Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save dochoffiday/6a0fabb7981237e8495a to your computer and use it in GitHub Desktop.
Save dochoffiday/6a0fabb7981237e8495a to your computer and use it in GitHub Desktop.
Jeep Wrangler YJ 1991+: Hard or No Start/Long Cranks in the Cold Explained

#Jeep Wrangler YJ 1991+: Hard or No Start/Long Cranks in the Cold Explained

**This is a rewrite of this post: http://www.jeepforum.com/forum/f12/91-hard-no-start-long-cranks-cold-explained-490788/ - so, all credit belongs to that author!

Ok, with the weather turning colder it seems like everyday people are popping up wondering why their jeeps aren't starting like they were in the warm weather. This post is going to try and explain one of the more common problems the YJ is known for and hopefully give you all some way to narrow it down before throwing parts at it that aren't needed. The underlying problem here is that the computer is failing to power up when you turn the key on.

Symptoms

You guessed it, the jeep doesn't always start right away. Instead it seems to sit there and crank forever. Usually (but not always) this is followed by a rapid clicking sound coming from under the hood and then miraculously the jeep will start. If you're really observant you might have noticed that the check engine light (CEL) doesn't come on at first and usually only comes on when you start to hear the clicking noise. The fuel pump also doesn't prime itself as usual and will also coincide with the clicking. If you've done some digging into the engine bay you might have also found that you have no spark while cranking prior to the CEL coming on. All of these things are connected my friend. So we have 4 symptoms to check off when you're trying to decide if this is your problem or not:

  1. No check engine light when you first turn the key to ON, but before START.
  2. No fuel pump priming (If you're not sure on this one, check with a gage)
  3. No spark prior to the CEL coming on.
  4. A clicking noise from the engine bay followed immediately by the jeep starting normally

A check of the trouble codes that the computer stores and returns to you by flashing the CEL with the key trick won't help you with this problem. The whole reason for that is a) symptom number 1; no CEL means no codes either while the problem is on going, and b) even after you get the jeep started, the computer only sees this as a normal start once it gets power, as if you had just turned the key on right then instead of 10 minutes ago. It doesn't realize there's a problem so there's no code stored.

Bottom line is, if you don't have these symptoms (the clicking noise is the ONLY one that is optional) or are dealing with a pre 91 vehicle, then what follows isn't going to apply to you. Sorry but you have other issues that I'm not covering here, like faulty fuel pumps, bad coils, burnt fusible links and whole slew of other possible choices. And for the earlier vehicles, while it certainly could be your computer causing the trouble, all of the tests below with the exception of the ignition switch test, just don't apply to your vehicle. You don't have the same components.

What's REALLY Happening

The whole problem here is that the computer isn't powering up like it should when you first turn the key on. The reason the jeep still cranks is that the starter motor circuit is totally separate from the computer. Without the computer though there's no fuel and no spark, so the engine merely spins without starting. When the key is first turned to the ON position before going to START they computer comes on to prime the fuel pump, send power to the ignition coil for the spark plugs, and turns the CEL light on for the diagnostic check so you know the light is still good (and a few other things that don't apply here). The clicking noise you hear is the auto shutdown (ASD) and the fuel pump relays coming on and switching rapidly as the computer tries to power up. As soon as the clicking stops, it means the computer is now ready to go, so you see the CEL come on and the pump start to prime itself. With the computer now powering the ignition coil you get your spark and your engine will now fire.

Tracing the Problem

So, we know that the computer isn't coming on. What we have to do now is figure out why. With the jeep it's not just a matter of "Oh the computer doesn't come on, it must be bad." There's several tests you want to run before you rush out to drop $200+ on rebuilding your computer only to find out that it wasn't the problem.

You're going to need some form of electrical testing device for this part. Personally I like one of these small multimeters that you can get just about anywhere these days. I'm not talking about a $200 Fluke meter here, just a small analog meter. I mean come on, even Wal-mart has them in the car department. They're not more that $20 and they'll make tracking down electrical gremlins so much easier than a test light. But if you really want to you can make do with a simple 12V test light to test for power, and a battery powered continuity tester with a light or a buzzer.

