What is an electric isolator?

[houkensjtu]

Hi !
I need to measure a voltage singnal which is around 10mV(output of a temperature sensor, when applying a constant stimulate current (usually 10 micro Amp) to it) , and recently, I found the voltage sometimes shifting even when there is no temperature change.
So we think there maybe some kinds of "noise" during our measurement, but we have totally no idea what kind of noise it is and where did it come from.
Someone suggested the following product to us, it said in its datasheet that it can somehow block noise. After reading the datasheet myself I still don't understand WHAT KIND of noise it is talking about and HOW WELL could it block the "noise".

http://www.watanabe-electric.co.jp/en/product/show/productCode/854/

Also i also appreciate if you could help me to specify what KIND of noise we are facing. THX!

[RJFreeman]

Someone suggested the following product to us, it said in its data-sheet that it can somehow block noise.

That would be noise generated by ground loop issues or where your sensor needs to float at a different voltage than your measurement or control device.
That appears to be an isolator for 4-20mA current loop systems so is unlikely to be applicable to your application.
http://en.wikipedia.org/wiki/Current_loop

Also i also appreciate if you could help me to specify what KIND of noise we are facing. THX!

well that is the question, isn't it?
Unfortunately without more information it is unlikely anyone can guess, it sounds like you are using something like an LM335 but 10uA as the data sheet suggests >400uA is needed.
Also what are you measuring and where?
is it a noisy electrical environment?
are you running long leads to the sensor?

Sometimes the simplest fix is to apply a low pass filter - the trade off being that this will slow measurement times

[houkensjtu]

Thx for your help! I will read that wiki article. For now maybe I'd better to provide some more further information of my measurement.
As you can see in the attached pdf file is my naive measurement scheme. There are many other machines and devices including large cooling pump, inverter motor, air-conditioner and many else in our lab so I suppose the noise environment is not good. Even the ground line in our lab is said to be shifting because there are quite huge power-consumption machine nearby.
And the voltage measurment lead line is not that long, say about 1 meter, and we are using twisted lead cable with shield.

[RJFreeman]

It is still a bit vague;
what is the sensor? is it a PT100/1000 etc or an LM335 or similair?
what is the meter?
     a hand-held multimeter? (in which case it is already effectively and isolated circuit)
     a mains powered bench multimeter?
     something else?
what is the 'constant current source?
    a current diode?
    a voltage regulator with series high resistance?
    a battery with series resistance
    a more complex circuit?
what are you measuring?
   a motor?
   a switchboard?
   a heatsink?
what form is this noise?
   a long term drift?
   a short hit?
   constant wandering measurements/doesn't settle?

the immediate thought is that 10uA probably doesn't take much interference to make your measurement  problematic