RTD P1000 wire was picking noise and is giving unsuitable reading

1: avoid noise:

• use a clean power source (a battery is usually very low noise)
• use a front-end that's designed for precision measurement, and fit for this purpose

2: filter noise

• find the nature of the noise. Frequency, signal noise ratio, is it drift or noise?
• it may be possible to filter out known noise with sampling / acquisition techniques, analogue and digital filtering...

Overall advice: the front end has to be good, and the same for the ADC part. Everything between the sensor, and having its value in a variable in your µcontroller.
Investigate RTD front-end designs from IC manufacturers. They have app notes.

You can also just use a running average of a bunch of measurements.  This "technique" can be useful if you are looking for a relative shift in temperature more so than an accurate temperature reading.  The noise is in both readings so the difference between two running average measurements should be reasonably accurate.

Even if you are looking for a single reading, the average of the readings compared to a simple thermometer will reveal if there is an offset - that could be compensated for with a correction factor or if the temperature reading is close enough to be acceptable.  Remember, just because the microcontroller's analog to digital converter gives you a value with a bunch of decimal places, it doesn't mean that the value is accurate or precise.

You don't give enough context.

How are  you using the RTD - 2 wire, 3 wire or 4 wire connection.

How are you conditioning and measuring the signal (description and schematics, maybe software explanation as well).

What is the environment (thermal, electrical  etc) and how long are the cables.

What noise level are you looking for in your system (in equivalent degrees K)

Are there any cost constraints.

MK

RTD- 2 wire

measurement - thermal.

used moving average but still not reliable. looking for something better filter.

You have difficulty with solving problems because you do not have a good scientific approach.

First you must define the problem in a complete and an-ambiguous way.

It is IMPOSSIBLE to solve an inadequately defined problem.

I asked you several questions about your noise problem and you have not answered even one of them.

I asked:

How are you conditioning and measuring the signal (description and schematics, maybe software explanation as well).

You replied:

used moving average but still not reliable. looking for something better filter.

But this is not an adequate description

I was expecting something like:

Buffering and amplifying the signal with (schematic with values) having a gain of  X and a bandwidth of (Y)

Sampling with a  12 bit ADC (type specified here) at Z Hz.

Then processed by moving average filter averaging over N samples.

You also need to specify the noise that is causing you problems:

The data is extracted from the filter every Mth sample and the peak to peak noise of the extracted filtered data is P samples.

This is equivalent to K degrees.

You will need to use a scope or similar instrument to measure the spectrum and amplitude of the noise at the input of the ADC.

If you present your problem as I've outlined above I will be able to help you and happy to do so.

If you won't make the effort then neither will I.

MK