# Plot the Temperature Characteristic of a Conductive material

#### kalemaxon89

Joined Oct 12, 2022
217
I have a resistive strip made of Polymide / Nylon metal plated with silver (approx. 3cm long and 4cm wide) that has an electrical resistance of 100 +-30 ohm/meter.
I do not have the datasheet (I think it is not really available), this information is all I have available.

I will have to make a circuit (it's not the purpose of this thread) that heats this material in a controlled way depending on how much current flows (or how much voltage I apply to it).
Reasoning a bit I thought that, before making such a circuit, I think I need to understand:
- how the resistance of this material varies as a function of temperature .. does it behave like a resistor? Is the resistance-temperature curve linear?
- maybe also the resistance-current curve characteristic (?)
- the temperature-voltage curve characteristic (?)
1) What other characteristic curves should I add (or remove) to the bulleted list I have just written?
Maybe just the first one I listed is enough, or the second ... you tell me.
(this is the first time I have come across a component that does not have a datasheet and therefore I have to characterise it myself)

If I am right about this first step to do and while we add (or remove) the characteristic to the bulleted list ..
I ask your anotehr advice about 2) How to make a measuring circuit in order to plot this relations (characteristics curves) that I've just written int the bulleted list.
My first idea of the setup circuit is this one:
(the classic setup circuit for an NTC sensor measurement) I feed the conductive material and, depending on how much current flows (or how much voltage at its ends), the resistance of the NTC (or another temperature sensor) will vary. With a simple divider I measure the voltage and, by turning the formula, find the temperature.
Will this simple circuit be OK? Or do we need something more precise?
Doing some researches .. I found this test circuit which (I think) is better than a simple resistive divider
What do you recommend?

Thanks!!

#### ericgibbs

Joined Jan 29, 2010
18,268
n electrical resistance of 100 +-30 ohm/meter.
Hi K89,
What accuracy of resistance measurement are you requiring?
Metals usually have an increase in resistance with temperature
I would not use a NTC thermistor, as they have a non-linear resistance versus temperature characteristic.
E

• kalemaxon89

#### DickCappels

Joined Aug 21, 2008
10,104
- how the resistance of this material varies as a function of temperature .. does it behave like a resistor? Is the resistance-temperature curve linear?

Depends on how hot you get it but around room temperature the resistance is pretty linear.

You have 100 feet of 20 gauge wire and its resistance is 1.015 ohms at 20° C (room temp). If the temperature of the wire goes up 10°C, the resistance will change by 0.0399 ohms (10 degrees * 0.00393 per degree * 1.015 ohms = 0.0399 ohms).
The wire resistance will now be 1.015 ohms + 0.0399 ohms = 1.0549 ohms. (thermal units are Celcius or Kelvin)

Different metals have different characteristics.

- maybe also the resistance-current curve characteristic (?)

I don't think so, until the current becomes enough to make the conductor noticeably warmer.

- the temperature-voltage curve characteristic (?)
That depends a lot on the composition and structure of the conductor, but it is related to the temperature that results from the I squared R dissipation of the conductor.

If possible use a thermocouple to measure the temperature of your conductor because it would be much more linear than a thermistor, and therefore easier to calbrate or check.

Now, about really high temperatures, here is what a tungsten light bulb does (used to do, when we could buy them)  source: https://www.researchgate.net/figure...tungsten-against-temperature-4_fig2_273005745

• kalemaxon89

#### kalemaxon89

Joined Oct 12, 2022
217
Hi K89,
What accuracy of resistance measurement are you requiring?
Metals usually have an increase in resistance with temperature
I would not use a NTC thermistor, as they have a non-linear resistance versus temperature characteristic.
E
I thought so, I can try an RTD or thermocouple.
I saw that both (usually) have metal connectors ... do I connect them directly to the conductive textile strip or do I have to put something in between?
@DickCappels

#### DickCappels

Joined Aug 21, 2008
10,104
I have had no problem connecting a floating thermocouple and meter to an electrified metal part, such as the metal tab on a power transistor.

high voltage would have been a problem.

