Cooling fans dont shut off

P0117 - Engine Coolant Temperature Circuit Low Input

Five volts is not a correct value. That will trigger the radiator fan relay and set a fault code for CTS voltage too high. I have a feeling you inadvertently set code 117 during the diagnosis because it disagrees with that 5.0 volts. Did you unplug the sensor to get that reading or did you back-probe through the connector next to the wire?

Does your relay look like the left one in the second drawing?

Definitely not supposed to have 5.0 volts. What you might have read is that circuit is indeed fed with 5.0 volts, as are most engine sensors, but with any two-wire temperature sensors, part of the circuitry is tucked away inside the Engine Computer. If you understand electrical theory, this will make sense, otherwise I apologize for making this sound complicated.

This is one big series circuit. It starts at the 5.0-volt supply inside the computer, goes through an internal "dropping" resistor, then it comes out on terminal 26 and goes to the sensor, which is another resistor, but this one is temperature-variable. From there it goes back to the computer on the ground return wire on terminal 13, along with some other sensor grounds, through some internal monitoring circuitry, then to ground after that.

The ground monitoring circuitry is pretty much irrelevant to this story, but explains why that ground wire doesn't just go to ground right on the engine.

That internal resistor is shown above terminal 26. The entire 5.0 volts has to be used up across those two resistors, the one in the computer and the one inside the sensor. As temperature rises, the resistance of the "thermistor" in the sensor goes down, so less and less of that 5.0 volts is dropped across it. There's another chapter to that story too, but there's no need to involve that here. As less of the 5.0 volts is dropped across the sensor, more is dropped across the internal resistor. Those two voltages will always add up to exactly 5.0 volts.

It's when there's a break in the circuit on either wire that current flow through the circuit stops. With no current flow through a resistor, there is no voltage dropped across it. Since we started in the computer with 5.0 volts, and nothing is dropped, or used up, across the internal resistor, we end up with the entire 5.0 volts right at terminal 26 and that yellow wire. That's the voltage that would be displayed on a scanner, and it is what the computer sees and bases its decisions on.

The only way possible to have 5.0 volts on the yellow wire is when there's a break in the circuit. We'd find 5.0 volts before that break and 0 volts after it. (Actually 0.2 volts). This is where even experienced mechanics make the fatal mistake of unplugging the sensor's connector so they can touch the voltmeter's probe to the terminal to take the reading. That just created the break in the circuit, then they're stumped when they find the full 5.0 volts there. Other types of sensors have their own strategies for putting the full 5.0 volts on their signal wires when they're unplugged. There has to be a way to force a bad condition so it will be detected and the appropriate fault code can be set.

Every sensor voltage has to be taken by back-probing through the back of the connector body alongside the wire, while it is plugged in. Any other condition is not valid.

The normal range of sensor voltages is from 0.5 to 4.5 volts. That's for training purposes. In actual practice, you might find one as low as 0.42 volts or as high as only 4.3 volts, but the important point is it can never reach 0.0 or 5.0 volts as those are the two that trigger the fault codes. Throttle position sensors do this with mechanical stops inside them that limit their range of signal voltage from idle to wide-open-throttle. Here again, the signal voltage can only go outside that 0.5 to 4.5 volt range if it is unplugged, or if there's a break in the circuit to include the physical connecting joints inside the sensor.

When you still find 5.0 volts on the yellow wire for the coolant temperature sensor, and while it is plugged in, there has to be a break after that point. By far the most common cause is a stretched terminal in the connector. Those are usually too mall to fit a small pick in there to squeeze them tighter. Instead, there's a colored locking wedge that has to be pried out on the exposed terminal end, then a plastic finger can be pried up to allow the terminal to be pulled out by the wire. Now you can squeeze the terminal a little so it will make a more solid contact. If that doesn't work or if the terminals are corroded, you can order a new connector from any auto parts store, or you can snip one off a similar car in the salvage yard.

Sensor ground wires don't seem to cause as much trouble, and in this case the coolant temperature sensor and the MAP sensor share a common ground. The CTS ground wire could have a break up to the splice, but anywhere after that, it would affect the MAP sensor too.

Your relay is the common GM relay. It threw me for a loop that it uses the same terminal designations that are found on Chrysler relays. Those are the ones in the drawing I posted. For now there's no reason to dig into the relay circuit. The Engine Computer is turning the radiator fan relay on because of that defective condition related to the coolant temperature sensor. That is the default condition because with an invalid signal voltage, the computer can't know engine coolant temperature. It runs the fan for safety just in case the engine gets too hot.

The service manuals always include a lot of resistance / continuity tests in their troubleshooting charts, but those have you do a whole bunch of little mind-numbing tests that can be skipped over. It is much for efficient to use the voltage tests to locate the cause of a problem, then you can use resistance tests to verify what you found or to isolate it to a small area.