Since the meter is what I prefer, it's what I'm going to use. And since this article is geared to everyone, non-electrical guru's included, I should probably mention just how to test with a meter. (those of you who already know how to read one or are using a test light, skip these 2 paragraphs). Now if I tell you to test for 12V, set your meter to measure on the DC scale (AC is for household current). Some meters are scaled differently so pick the one that will let you read the smallest scale to still read 12V without being too small. For example, if you have a meter that reads on 10, 50 and 200V scales then you would want the 50V one. 10 is too small, and 200, while still giving you a reading, would only give you a very small change in needle position making it hard to tell if you're getting 7 volts or the full 12. A digital meter will usually still give you small readings even on the larger scales, but use the closest one just to be sure. Once you have your scale, take your black/negative probe and clip/touch it to a good ground. For me, I usually keep a long jumper wire with alligator clips on both ends and use that to jump the lead right back to the negative terminal on the battery. This eliminates any chance of getting a false reading because of a bad ground (ie trying to connect through paint on the body, rust on a bolt, etc). If you don't have a good ground you could see the test as indication no power and thus end up chasing a problem that you don't really have. Doing it this way also means I only have to hold on to one probe, leaving the other hand free to hold whatever it is I'm testing. Once you have your ground hooked to the meter, simply probe the wire in question with the red/positive probe and watch your meter for a signal, usually 12 volts. If you get a reading on the meter that is less than specified, it could point to a bad connection somewhere or another problem (sensor voltage can actually be a range from 1 to 5 or 1 to 8 volts but for the tests that follow 12V will be the norm). Check your probe connections and then repeat the test to confirm the reading and go from there.

Checking for continuity or lack there of is also pretty simple and again you may want/need to make up a jumper wire depending on where you need to test. For example you may need to test between something in the cab and the engine bay. Your probe leads probably won't reach by themselves so you need to jump with a longer wire. The good news is continuity usually works either way when it comes to polarity so you can jump either the positive probe or the negative and still have it work. The only thing you really want to do to be on the safe side is to disconnect the negative battery cable from the battery to shut down all power. Otherwise you could risk sending power through the meter when it doesn't want to see any power, which could damage it. A good example of this would be testing a wire to a light bulb to see if the wire is broken while not realizing it's got power. The light won't work with a broken wire so you don't realize it's powered, but as soon as you put your meter in line you complete the circuit and the power will flow; except it flows through the meter instead of the wire. So be sure to disconnect that cable before testing continuity. Just remember to hook it back up when it comes time to test for 12V power again or you'll be scratching your head. To actually do the test, set your dial to Ohms/resistance/continuity and simply touch one lead from the meter to your starting point, and the other to your ending power and watch you meter for whatever it does when there is continuity. Usually on the analog meters you'll see the needle sweep across to the other side indicating there is continuity but yours might have a buzzer to tell you without having to look. On a digital meter, you're looking for a reading between 0 and 10. Much more than 10 ohms can mean that there is a corroded connection between your probes somewhere. A null reading means no continuity which is a broken/loose wire. Ok, now that we know how to read the meter, on with the testing.

Test A

First order of business is to check your fuses! Now by check I mean with a meter. Don't just look at them, put your continuity tester across the blades and make sure they're good that way. I've seen more than one bad fuse that didn't burn the element out when it blew so a visual inspection isn't fool proof. You'll save yourself a lot of hair pulling later in the game by checking the fuses with a meter now. The first fuse in question is under the hood in the Power Distribution Center (PDC) right next to the battery. Open the cover and look at the diagram on the under side of it. Find the one that's labeled as the Fuel System/ECU. For those of you who don't have a cover, go get one to keep the elements out, 😛. But until then if you look at the box with the relays running across the top (those little black boxes) and the fuses under them, the fuse in question should be all the way to the right. It should be a green 30A fuse. Test that one. If this one is blown though, you probably won't be seeing the problems listed above since your jeep won't start at all. But you could have some corrosion building on this fuse so check it anyway while it's out and also check for any white/green powder like substance on it and also the terminals inside the fuse box. Also, take a tooth pick and see if the terminals wiggle easily inside the box. If they do it could indicate a broken/loose terminal that could cause intermittent problems. All good? Ok, time to get in cab. Leave that cover off because you might be coming back in a minute. Under the dash is your other accessory fuse box. No cover here, just look up above your left foot on the driver side and you'll see it. You should also see that each fuse is labeled beneath the fuse itself. Now here's where my info gets a little fuzzy. Some of the diagrams I've seen say that the "Dome" fuse powers the computer, some say that the early models were powered of the "Ign-Lps." They also say that the 91 and 92 models don't have a Dome fuse but my 92 does so who knows which one is right. But for the sake of thoroughness, test them both. The consensus on the boards seems to be the "Dome" fuse is usually the culprit here. First, while you have these two fuses out, get your meter and test the terminals that they bridge for 12V when the key is turned to ON. Probe one terminal and then they other. One of them should be hot with the key on. If not, your problem probably isn't the fuse but more on that in a minute. In any case, test both fuses with the meter and again look for signs of corrosion and try the tooth pick again to test for loose terminals. If you find that the fuse is loose, or even if you just want to be sure, take a pair of needle nose pliers and bend the blades of the fuse just slightly so that they're angled in opposite directions. Not too much or else you won't get the fuse back in. This will help insure that the fuse makes a tight connection with the terminals and should eliminate any problem with a loose fuse. In an extreme case, it may be necessary to replace the fuse box, or at least rewire the connections to bypass the box with an inline fuse. In any case, if your fuses all check out, and nothing seems amiss there, it's time to dig deeper.

