Failing thermistor ?

[axero]

I have a refrigerator that doesn't cool properly. Inside is a sensor of type M2020/5. A datasheet for this sensor is available here:

http://www.mouser.com/ds/2/400/NTC_Probe_ass_M2020-595804.pdf

So when I measure a temperature of 12°C near the location of the probe inside the fridge, I can measure a resistance of about 11.66 k\$\Omega\$ over the probe. According to the datasheet above and some number crunching using Python, I get that the temperature at that resistance should be 7°C. This is the temperature shown in the display on the fridge as well.

So the control unit is interpreting the probe properly but it seems like something's wrong with the probe. Is it common with this type of failure of the probe or am I missing something? The fridge operates by blowing cold air through a fan rather than cooling directly with the evaporator.

[joseph nicholas]

I have this type of refrigerator and I needed to replace this mechanical part along with the microcontroller.  They can be checked by freezing it and warming it gradually.  Not difficult to determine if it needs replacing.  You should also test the heating element as well.  There are alot of repair videos about this on youtube.  Good luck.

[Armadillo]

You will need to know the thermal capacity and response time of your hand held probe and allow it to stabilize inside the fridge before making the comparison. No idea what kind of probe you used, whether is insertion probe, contact probe or air probe? Read the manual.

[Seekonk]

I don't see why you would want to be at either of those two temperatures. I turned a 7CF freezer into a fridge at my camp. Even with the sensors within an inch of each other I get wide swings of temperature every time the door is opened for just a short period. Very close to the sensors I have a fan that runs all the time to prevent stratification.  I'm used to a world of precise numbers and these two sensors hardly ever match. Without further information I would have to think thermal response and location.  Have you sensed other locations in the fridge?

[axero]

The fridge has a glass door and is designed to maintain temperatures within the range of 8-18°C. The glass door means that I don't have to open the door to see the temperature as shown by the thermometer inside.

If I set the desired temperature to 12°C, then the temperature inside will be measured to 18°C, if I set it to 8°C and then see 7°C on the display, then the measured temperature is 12°C. I measured as close to the probe as possible. At other points the temperature is even higher.

When I measured the 11.66 k\$\Omega\$ resistance, the reading was pretty stable, i.e. it wasn't fluctuating.

Is it common that a thermistor fails by an offset like this?

[Armadillo]

Why don't you immerse the thermistor into a glass of ice with water, wait for at least 4 Tau = 4 X 35sec, say approximately 4 minutes to see if the fridge readout shows close to 0 deg.C or [16330 ohms]. If it still deviate few deg.C somewhat, then replace the NTC thermistor.

Or alternatively, put a thermometer into a glass of water, then put it into the cooler for at least a day, then after readout the thermometer to be close to the set temperature. This method takes longer time but sure represents the temperature of the content being cooled.

Also sight the cooler away from any radiant heat.

[joseph nicholas]

I looked at the data sheet you posted.  It tells you everything except what type of probe it is an how it actually works.  This may seem like a dumb solution but just use a completely different sensor.  The kind with a bi-metalic strip that clicks.  Do this if the microcontroller will allow it.  Please tell us more about how the microcontroller works, if you can.