• kalemaxon89

#### kalemaxon89

Joined Oct 12, 2022
217
If possible use a thermocouple to measure the temperature of your conductor because it would be much more linear than a thermistor, and therefore easier to calbrate or check.
I have had no problem connecting a floating thermocouple and meter to an electrified metal part, such as the metal tab on a power transistor.

high voltage would have been a problem.
I was thinking that with a thermocouple I need a more complex conditioning circuit than the classic resistor in series with the NTC ... am I wrong? (for example an AD590 and an instrumentation amplifier to get an appreciable output voltage)

I have never used a thermocouple and could (mistakenly) attach it directly to the conductive wire .. I have no idea how to make the connection, I can do some testing.

Can you link some examples of the thermocouple you are talking about? There are so many on Digikey and I don't really know where to start .. maybe it's easier than I think

#### DickCappels

Joined Aug 21, 2008
10,104
I failed to mention that some digital voltmeters have a setting to measure temperature with an included thermocouple. You can buy an inexpensive one an expensive one, your choice, they are both useful.

You can also build your own thermocouple amplifer but it a very involved affair, and not recommended for...anybody. How do you make the connection?
1. Plug the thermocouple into the meter
2. Switch to temperature measurement mode
3. (the thermocouple is already attached to the subject, so look at the meter and see the temperature.
3A. If your thermocouple is something at some would call high voltage, don't touch the thermocouple when the circuit being observed is energized, and don't touch the meter either!

Does that clear things up?

• kalemaxon89

#### Alec_t

Joined Sep 17, 2013
14,012
I have a resistive strip made of Polymide / Nylon metal plated with silver (approx. 3cm long and 4cm wide) that has an electrical resistance of 100 +-30 ohm/meter.
For a given current through that strip its temperature will depend largely on what surrounds it.
Is it in free air? What supports it? How much heat is conducted away by the connecting leads?

• kalemaxon89

#### kalemaxon89

Joined Oct 12, 2022
217
For a given current through that strip its temperature will depend largely on what surrounds it.
Is it in free air? What supports it? How much heat is conducted away by the connecting leads?
It is in free air at 25°C.

Obviously in the 'thermal model' of the measuring system I have to take into account a resistor between sensor and strip, another resistor between sensor and environment and the effect of self-heating on the temperature sensor I will use.

#### MrAl

Joined Jun 17, 2014
10,927
Hello there,

Thermistors are non linear but their characteristics are very well known.
There are some regular thermistors and some made just for measuring temperature and are accurate to within about 1 percent.
The 1 percent type you can find easily have a 10k nominal resistance at 25 degrees C and a constant they often refer to as Beta or just 'B'.

Here is a formula for using such a thermistor connected to ground with a fixed resistor R2 connected from +5 volts to the other end of the thermistor, and measuring the voltage 'v' at the junction of the two:
Rx=(5*R2-v*R2)/v
degC=B/(ln(Rx/R1)+B/T1)-273

where
B is a constant of the thermistor (4100 for example),
R2 is the fixed resistor resistance,
R1 is the resistance at 25 degrees C (10k typical),
v is the measured voltage at the junction of the thermistor and fixed resistor,
'5' is the voltage supply (5v typical so 5 was used here),
T1 is the actual temperature when the resistance is R1 (25 C typical).

The best linearization occurs when R2 is about 17k for a temperature range of 0F to 100F. The reason for using this value is not to get a perfect linearization (the formula takes care of that) but to get the spread between successive temperature degrees to be as equal as possible, which means if you obtain 2.52v at 56 degrees F and 2.42v at 58 degrees F, then you would hope to get close to 2.32v at 60 degrees F, which gives you about -0.1v change for each change of +2 degrees F. That means the resolution is about the same over the expected temperature measurement range.
If your temperature range is less than 0F to 100F then you get even better results.