Test B

Remember that 12V test on the fuses before? Well if you found that you have 12V then you don't need this test. Skip to test C. If not, then there's a good chance that you have a faulty ignition switch. First though, go back under the hood and locate the Ignition System fuse. Counting in from the right side (using the same orientation as before, meaning the first fuse you count is the one we just tested before) it will be number 4. It should be an orange 40 amp fuse. Test that one and again look for corrosion and loose terminals. If it's blown the ignition switch won't have power and thus neither will your under dash fuse box. If the fuse checks out, locate the ignition switch itself at the base of the steering column. It's not the same as the lock tumbler where you put the key. The actual switch is connected to the key with a rod and is a narrow, usually white rectangular affair with about six to eight wires coming off it. The two wires in question here are the heavy gage (thick) red wire and the yellow wire. First, test the red wire for 12V. Get a sharp, fairly heavy sewing needle and stick it through the wires insulation to test it without stripping off the insulation or pulling the harness apart (your probes might also be sharp enough to push through). Once the needle is in, just touch your probe to it being careful not to ground the needle on anything while it's in the wire. This wire should be hot (12V) all the time, with or without the key on. If it doesn't, you've got a break between the orange fuse we just tested and the switch. Verify this by testing for continuity between the fuses terminals and your red wire where you tapped it near the switch (jumper wire time and be sure to disco that negative cable like I said before). I'm not sure which of the two terminals will show continuity, but one of them should. If you do have 12V don't bother with the continuity test, simply move on to the yellow wire. With the key on this time, test for 12V again on the yellow wire. No voltage on this one but a red wire that passed its own test indicates a bad ignition switch. If you do have voltage here, but failed the 12V test in Test A then you want to test for continuity between the yellow wire tap, and the terminals of the fuses in Test A. Again, one of those terminals should show continuity to your tap if all is well. If not, you've got a good switch but a bad wire between the fuse box and the switch. Time to start tracing your wires to find the break.

Test C

Part 1

If you've made it this far, congrats, it's starting to look like you're going to need a new computer after all. Yippee, right? Anyway, for this test you need to pull the harness of the computer itself, which is located behind the washer fluid bottle on the firewall. Not that big clump of wires you see when you lift the hood, that's the bulkhead connecter. The computer is down below that. You'll probably have to pull the bottle in order to get clearance to pull the connector off. Once you have the connector loose, you'll see that it has a shoulder on one side, and it a smooth line on the other. The terminals themselves are arranged in 3 lines of twenty, separated evenly in the middle by the retaining bolt so there are 6 banks of ten terminals for a total of 60 pins. Now, it's been a while since I had one of these off, but if I remember right, with that shoulder positioned in the upper right corner when looking at the terminals themselves, then pin 1 is in the upper left corner. Maybe someone who's done this more recently can confirm or correct that but our first test should tell us one way or another. Counting to the right from pin 1, pin 20 is all the way to the right, with the bolt hole separating pins 10 and 11, and then it wraps back to 21 all the way on the left under pin 1. That pattern continues all the way through to pin 60 in the lower right corner. Now, the pins we're interested in are pins 3, 9, 11 and 12. Respectively they are: main power from the battery, ignition sense, and main grounds. First test pin 3 for 12V. This should be hot all the time. If for some reason you don't get 12V on what I'm calling pin 3, then count (by my orientation) to pin 18 and try again. Then repeat for pins 43 and 58 (again with my orientation). This will make sure that I'm not orienting wrong. None of those other pins will give you a 12V signal without the engine running (which it won't without the computer hooked up 👍) so if you see 12V on any of them then my orientation is off and we'll need to adjust. If you get no signal from any of them, then we just found out why the computer isn't working. It's not getting a steady power feed! Since we already tested the green fuse in the PDC in Test A we know the power is getting through, so we must have a bad connection in the wiring. Verify this by testing for continuity between pin 3 and the fuse terminals in the PDC (the green one again). With no 12V signal you shouldn't get continuity. If that's the case, trace your wires to find the break. The trick here though is that this wire doesn't go directly back to the fuse. Instead it goes to either the fuel pump relay or the ASD relay which get their main power from that same fuse (hence the Fuel System designation). So don't be confused when your wire doesn't go direct to the fuse. A bad connection at the relay could be the culprit as well.

(Test C) Part 2

If pin 3 checks out, test pins 11 and 12 for continuity to ground (btw, if you're going back to the battery for this test, clip to the removed battery cable end, not the terminal itself). You should see continuity on both pins. If not, trace the wires back and you should find a break or a loose connection. If memory serves they both go back to the engine ground behind the distributor. While you're there, test that ground terminal with the end of the battery cable. You could have a faulty connection back to the battery from the block ground that is causing problems.

