How does an air inlet temperature sensor function?

[Silvic]

I thought this would be a good place to post this. There is such a large community of knowledgeable folks that maybe someone can help.
A bit long winded but here goes:

I have a piece of equipment (CTL) that has a Perkins engine in it. There is a gauge (LCD style) that displays the inlet air temp. This is not the manifold temp but the inlet to the turbo temperature sensor.

Recently it began a intermittent issue in which the inlet air temp starts dropping moderately fast when the engine is at full throttle. In my application this engine is at full throttle almost full time (except warm ups and cool-downs.) Normally the temperature tracks engine load and throttle directly. When this "drop in temperature" occurs the ECM commands more fuel as it senses that it needs the fuel due to the temp dropping. This results in an unburned fuel smell. If I idle down while it is still dropping it drops below what would be normal for a full power idle down and if the throttle is then increased the temp starts decreasing again. Almost like it is reverse sensing temperature at this point. Once I turn the engine off and let it sit for 10 seconds and restart it, everything reads and runs normal until the next occurrence.

I have talked to the Perkins rep and the service manager of the dealership who happens to be 3 and 5 hours away respectively and they are at a loss as well. No codes are being thrown and that prevents any trouble shooting with codes. There is a Perkins software package that can run some tests but the Perkins facility is 3 hours away and the software is about $5K so I am rather reticent to purchase it. (not certain they would even sell it)

Don't really want to shotgun a temp sensor at the problem ($140). Does anyone know just how an air inlet temperature sensor functions and is it possible with some intermittent malfunction that it could start reading inverse temp/voltage. The Sensor has a 5 volt energizing signal from the ECM but with only two wires I don't know haw it sends data back, unless it plays a plus/minus game with the 5 volts and that is converted to a digital temp reading.

Not only is it annoying, I and the service manager have a concern that long term it will ruin the DPF due to the unburned fuel that is suspected of being the source of the smell.

 

[RandyT]

I would say you have 5 volts input and as the sensor heats up and cools down that voltage gets reduced which is interpreted by the ECM as a temperature. The sensor may be going bad and causing erratic resistance readings which the ECM interprets as a temperature. It won't set a code unless what is received is outside of it programmed range. A graphing meter connected to the return/sensing wire to the ECM may be able to show what the sensor is doing and you could watch the voltage changes in real time and if it is the sensor could what it go wonky in real time.

 

[Bavarian]

I understand your pain on a new sensor. I do agree to the previous poster. I would check the resistance of the sensor before you run the engine and after it acted up and put the engine on the rich side. Assuming the ambient temps don't change much, it should show almost the same resistance.
There is an equal chance that the wiring harness is starting to short out and causes a faulty reading.

My two kids, both of them got a used Mini Cooper where the ambient temp sensor went south. On both, the display showed -40. There is a tendecy that the reading shows colder temps than it actually is.

 

[Silvic]

I would say you have 5 volts input and as the sensor heats up and cools down that voltage gets reduced which is interpreted by the ECM as a temperature. The sensor may be going bad and causing erratic resistance readings which the ECM interprets as a temperature. It won't set a code unless what is received is outside of it programmed range. A graphing meter connected to the return/sensing wire to the ECM may be able to show what the sensor is doing and you could watch the voltage changes in real time and if it is the sensor could what it go wonky in real time.

In your opinion and understanding, could the sensor going bad actually flip the resistance that the unit shows thus causing the Apparent "reverse" temp sensing? There are two types (PTC and NTC) but for this next question it may not matter.

If higher temperature reduces the resistance normally for the given style (NTC) can a failing sensor cause a higher temperature to increase resistance instead?
And if that is the case why does a shutdown of the engine and then restart "fix" the problem, at least for a while.

Getting a probe into the area during operation with the intermittent and unpredictable period between "failures" is going to be a problem with the location and the type of work (mulching) that the machine is used for. Problem never surfaces during non active time (at least not yet). If it was a hard fail it woulds be so much easier to deal with. The "fixing" of the problem with the de-powering as the unit is turned off makes it even more challenging.

 

[dodge man]

My guess is the sensor is bad but it could be something else, a wire or the way the computer is reading it. What’s cheaper getting it to the Perkins guy or getting a new sensor?

 

[radios1]

it could be the "hot wire" type of sensor, where the "hot wire" is in the airflow, and it's resistance changes with temperature(hotter temperatures increase the resistance, and colder temperatures decrease the resistance). it's usually a nichrome wire.. I've seen these go bad on cars and trucks before.. also, you need to see if it's an ECM problem!..

 

[dodge man]

A short story, we had a Chrysler 300c with a bad egr valve, but I didn’t know what it was. I had the shop manual and it seemed like there was about 20 pages to pin the problem down and scan tools I didn’t have. A new egr valve was $35 at Autozone. I hate to throw parts at something but if you are doing the work it’s often the cheapest.

 

[BleedingOrange]

Is there any possible way to clean up the outside of the sensor ? Any chance it is coating with something that might be affecting the sensor being able to read the true temperature ?

 

[radios1]

Is there any possible way to clean up the outside of the sensor ? Any chance it is coating with something that might be affecting the sensor being able to read the true temperature ?

possibly.. maybe carb cleaner or sandpaper.. if it's the hot wire type, the end connections may have a bad connection, also.. finding the standard resistance values for that sensor, then comparing them to what you have is key..

 

[s219]

it could be the "hot wire" type of sensor, where the "hot wire" is in the airflow, and it's resistance changes with temperature(hotter temperatures increase the resistance, and colder temperatures decrease the resistance). it's usually a nichrome wire.. I've seen these go bad on cars and trucks before.. also, you need to see if it's an ECM problem!..