The formula is used in conjunction with a microcontroller which is the best way to do this anyway.

There is actually a third constant you can use to get even better accuracy, and of course a lookup table with interpolation gets you really good accuracy if you think you really need that.

As with any temperature measurement, the probe has to come into contact with the item to be measured, and if the item is long you may need to use several probes and calculate the temperature at several points along the item length. You can then use all the individual measurements or take an average.
Cooling can also be something to think about.

• kalemaxon89

#### kalemaxon89

Joined Oct 12, 2022
217
I failed to mention that some digital voltmeters have a setting to measure temperature with an included thermocouple. You can buy an inexpensive one an expensive one, your choice, they are both useful.

You can also build your own thermocouple amplifer but it a very involved affair, and not recommended for...anybody. How do you make the connection?
1. Plug the thermocouple into the meter
2. Switch to temperature measurement mode
3. (the thermocouple is already attached to the subject, so look at the meter and see the temperature.
3A. If your thermocouple is something at some would call high voltage, don't touch the thermocouple when the circuit being observed is energized, and don't touch the meter either!

Does that clear things up?
Hello again, everyone. @MrAl @DickCappels
I have taken some measurements on the conductive patch and have a technical doubt.

I fed the conductive textile material (it has about 80 ohms +-30 ohms) with different voltage values (from 1V to 8V) with a bench power supply and, for each V, I obviously noticed an increase in temperature and current. Around 45°C the bench power supply showed 8V (I set it) and 0.11A on its display.
If I'm not mistaken, 0.11A is the current flowing in the circuit ... in fact another digital multimeter also indicated about that figure.

Correct me if I'm wrong, if I want to supply the material until it reaches, say, 45°C .. I have to supply it with a voltage of 8V (which will make a current of 0.11A flow).
If I use a PWM-controlled mosfet and put conductive material between Vcc and drain ... by changing the Vgs I can decide how much current to run in the drain (and thus on the conductive material).
But what if I directly connect a PWM voltage (from a micro) to the strip/conductive material? Wouldn't that simplify my life?

#### Alec_t

Joined Sep 17, 2013
14,012
But what if I directly connect a PWM voltage (from a micro) to the strip/conductive material? Wouldn't that simplify my life?
You would expect it to. Providing the material has enough thermal mass it will behave as though it receives the average voltage provided by the PWM. However, the micro is unlikely to be able to supply directly the 0.1A current, so you would still need the MOSFET.

• kalemaxon89 and DickCappels

#### MrAl

Joined Jun 17, 2014
10,927
Hello again, everyone. @MrAl @DickCappels
I have taken some measurements on the conductive patch and have a technical doubt.

I fed the conductive textile material (it has about 80 ohms +-30 ohms) with different voltage values (from 1V to 8V) with a bench power supply and, for each V, I obviously noticed an increase in temperature and current. Around 45°C the bench power supply showed 8V (I set it) and 0.11A on its display.
If I'm not mistaken, 0.11A is the current flowing in the circuit ... in fact another digital multimeter also indicated about that figure.

Correct me if I'm wrong, if I want to supply the material until it reaches, say, 45°C .. I have to supply it with a voltage of 8V (which will make a current of 0.11A flow).
If I use a PWM-controlled mosfet and put conductive material between Vcc and drain ... by changing the Vgs I can decide how much current to run in the drain (and thus on the conductive material).
But what if I directly connect a PWM voltage (from a micro) to the strip/conductive material? Wouldn't that simplify my life?
Hi,

You can also use a buck switching regulator to power the device(s) and that would give you better efficiency in most cases too, if that matters that is.
If efficiency isnt an issue at all then yes PWM or a linear will do it.

If you did mean direct drive from a microcontroller that would probably not work because most microcontrollers have a max output current PER I/O PIN of about 20 ma. You may get away with using 5 or 6 I/O pins but you also have to pay attention to the max Vcc and ground currents too, which could be just 100ma. Since it is not hard to add a bipolar that's probably going to be the simplest method.