(Test C) Part 3

This is it. If you've made it this far then there's just one more test to determine once and for all if your computer is the culprit or not. From the battery, power to pin 9 flows to fuse F4 in the PDC (the orange one), through the ignition switch, to the Dome (or the Ign fuse according to my diagrams) and finally to the computer at this pin. That's why you've been testing all those other things first. So, with the key on, test pin 9 for 12V. This is the one that the computer actually uses to see if the key is turned on to know if you're trying to start the jeep or not. Without power here then the computer just sits there doing whatever it is that computers do when they're not needed; playing ball with the kids, making time with the wife, sending out spam, that sort of thing. So the lack of 12V here with the key on, and having passed every other test, means that you've got a break between the under dash fuse and the computer. Turn the key off and test for continuity back to the Dome or Ign-Lps fuse and this pin. With no 12V signal you shouldn't see any. If you do, somehow the laws of physics have ceased to exist in your jeep and you should call in a priest for an exorcism as it is probably possessed (or else you somehow screwed up one of the earlier tests). Trace your wire back to the box to find the break or the bad connection. You may need to pull the back off the box to get a look at the terminals to find the break. But in the end that's what your problem is in this case.

Now, if you've passed every test listed so far, then congratulations, you're screwed. At this point there's no other thing that can be causing this problem except for the computer itself.

What To Do

Well, it's decision time here. You've got about 3 options that I'm aware of.

  1. Pretty obvious here, get a new computer. Usually though it's not a "new" computer but rather, you send yours out to get refurbished. Most parts stores will offer this service. You bring them the old one, they send it out, you wait 2 weeks, you pick it up and reinstall it and you're good to go for another 15 years (hopefully). Expect to pay $200-300 or more for this service if I remember right.
  2. Get a used one. This one has it's pros and cons. On the plus side, a used one is usually cheaper than getting yours rebuilt or buying a brand new one if you can find it. The down side is you may or may not get a guarantee on it (bought from a private owner parting out a jeep) or it may be hard to find one (jeeps being quite popular and not hanging around very long in junk yards). Also, you may only get a year or two before you new to you computer starts doing the same thing and you're back to square one. If you're lucky enough to have a 4.0L engine, you can probably find one from a Cherokee of similar vintage fairly easily from a local junk yard. The 2.5L XJ's though are harder to come by but they are out there. In any case, I would at least call around and maybe look on Ebay or the jeep bulletin boards' classifieds to see what's available before making a decision. Expect to pay around $50-100 or more for this depending on where you buy it.
  3. The last option available is repairing the computer yourself. Now, personally I've never tried this myself, however I've seen at least one write up on how to do it on the web somewhere. If anyone has a link or has done it themselves and wants to do a write up I'll be happy to include it in this one. But anyway the heart of this problem is that the capacitors inside the computer break down over time and fail. If you can solder to a circuit board then theoretically you should be able to replace said capacitors and be on your merry little way, having saved yourself a good chunk of cash. Caps are only like a buck or two at radio shack. The big deciding factor here is how comfortable you feel tearing into the computer's internals. I've never heard of anyone trying this and failing only to send out their computer to have it rebuilt, so I don't know if the rebuilders will accept your unit as a valid unit if you try this. You might want to call ahead first to confirm. The other thing about this is that I've heard conflicting information about what the actual cause of the computer failure is. Besides the caps failing, I've also heard that the circuit board itself will warp, either cracking or pulling away from terminals causing an intermittent connection, particularly in the cold weather when things start shrinking from the temperature. If this is the case, no amount of soldering is really going to help. In any case I suppose it can't hurt much to open the case and have a look at things to see if you can identify the problem at least. From what I've read, leaking capacitors leave a tell tale of electrolyte on the circuit board if they're bad, so if you see that, then you might be more inclined to try replacing them than if you don't find any obvious sign of failure.

UPDATE: For those of you who want to attempt changing the capacitors yourself, here's an excellent write up on doing so: http://www.jeepforum.com/forum/f12/crash-course-brain-surgery-replacing-capacitors-ecu-492597/

So there you have it. Three (really 5) relatively simple tests to determine once and for all if your computer is going on you without having to buy an expensive tester or take it to a shop or spend hundreds of dollars throwing sensors and fuel pump and all manner of other parts at it hoping to get lucky. Just remember, if you're not getting those first 3 symptoms then this isn't your problem, save yourself some time and look elsewhere. If you've got no spark but the pump is priming, the computer is on. No pump but a check engine light is on, bad pump or bad relay. It's only a combination of all 3 that points directly to the computer not powering up. If you guys think I missed something, or if you've got questions, post 'em up!

(Whoo hoo! 23,626 Characters! 😃)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment