the computers can wise up, figure out whats up, and then ignor a sensor that is giving it data that is causing operating problems. Or if the sensor output is out of bounds of reality then the computer simply ignors it.

But if the engine is at 210, and the temp sensor says it is 10 F, then the computer acts on that 10 F data and the engine runs poorly if at all, or it fails to start. In that case disconnecting the sensor would be a good diagnostic tool, and might get him running in a pinch while he is waiting on a new sensor.


No argument here, YOU are corect!!!;)

Flash.
 
Ecomike said:
If you want to really know for sure start the jeep, then disconnect the MAP sensor (electrical connection only) for a few seconds. As I recall it barely runs if at all and probably won't start at all with out the MAP sensor.

The real issue probably is whether the map sensor is accurate. That would require a vacuum gauge and vacuum tool along with your MM. I don't know anyone who has fixed a jeep yet by replacing the MAP sensor.

andyr354 seems to have answered your second question.

Thanks both to Ecomike and andyr354 for your helpful comments... and to others who have chipped in. We're making progress!

The good news is that my MAP sensor seems OK. It did seem bizarre that the Haynes manual should say to disconnect it then see if you get voltage readings from the terminals with the ignition on. The crucial sentence the manual omitted is - after you disconnect, you have to use bridging or jumper wires to reconnect the A and C pins (which then allows you to get the meter clips on to the relevant places), then turn ignition on and test. I did so, and sure enough, got the right readings (A to C about 5v, and B to ground also about 5v).

So then I tested the coolant temp sensor which is fitted on the t/stat housing (all you need to do is disconnect then test the resistance - ohms - across the two sensor pins). Should be 7000 ohms cold, and <1000 ohms at operating temp. Well, I only got about 3000 ohms cold... and I don't yet know what I get when it's hot, as night fell and I had to pack up and deal with some other stuff. So I'll check the hot sensor resistance tomorrow. But that coolant sensor is looking like the culprit: it may just be erratic and delivering variable signals (which might explain why I only have the hard-starting problem sporadically).

Now I just hope they sell replacement sensors over here (UK) for a price not too far off the $15 or so which US suppliers charge!

Stand by for final (?) instalment in the story.

br1anstorm
 
Update on my coolant temp sensor tests. Resistance reading when cold, 30000 ohms; and when hot 20000 ohms. Neither is in the range the Haynes manual suggests. I did wonder if I'd set my analog meter scale wrongly (looks as if there's an extra zero, but I did double-check). Even 3000 and 2000 would still be abnormal.

So I'm about to go looking for a replacement sensor...

br1anstorm
 
It should be easy enough to get. GM uses the same sensor on alot of its injected vehicles of similar vintage.

Andy
 
Back
Top