4.99 volts is in effect, 5.0 volts, and the only way that can be at the sensor is there has to be a break after that point. The next step is to back-probe the black wire on the coolant temperature sensor. That one must have 0.2 volts. If you find 5.0 volts, the break is after that point.

Next, I'm still going on the assumption the MAP sensor is not setting a fault code, so that rules out a problem in the part of the ground wire both sensors share. That means the break has to be between the sensor, (blue arrow), and splice 186, (red arrow).

I just noticed another connector, C126, right below the blue arrow, in that ground circuit. They show it as terminal "A", which could mean it's a single-wire connector, or its the first one of multiple wires in that connector.

The second drawing shows where the two sensors are located. Logic would dictate the part of the ground wire with the break runs between them. The connector will be in that area too.

If you do find 0.2 volts on the CTS black wire, the break has to be between one of those pairs of mating terminals, or the sensor is open. Do a resistance test on the sensor between those two terminals. The exact value is irrelevant but expect to see anywhere from 1,000 ohms to perhaps as much as 30,000 ohms.

What's the voltage on the black wire at the sensor?

On older models there were two coolant temperature sensors. The sensor for the dash gauge always had just one wire. It was grounded through the engine block. Coolant temperature sensors for the Engine Computer have to be more precise. They also always have two wires. The ground wire goes through the Engine Computer before it gets to ground.

Between the late '90s and mid 2000s, they started using just the two-wire sensor for the Engine Computer. Typically engine sensor data is broadcast out on the data buss wires to all the other computers. The Body Computer uses some of that data, then sends it to the instrument cluster, which is another computer module. That means there's three computers involved in running the dash gauge.

I suspect you're letting the MAP sensor confuse you. That doesn't have anything to do with the coolant temperature sensor except that they share a common ground wire. That's no different than working on a dead tail light that shares a common ground with all the other lights on the car. The only thing the MAP sensor circuit is good for is to verify one small part of that ground circuit has to be okay.

By your voltage readings, the break has to be between the orange and red arrows in the diagram a few replies back. As far as the terminal designations, you'll see a "B" right at the top of the coolant temperature sensor, and an "A" right below it. Those signify there's two wires in that connector. C126 is a different connector, and they show the sensor ground wire as terminal "A", but that doesn't necessarily mean there's just one wire in that one. However, I looked it up, (drawing below), and I was surprised to see it's a two-wire connector. Then, after looking closer at the diagram, I see terminal "B" in C126 is right above the orange arrow. It looks like they added that short pigtail either to create easily-accessible test points, or because there was a change made in the placement of the sensor after the wiring harnesses had been manufactured, and the connector didn't quite reach.

It looks like the connector in your photo is C126.

Regardless, there has to be a break between the orange and red arrows. Starting at the top at terminal "B" in C126, (yellow wire), and working your way down to the splice at the red arrow, if you find 5.0 volts, the break is after that point. If you find 15 mv, the break is before that point. Remember, these voltages are only valid when all the connectors are plugged in and you are back-probing next to the wires. When the meter's probe is too fat to fit in there, I use a stick pin or sewing needle, then touch the probe to that. Also be aware if you get a reading of 0 volts, there's a good chance the probe isn't touching the terminal. I've gone down the wrong path many times from making that mistake.

You're making this way too complicated, although now I'm willing to add the sensor itself to the list of suspects. I made this drawing to show everything that is involved with the coolant temperature sensor. That is everything the Engine Computer uses to know engine temperature. Nothing more; nothing less.

284 degrees is the default reading it goes to when there's an open circuit, meaning a break in the circuit. As I mentioned early on, this is rarely caused by the sensor itself because there's just one resistor inside it, but Ford did have just that problem in the early '90s, and it was real common. It is possible this is what's happening to your sensor too, but it is just as likely there is an intermittent break between a pair of mating connector terminals.

When it goes to 284 degrees is when the signal voltage goes to 5.0 volts. What you might do now is when it does that, stop the engine, then turn the ignition switch back on and watch that reading over the next few minutes. As the engine cools down, if the reading suddenly drops to 110 degrees, and you're not disturbing anything under the hood, it is likely the sensor that is defective. If the same thing happens repeatedly, meaning the signal voltage always pops up from around 110 degrees to open circuit, that would also point to the sensor.

If the signal voltage changes at all different times, especially while you're flexing the wiring harness, that would point more to a connector terminal problem.

You don't have to tear the entire wiring harness apart. You already measured 5.0 volts, (4.99 volts), at some point along the yellow wire. Everything up to that point has to be okay. Continue working your way toward the sensor to find the first point that 5.0 volts is no longer there. That is when you crossed over the break.

To add to the story of the dash gauge, this drawing is a representation of six of the dozens of computer modules on your car. There is a pair of "data buss" wires connecting all of them. On some car models the data buss is a single wire. When it's a pair of wires, they'll always be twisted around each other, called a "twisted pair". Stray magnetic interference would induce voltages that would confuse the computers. By twisting the wires, that interference affects both wires equally, and the voltages induced cancel each other out.