#### kalemaxon89

Joined Oct 12, 2022
217
You would expect it to. Providing the material has enough thermal mass it will behave as though it receives the average voltage provided by the PWM. However, the micro is unlikely to be able to supply directly the 0.1A current, so you would still need the MOSFET.
Hi,

You can also use a buck switching regulator to power the device(s) and that would give you better efficiency in most cases too, if that matters that is.
If efficiency isnt an issue at all then yes PWM or a linear will do it.

If you did mean direct drive from a microcontroller that would probably not work because most microcontrollers have a max output current PER I/O PIN of about 20 ma. You may get away with using 5 or 6 I/O pins but you also have to pay attention to the max Vcc and ground currents too, which could be just 100ma. Since it is not hard to add a bipolar that's probably going to be the simplest method.
Pardon the possibly trivial question, but I do not understand a concept that is certainly very basic.

You rightly say that the pin of a microcontroller does not produce the mA that I would probably need to heat the material.
But it occurs to me that, given the material's approximately 80-100ohm resistance .. it should 'act' as a resistor .. so if I simply supply it with a voltage (say 5V from the micro) it will produce I=V/R and P=R*I^2 and will inevitably have to heat up (as a resistor would).
If this reasoning is correct, the limitation here is the 5V that a micro usually provides .. so probably I am forced to use a DC buck or BJT or PWM mosfet etc.

Because I often hear about how much current I have to provide into the load, but I don't understand why it's not enough to say "let's give a voltage to the load" and the current will be generated by the load itself thanks to the Ohm's law .. It is probably the same thing(?) and my doubt does not make sense

#### MrAl

Joined Jun 17, 2014
10,927

Pardon the possibly trivial question, but I do not understand a concept that is certainly very basic.

You rightly say that the pin of a microcontroller does not produce the mA that I would probably need to heat the material.
But it occurs to me that, given the material's approximately 80-100ohm resistance .. it should 'act' as a resistor .. so if I simply supply it with a voltage (say 5V from the micro) it will produce I=V/R and P=R*I^2 and will inevitably have to heat up (as a resistor would).
If this reasoning is correct, the limitation here is the 5V that a micro usually provides .. so probably I am forced to use a DC buck or BJT or PWM mosfet etc.

Because I often hear about how much current I have to provide into the load, but I don't understand why it's not enough to say "let's give a voltage to the load" and the current will be generated by the load itself thanks to the Ohm's law .. It is probably the same thing(?) and my doubt does not make sense
Hello again,

A typical microcontroller has a typical current limit per pin of 20ma. That's the most it should conduct. It does not matter what the load is as long as the load does not draw more than 20ma. A load does not 'generate' current, only the microcontroller does. The microcontroller i/o pin must be rated for the current that the load is going to draw.
With a 5v microcontroller a 100 Ohm resistor would draw 5/100=50ma of current, which would be too high. It's as simple as that. With that high a current it may burn out the pin.
To keep the current at 20ma or less, you would need a resistor of value 5/0.020=250 Ohms. Anything less than 250 Ohms may burn out the pin.

That's a typical microcontroller, you may find some that can go to 25ma or maybe you can find one that goes higher. You have to check the spec on your microcontroller.
Also note that is for ONE i/o pin. It may be possible to use more than one pin, but a transistor is the better way to handle this. Drive the transistor with an i/o pin and let the transistor handle the extra current. It may be possible to use one single NPN transistor as a voltage follower and get the output current you need, as long as the voltage is high enough for your device. You could also consider driving the base with a resistor and use the NPN as a common emitter inverter to drive the load as you get higher output voltage across the load that way. This is all if you intend to use PWM unless you have a DAC peripheral on the microcontroller. If you do have a decent DAC then you can use the transistor in linear mode if you care to do that, but you'll have to watch out for the power dissipation in the transistor also.

