[Study] PTC/NTC Compensation

Cover from Danny Nicholson (Flickr)

All the components we use are sensitive to thermal gradient. It’s the laws, laws of physics.

Most of the time , efficiency decrease when the temperature increase. For correcting that you can put some cooling air , thermic-dissipation , isolation , etc…

Especially that this drift is expressed on data-sheets so it’s easy to put number on it.

It works, but it could became a real hell when you want to build an embedded system , low-cost device or any compact / amateur thing.

So the point is to use PTC and NTC , who can have sensitivity value in function of temperature in a linear way and in the slope we want (PTC => + / NTC => -) for correcting the drift of an another device.

For this study i will take a photo-diode SFH2701 .

This photo-diode have a drift of sensibility of -0.01% by Kelvin (exposed at 780nm light)

DriftSFh2701

It means that the sensitivity of 0.5 A/W decrease by -0.2 % every 20°.

If we use it with a trans-impedance circuit (conversion current/voltage), it will normally decrease the output signal of -0.2% for a gradient of [0° – 20°].

Let’s draw this for an 500 kOhm trans-impedance circuit at 7.5mW exposure (850nm)      (X = Celsius ; Y = Volts) :

C2

The drift is very low but it’s a real pain in the ass when all your system is dependent of it.

The data-plot shows that an increase of 20 degree Celsius decrease the output of 3mV which mean 1000 Ohm transposed to the resistor.

All the magic begin here : I will replace the 500 kOhm resistor by a 490 kOhm resistor followed by a PTC of 10 kOhm (20°). So the amplification will be the same but i introduce the PTC sensitivity.

For a perfect compensation, i need to find a PTC with the closest amount of “Ohm by 20° Celsius” compare to the resistor equivalent-drift => 1000 Ohm.

When it’s done let’s plot the compensated output :

C32

Not perfect , but much more linear 😉

The real limitation of this technique is the range of some PTC / NTC, who can’t handle  fluctuations (Sensitivity to low / Resistor Value to low).

I haven’t try this method on other things than optical system, but feel free to share if you use it too !

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s