Some models today use up to three separate data busses. There is so much information being passed along them that it would take too long to share critical or time-sensitive data. An example would be for the air bag system. The deployment of the air bag is timed to milliseconds, and there isn't time to waste until it's time for that computer to send and receive information.

In this story, the Engine Computer knows all the operating parameters of the engine, including load, rpm, fuel volume, throttle direction and rate of change, along with all the data from its sensors. It controls the cruise control servo, so it needs to know when the brake pedal is pressed, when steering wheel switches are pressed, and it needs to know road speed.

The Transmission Computer needs to know engine load and speed, and coolant temperature. It already knows road speed from its dedicated vehicle speed sensor.

All of these computers know a lot of data by themselves, then, rather than having duplicate sensors for each computer, they get some of their needed information from the other computers.

Each computer takes turns broadcasting its information out onto the data buss, then it can be seen by all the other computers. Starting with the Engine Computer, when it is its turn, it sends out sensor data and all engine operating conditions such as rpm, charging system voltage, and which relays and solenoids it has activated. Among that data is coolant temperature.

The Transmission Computer, along with all the others, looks at that data and picks out what it needs and ignores the rest. It needs to see throttle position and engine load along with the road speed it already knows, to calculate shift points. It needs to know coolant temperature to know when it is okay to engage the torque converter's lock-up clutch. It ignores information related to the knock sensor, ignition timing advance, and the number of milliseconds the injectors are being pulsed open.

A few milliseconds later, the Transmission Computer gets its turn at sending out data. It sends road speed, which gear it's in, whether torque converter lock-up has occurred, and data related to the amount of wear on the clutch plates.

Much of that data is ignored, but another computer can be added to the data buss. That is the scanner we use to read this data and to read diagnostic fault codes.

The Body Computer also is able to read all the information on the data buss, but it too just picks out what is needs and ignores the rest. It has its own sensors too to get information that it shares on the data buss. Fuel level is one of those. In this story, the Body Computer picks out the coolant temperature reading from the message it saw from the Engine Computer. It analyzes that digital signal, interprets it, then translates it into a different digital signal. When its turn comes to broadcast its information onto the data buss, most of it is ignored by the other computers, except for the instrument cluster. That is the only one capable of understanding the digital signal related to coolant temperature. Logic would dictate the instrument cluster could take that data right from the Engine Computer's message, but to do that for many dozens of gauges, warning lights, and other functions would make the cluster more complicated than it already is. Since that data is already being processed by the Body Computer, it is sent to the cluster, but in a modified form that will be ignored by the other computers, otherwise they would be getting two different coolant temperature messages from two different sources. The transfer of data from the Body Computer to the instrument cluster is not time-sensitive, meaning it doesn't matter if it shows up right now, or a fraction of a second from now. Compared to engine performance, or the critical timing of air bags and anti-lock brakes, Body Computer data can be sent on a low-level data buss of its own, or it can just wait until its chance to broadcast comes along on the higher-level data buss.

Once a packet of data is transmitted onto the data buss by the Body Computer, the instrument cluster takes that and analyzes it, then turns it into the voltages that run the gauges and warning lights. This is where the coolant temperature gauge responds to what the Engine Computer is seeing from the coolant temperature sensor.

All of this data sharing became more and more involved over the years. On older cars up to around the mid to late '90s, the instrument cluster wasn't such an unnecessarily-complicated computer module. That was when, instead of interpreting digital signals, they just used their own, very reliable circuits and sensors to run the gauges. That was when there was a second coolant temperature sensor just for the dash gauge, and it always had just one wire connected to it. Those gauges were never meant to be particularly accurate. Their main purpose was for the driver to notice when something was out-of-the-ordinary. One coolant temperature gauge could run at half-scale, and another one could run at 3/8, and both engines could be running perfectly.

Coolant temperature sensors for Engine Computers have to be very accurate, but with all the computer circuitry involved before the gauge's pointer is placed to the desired position, there are just too many variables to expect the gauge reading to be right on the money. The only way to know the exact coolant temperature, or any other sensor reading, is with the scanner. That takes the information from the data buss exactly as it was transmitted by the computer, and interprets that so we can read it.

Pursue the wrong voltage at the coolant temperature sensor. You can't expect the dash gauge to read correctly when the sensor circuit isn't working. GM has had a lot of trouble with their stepper motors that run their gauges, but the positions of the pointers are not monitored. The instrument cluster computer sends varying voltages and polarities to those motors, then it just assumes the pointers moved to their intended positions. If they don't, no computer knows that or responds to that.

The radiator fan being switched on is the second clue. That is also a result of the sensor circuit problem, but it has nothing to do with the dash gauges.