Display instability for temperatore sensor

[exwau]

I bought some 3-digit thermometers from China.
the problem is that the display is not stable and at times jumps from +/- .1 or .2 degrees twice or three times a second.
I added a capacitor to the sensor but that makes no difference.
The sensor is resistive with a negative coefficient (...).
Does anyone have a solution?
I would like the display to not change for at least 3-5 seconds.

 

[kellys_eye]

This is due to the "sampling time" of the processor that reads and displays the value.

You might be able to slow the processor down - depends on the way it's been designed. Got a schematic or a picture of the insides?

 

[Alec_t]

What is the power supply for the device? Perhaps the device is sensitive to supply variation.

 

[shrtrnd]

Good ideas posted above.
Here's one more to consider, a sample and hold circuit that clears itself every 3-5 seconds.
The problem I have with trying to figure-out what to do, is that I don't know your application or specifications.
+/- .1 or .2 degree fluctuation (depending on the accuracy of the '3-digit thermometers'), does not seem to me at first glance much of a problem. If this is a critical application, I wouldn't have bought Chinese thermometers.
For all I know, your environmental set-up might actually be causing a +/- .1 or .2 degree change.
I take it you're asking for a way to make the display LOOK stable. Depending on your application, I might
consider just going to a 2 digit display, or disabling that third least significant digit.
If you actually need to monitor +/- .1 or .2 degree change, that's a darned tight spec to try to maintain practically.

 

[timff]

One source of variation can be electrical noise on the supply line that a capacitor would reduce. Since you have tried that, it may be that since 0.1F is a fairly frequent change in most environments, the software does not make effective use of a "rolling average" that can smooth out readings. If this is the case and if your temperature display has the sensor integrated as a monolithic unit, you may still be able to smooth out the readings by applying an insulating material that can prevent rapid ambient change from being detected by the sensing element. In effect you are using some thermal mass to slow down changes to the sensing element.

 

[shrtrnd]

What timff is describing is what is called in the industry a 'thermal well'.
Basically a covering for the temperature sensor you're using itself. In industry it's used to isolate the actual
temperature element from direct exposure to any physical contact with whatever it is you're trying to read the temperature of.
Like I mentioned before, we don't know your specific application, if you're still here and can give us a little more
information. We don't know if you're trying to read ambient air temperature, or something else.