Agree!

 

[Tinhack]

it could be the "hot wire" type of sensor, where the "hot wire" is in the airflow, and it's resistance changes with temperature(hotter temperatures increase the resistance, and colder temperatures decrease the resistance). it's usually a nichrome wire.. I've seen these go bad on cars and trucks before.. also, you need to see if it's an ECM problem!..

Ditto also.

There should be a visible coil or wire in a port (hole). On automobiles, I've cleaned similar sensors with brake cleaner on a Q-tip. Most times it works. The sensors get dirty even behind an air cleaner because they use a heated wire process. Be gentile cleaning. The wire can be brittle.

 

[skipmarcy]

The "hot wire" sensors you folks are describing are not air temp sensors, they are mass airflow sensors - measuring amount of air taken in. Automotive applications have a separate air temp sensor which is a thermistor type and a lot of times it is incorporated into the same unit as a MAF sensor to read ambient temp. The thermistor will drop that 5 v reference voltage according to temp and are usually extremely sensitive. Anything else that changes the voltage the computer reads on that circuit can affect that reading. Automotive uses that 5 v ref on throttle position sensors, egr position sensors, coolant temp sensors etc. and any problems in those circuits can have an effect to the computer. Throwing parts at it is the typical inexperienced approach, proper diagnosis is the right way. With automotive I would use my scanner to actually read the temp/voltage the computer is seeing, see if the 5 v ref voltage is correct and all the other pertinent info is correct. I might use my dvom to confirm voltage readings at the proper pins at the computer and then trace back towards the individual components to see where a problem might be in the wiring, connectors or component.

 

[dodge man]

The OP said the software he would need is 5k. Hauling it there and back and paying to have it fixed is probably going to cost a lot more than the sensor. I agree throwing parts at it isn’t the best choice but my point is it might be the cheapest.

 

[Silvic]

I appreciate the responses.
I am still bugged by the fact that the problem has repeatedly gone away on a power/key cycle. I do not understand how a bad wire or bad sensor could be "fixed" albeit temporarily by what is affectedly a reboot unless it is the ECM. But if it was the ECM I would think I would see other issues and codes would be thrown.

Yesterday I fought with it for about 30 minutes in which it did not reset with the power cycle and seemed to be a hard fail. Probably at least 8-10 times the power cycle had no effect. Decided that I finally had a hard fail and maybe I could finally get somewhere. After sitting for 5-10 minutes I would power up and the inlet sensor saw 150 degrees (way higher than normal for sitting that long) As the engine started the temperature in idle started to drop rapidly to 100-105 degrees. Definitely a reverse reading and I decided in that permanent fail mode I would pull the sensor and take it home and run some test with a multi-meter. Being in the field I started it to move closer to the road and the work truck. It started fine and worked normally for the next 3 hours without any oddities. Heavy rain forecast for the next several days so I decided not to pull the sensor as I did not have any good method of sealing the open port with me in the truck so will wait to pull the sensor.

I agree with Dave right now. I am leaning to throwing a new sensor at it with all the factors of distance and s/w costs that are involved. I am starting to debate on purchasing the test equipment/software as it may pay for itself in lost time at this point.

Just hoping that it is not the ECM.

 

[finn1]

I致e been away from calibration for almost 12 years, but here痴 a little input.

First, Diesel engine痴 always run lean. The governor controls fuel injector pulse width so intake manifold temperature has nothing to do with air /fuel ratio. It痴 simply commanded fuel delivery and turbo boost level , which is a function of fuel delivered, wastegate position, and engine speed. Temperature is also used to increase pulse width, or modify pilot injection profiles for cold start, until the governor take over that function.

One thing the ambient air temperature sensor may be compensating for is the injection timing requirements at low temperatures, ie cold starting and cleanup conditions call for advancing timing to compensate for increased ignition delay at cold conditions. Most calibrations use engine oil temperature, though, or even coolant temperature, as diesel combustion tracks better to those than air temperature during operation.


Not sure if Perkins uses ambient air readings for egr control, but they may have a strategy to modify egr control valve pulse width at extremely low temperatures, so a faulty sensor could impact that.

Best bet is to monitor sensor voltage in operation to verify that there is, in fact a voltage change that corresponds to the drop in gauge reading.

As to the cost of scanners and software from Perkins, there are generic scanners out there for the heavy duty market. They are still not cheap, and won稚 do everything the factory scanner does, but a good one can be found for $300-1200.

 

[ericm979]

Does the machine have an OBD2 port? I've diagnosed similar issues that did not throw codes using an OBD2 reader set to display the values from the sensor in question and monitoring the outputs. A generic reader may be able to read the sensors values as you operate the machine.

 

[Silvic]

Does the machine have an OBD2 port? I've diagnosed similar issues that did not throw codes using an OBD2 reader set to display the values from the sensor in question and monitoring the outputs. A generic reader may be able to read the sensors values as you operate the machine.

OBD2 port : no
J1939 port to read with proprietary software and hardware :Yes

 

[def38]

If the sensor is a thermistor (they are popular for automotive applications and they are cheap) there is a self heat phenomena that may be causing your problem. This means there is some failure in that circuit that causes the thermistor to self heat or there is an unwanted resistance somewhere. Thermistors are typically easy to check with a quality VOM. Many thermistors are in the 1K Ohm to 100K Ohm range and zero power resistance is often are rated at 25C. We used thousands of them and did a quick QC check with a VOM at room temperature. Ours were from Betatherm. Their Curves are on the internet.