Some microcontrollers do not even allow 20ma per pin, you may run into one that only allows 10ma per i/o pin. You really have to check your actual part to get this information. I also have to wonder now what microcontroller you are going to use. You also have to watch out for the i/o pin voltage sag which would mean the pin HIGH level voltage which is supposed to be 5v may load down to 4.5v or something like that, which may or may not be a problem with whatever you are driving with that pin.

Last edited:
• kalemaxon89

#### kalemaxon89

Joined Oct 12, 2022
217
Hello again,

A typical microcontroller has a typical current limit per pin of 20ma. That's the most it should conduct. It does not matter what the load is as long as the load does not draw more than 20ma. A load does not 'generate' current, only the microcontroller does. The microcontroller i/o pin must be rated for the current that the load is going to draw.
With a 5v microcontroller a 100 Ohm resistor would draw 5/100=50ma of current, which would be too high. It's as simple as that. With that high a current it may burn out the pin.
To keep the current at 20ma or less, you would need a resistor of value 5/0.020=250 Ohms. Anything less than 250 Ohms may burn out the pin.

That's a typical microcontroller, you may find some that can go to 25ma or maybe you can find one that goes higher. You have to check the spec on your microcontroller.
Also note that is for ONE i/o pin. It may be possible to use more than one pin, but a transistor is the better way to handle this. Drive the transistor with an i/o pin and let the transistor handle the extra current. It may be possible to use one single NPN transistor as a voltage follower and get the output current you need, as long as the voltage is high enough for your device. You could also consider driving the base with a resistor and use the NPN as a common emitter inverter to drive the load as you get higher output voltage across the load that way. This is all if you intend to use PWM unless you have a DAC peripheral on the microcontroller. If you do have a decent DAC then you can use the transistor in linear mode if you care to do that, but you'll have to watch out for the power dissipation in the transistor also.

Some microcontrollers do not even allow 20ma per pin, you may run into one that only allows 10ma per i/o pin. You really have to check your actual part to get this information. I also have to wonder now what microcontroller you are going to use. You also have to watch out for the i/o pin voltage sag which would mean the pin HIGH level voltage which is supposed to be 5v may load down to 4.5v or something like that, which may or may not be a problem with whatever you are driving with that pin.
Are there any mosfets that can deliver a maximum current of 0.2A in the drain?
That is, you adjust the PWM on the gate to have about 200mA on the drain. Because I have usually seen Id-Vds-Vgs with currents of tens of amperes.

If the answer is no, I have to think about a bjt as you suggested

#### MrAl

Joined Jun 17, 2014
10,927
Are there any mosfets that can deliver a maximum current of 0.2A in the drain?
That is, you adjust the PWM on the gate to have about 200mA on the drain. Because I have usually seen Id-Vds-Vgs with currents of tens of amperes.

If the answer is no, I have to think about a bjt as you suggested
Hi,

Yes, but that's if you switch it. If you want to operate in the linear mode you'll have to add feedback.

#### kalemaxon89

Joined Oct 12, 2022
217
To keep the current at 20ma or less, you would need a resistor of value 5/0.020=250 Ohms. Anything less than 250 Ohms may burn out the pin.
Shouldn't it be [5-Vbe(min)]/0.020 ?

#### MrAl

Joined Jun 17, 2014
10,927
Shouldn't it be [5-Vbe(min)]/0.020 ?
Hi,

Yes if you used a bipolar transistor, and with an LED then (5-vLED)/0.020, etc.
However, when we talk about general loads we often talk about pure resistance. A pure 250 Ohm resistance at 5v will draw 20ma.
We can explore other types of loads if you wish. A MOSFET is quite a bit different though we have to consider the switching dynamics and the threshold voltage carefully.

#### Janis59

Joined Aug 21, 2017
1,783
RE:""2) How to make a measuring circuit in order to plot this relations (characteristics curves)""
The simplest solution is to apply the Picotech Data box, having already the Thermocouple input and output to PC thus the PC keps data in excel. This wonderbox is rather cheap.