NTC Thermistor Resistance Specs tolerance range at Zero PowerResistanceAt 25°C

crutschow

Joined Mar 14, 2008
34,285
How are you measuring the resistance?
A standard ohmmeter will not be very accurate at such low resistance, due to varying contact resistances.

You really need to use a 4-wire (Kelvin) resistance measurement technique.
You put a known small current through the resistance (say 10mA) and measure the voltage directly across the resistor, from which you can calculate the resistance.
 
Last edited:

Ian0

Joined Aug 7, 2020
9,671
I don't see a tolerance value for the resistance specified in that datasheet.
This is a similar product made by EPCOS
http://www.farnell.com/datasheets/2921099.pdf
It has a tolerance of ±20%, which is probably typical for that sort of product.
What does your spec say is the maximum acceptable tolerance of the resistance? If the tolerance is not specified, it is a poorly written spec.
If that is the same for your product, that would indicate a spread of resistance of about 4Ω which is what you observe.
The mean value of your thermistors is 12Ω - what is the temperature in your lab? Have you subtracted the resistance of your meter probes?
Have you accounted for the contact resistance as @crutschow said?
 

Thread Starter

Lumenosity

Joined Mar 1, 2017
614
How are you measuring the resistance?
A standard ohmmeter will not be very accurate at such low resistance, due to varying contact resistances.

You really need to use a 4-wire (Kelvin) resistance measurement technique.
You put a known small current through the resistance (say 10mA) and measure the voltage directly across the resistor, from which you can calculate the resistance.
Thanks.
This information led me to read about 4 wire vs 2 wire resistance testing.
A little more educated each day.

Unfortunately, I don't have a 4 wire ohmeter, but now I can better appreciate your reply.
 

Ian0

Joined Aug 7, 2020
9,671
Thanks.
This information led me to read about 4 wire vs 2 wire resistance testing.
A little more educated each day.

Unfortunately, I don't have a 4 wire ohmeter, but now I can better appreciate your reply.
You don't need one - if you have a bench power supply and a voltmeter.
Wire up the NTC thermistor in series with the 100 ohm resistor to the bench power supply. Adjust the power supply so that the voltmeter reads 10V when connected across the 100 ohm resistor. Don't trust the readout on the power supply.
Now set the meter to the mV range and measure the voltage across the thermistor. Ohm's law will tell you the resistance of the thermistor.
 

Thread Starter

Lumenosity

Joined Mar 1, 2017
614
I don't see a tolerance value for the resistance specified in that datasheet.
This is a similar product made by EPCOS
http://www.farnell.com/datasheets/2921099.pdf
It has a tolerance of ±20%, which is probably typical for that sort of product.
What does your spec say is the maximum acceptable tolerance of the resistance? If the tolerance is not specified, it is a poorly written spec.
If that is the same for your product, that would indicate a spread of resistance of about 4Ω which is what you observe.
The mean value of your thermistors is 12Ω - what is the temperature in your lab? Have you subtracted the resistance of your meter probes?
Have you accounted for the contact resistance as @crutschow said?
Thank you.

I cannot find a datasheet for this component that offers a tolerance range.
I tried to get the lab temperature to 77F or 25C as specified in the datasheet.

My meter probes resistance measures 0.3 ohms, however, the accuracy is based on a two wire test.

What really throws a monkey wrench into my understanding is this.......

3 of the thermistors now consistently test at 10.4, 9.9 and 9.7 ohms (well within specs)
This is after apply heat to them to test them. I haven't tried applying heat to the others yet.
But if the others fall into the 10ohm range plus or minus 2% then maybe new thermistors need some kind of "break in" ?

With my two wire multimeter I get very consistent readings and those readings now indicate three of the thermistors are completely acceptable.

I will apply heat to some of the others and see if they read closer to 10 ohms afterwards.

Still, I understand that I will not get very accurate readings under 100 ohms with a 2 wire ohmmeter.
 

Thread Starter

Lumenosity

Joined Mar 1, 2017
614
You don't need one - if you have a bench power supply and a voltmeter.
Wire up the NTC thermistor in series with the 100 ohm resistor to the bench power supply. Adjust the power supply so that the voltmeter reads 10V when connected across the 100 ohm resistor. Don't trust the readout on the power supply.
Now set the meter to the mV range and measure the voltage across the thermistor. Ohm's law will tell you the resistance of the thermistor.
Thanks Ian0.

You made that simple enough for even a beginner like me to understand.
will do.

Cheers!
 

Ian0

Joined Aug 7, 2020
9,671
Generally, one is more interested in the ability of the device to limit inrush current and it ability not to explode whilst doing so, than its exact resistance!
 

Ian0

Joined Aug 7, 2020
9,671
Thanks Ian0.

You made that simple enough for even a beginner like me to understand.
will do.

Cheers!
I used the technique a lot - we used to have a dodgy cable supplier who like to send copper-coated-steel, copper-coated-aluminium or brass cable instead of copper. I needed a technique to check the resistance of the cable without having to unspool a whole reel. With a 1A test current I could use this technique on a 1m length of cable.
 
You end up measuring the current through the device and the voltage across the device. Unless you have a feedback ammeter, you really need the current meter (a simple resistor will do) in the circuit at all times. The common 10 meg ohms input impedance of a typically voltmeter across 10 ohms is negligible. 1/R=1/10M+1/10

When your measuring 10M or 1G ohms. the technique differs. You use a feedback ammeter which drops mV and a voltage source.

When things get really weird, I had to count coulombs for 30s and divide by 30 to measure currents around 3 pA.

All sorts of wacky things happen when measuring tiny currents like the piezoelectric effect and the triboelectic effect.
Cable movement generates current because it's a conductor in the Earth's magnetic field and physics tells us that generates a current.

You also have guard voltages and shields. The cables have two shields with a layer of graphite, The inner shield is at the same potential as the center conductor. The outer shield is ground,

I could measure the resistance of paper.

The 4-wire Kelvin technique is common. There is even Kelvin alligator clips and dual probes for probing IC;s.
They are common for LCR meters.
 

Alec_t

Joined Sep 17, 2013
14,280
Adjust the power supply so that the voltmeter reads 10V when connected across the 100 ohm resistor.
That puts 100mA through the thermistor, which will dissipate ~100mW. Isn't that far too high for a zero power resistance reading?
 

Ian0

Joined Aug 7, 2020
9,671
That puts 100mA through the thermistor, which will dissipate ~100mW. Isn't that far too high for a zero power resistance reading?
Thanks for pointing that out - I meant 1kΩ. If I knew the TS had a meter with a good mV range, I'd have recommended 1mA.
 
Top