How to design a filter for temperature measurement?

[baicy]

Hi, guys,

I measure the temperature by PT100. The signal is collected by PICO scope. The measurements are alright except the signiicant noise, which leads the ambient temperature vary between 20 and 20.2 degree, although the actual temperature remains quite stable. How to select a suitable filter to reduce the electrical noise?  Please find the VI figure in the attachment.

Many thanks,

Alan

[hrh1818]

On Jun 11, 8:40 am, baicy <x...@no.email> wrote:
> Hi, guys, I measure the temperature by PT100. The signal is collected by PICO scope. The measurements are alright except the signiicant noise, which leads the ambient temperature vary between 20 and 20.2 degree, although the actual temperature remains quite stable. How to select a suitable filter to reduce the electrical noise?&nbsp; Please find the VI figure in the attachment. Many thanks,Alan
>
> Temp.jpg:http://forums.ni.com/attachments/ni/170/330688/1/Temp.jpg

The simplest approach is to use a moving average. Say average the last
5 or ten readings. The more readings you average the better you noise
rejection will be. However, if you go overboard and average a lot of
readings you will have a considerable delay before you can detect a
significant change in temperature. It is up to you to make the trade
off between response time and noise rejection. Other types of
averages you may want to consider are given in the following article.
http://en.wikipedia.org/wiki/Moving_average

Howard

[baicy]

Thank you for you replay,Howard
<>

I think the problem is the scope I used, which is PICO PT 104. The noise is random, even after average the results for 60 seconds, there is a considerable noise in the measurement as present in the attachments.  Any filter in the Labview I can use to reduce the noise. Many thanks.

Alan

[Ray.R]

That's very noisy for temperature measurements.

If you want accurate results, you should look into signal conditionning or why the signal is so noisy. 

Trying to remove the noise by software is like using a bad bandaid.  The software will not know how to distuinguish an accurate reading from a noisy one.  Also, the signal should be calibrated so that you can rely on the measurements.

Try to improve the measurement environment first (hardware).

RayR

[baicy]

Thank you RayR,

Any suggestion how to measure the temperature accuratly? What I get is the PT 100 probe with 4 wires connection. Any NI instrument recommended to measure the PT100.

Thanks,

Alan

[Ray.R]

There are many reasons why you can have a noisy signal.  Signal conditioning may or may not be the answer for you, but it is the first place to look.  Typically, temperature measurements should not be that noisy. 

What is your voltage range?  Is your DAQ setup for that range (if input level is programmable)?  What temperature (voltage) swings are you looking at? 

There is a lot of useful information on techniques for minimizing noise in the Field Wiring guide.

Signal conditioning offers a few different techniques for improving signal quality including amplification and filtering.  There are many resources available to learn about using signal conditioning for different applications including the following: Improved Signal Quality via Conditioning and Increase Your Measurement Accuracy by 10X with Signal Conditioning.

Are you using the PT 100 probe from Ohmega?  If so, this is what they have to say:

"Wiring for the PR-11-2-100-1/16-6-E (3-wire 385 Pt 100 ohm sensor element) Probe:
One RED & two BLACK wires; the RTD sensor element is across the RED wire & either of the BLACK leads; the two black wires are connected to the same point on the RTD sensor and used for lead wire resistance compensation. At room temp the RTD will measure approx 108 ohms across the RED & BLACK leads, and measure a short across the two black leads. Beyond this, I cannot help you with the connections to the National Instruments. However, an RTD is a resistive element (resistance is directly proportional to temp increase) and does not require COLD JUNCTION COMPENSATION like a thermocouple."

Message Edited by JoeLabView on 06-12-2008 08:24 AM