Thermistor help

Hi,

I am just coming to the end of a long reprap build and am currently trying to get the extruder board to measure the temperature through a thermistor. Its not plugged into a nozzle at the moment as I am still waiting for the workshop to finish knocking up the extruder nozzle parts. So I thought I would try getting the extruder to measure ambient room temperature. At the moment the extruder is reading 19 degrees whereas my calibrated room thermometer is saying 21 degrees.

The thermistor I am using is a 100k one from Maplin, I believe it is manufactured by Siemans. At the moment I am wondering about the discrepancies between the extruder and the calibrated thermometer. Is this difference normal for a thermistor, what kind of accuracy should I be aiming for?

I grabbed the resistance at 25 degees and the beta value from the Maplin site and plugged them into the python script to get a lookuptable before I got a reading of 19 degrees. However, the Maplin site also gives the resistance for 0 degrees and 100 degrees, when I plug these figures into the java program on the reprap thermistor page I get a different beta value to that given by Maplin.

I'm thinking I should measure the resistance at two temperatures using a multimeter and calculate a lookuptable customised to my thermistor but was wondering if it will actually give me a more accurate temperature or should I stick with what I've got; basically, is it going to be worth the effort?

Thanks for any and all advice,

Milly

From the way you're describing it, the thermistor is hooked up to your electronics.

There is a resistor the thermistor is hooked up to (making a voltage divider). in most electronics setups it's a 4.7k resistor. The resistor is part of the formula for creating the lookup table. If you measure this resistor accurately, you may find that it's not quite 4.7k (usually it's a 5% tolerance resistor so it'll be anything up to +-5% of the value). Putting the accurate resistance reading in may help your readings.

That said, most people simply use the temperature reading as a guide to get the nozzle near the right temp, and then adjust it manually by a few degrees to compensate for any drift.

There is an interesting article that gives you an idea of how the resistance in a thermistor behaves, and how you can optimise your thermistor table at [www.brokentoaster.com].

PS: Also note that at room temp, glass tends to be just slightly cooler than room temp (metals too, but for different reasons). You may find that what you are reading from the thermistor is actually fairly accurate when you take this into account.

PPS: After a bit of thought, the above PS note is probably not relevant, as any possible effect I suspect would be in the sub-degree range.

Edited 2 time(s). Last edit at 02/14/2011 10:55PM by Cefiar.

Hi Cefiar,
Why are glass and metal below room temp? I thought they just felt cold because they are good conductors with significant heat capacity.

---

[www.hydraraptor.blogspot.com]

This is one of those things that's hard to quantify, but from memory it is related to the fact they are good thermal conductors and that they have significant heat capacity. I'm also assuming here that the air is what is heating the object, rather than the other way around.

The difference is minimal (should be less than a degree from what I have seen), and apparently depends on many variables. It's based around Non-equilibrium thermodynamics, which (to my eyes) is very complicated. I know 'of it', but I don't dare attempt to understand it - I'd need a lot more physics under my belt first.

As I said, the difference is minimal. But it is there. The world unfortunately is not an ideal place.

There is also the effect of Interfacial thermal resistance, which may be relevant to such a phenomenon. I still doubt it would be much of an issue at the temperatures we deal with, but the property still exists.

regarding glass:
one thing i remember reading when working with borosilicate for a hot end is that glass doesn't have good thermal conductivity in the same way that metal does. since the glass is clear, the radiative portion of the heat is transfered efficiently, but it doesn't actually conduct heat well as a conductor. or something like that...

Rocketscientist was saying that he could get the glass tubing yellow hot and be holding it in his ungloved hand 5 cm away.

@cefiar: your explanation does not account for the 2 degree difference that Milly is seeing. Nor does it make sense with regards to my understanding of thermodynamics. glass or metal will eventually come to thermal equillibrium with the surrounding air unless the temperature of the air is constantly changing. glass and metal at room temperature will feel cold to your touch since your body temperature is higher at 37C.

I think prusajr just posted a blog in the reprap wiki comparing a thermistor to a thermocouple and he saw +/- 2 degree difference from room temperature to around 260 C. This is consistent with what Milly saw. Also, we don't know how well calibrated the other thermometer is.

Edited 1 time(s). Last edit at 02/14/2011 11:11PM by brnrd.

brnrd; In an ideal model, It should come into thermal equilibrium. Nothing is ever ideal.

I edited my post with a subsequent comment that the issue would not cause the effect Milly is after an answer for. I've edited it again since to 'cross out' the original comment, so that people don't get the wrong idea.

My temperature readings at room temp also seem to be about 19C when the room is more like 23C. I have never bothered trying to get it better because I am only interested in it being accurate at the top end. The equation we use with Ro and beta is only an approximation and sometimes the manufacturer quotes more than one beta for different temperature ranges. I calibrate mine at about 240C and I always get a resultant beta slightly different from the datasheet.

---

[www.hydraraptor.blogspot.com]

Thanks for the interesting comments, would the beta value alter the accuracy in a given band, i.e. is it likely the thermistor will be more accurate at higher temperatures and if I mess with trying to change the beta value by measuring the resistance over two different temperatures and recalculating the beta is it therefore more likely to be accurate in the band I measured it?

Milly

Would an IR temperature gun like this one sold at amazon be good enough to measure the temperature of the extruder tip? I figure that I would have to hold it very close to the tip since it's only a couple of mm across.

Milly,
another point that I did not see when skimming though the thread is that thermistors are notoriously not accurate. When you initially said that you were getting 19C on the thermistor and 23C from a room thermostat. my first thought was "wow, you already have it dialed in!". Thermistors change resistance over several orders of magnitude, and not linearly. Sometimes not even uniformly. That is why you calibrate one by taking several readings at widely separated, but very accurate temperatures, or use a more precise temperature reference like a thermocouple.

If your thermistor currently reads 4C less than the room temperature thermostat, then it will likely always read 4C low near room temp. The idea is that once you have a working extruder, you tweak the control circuit until the extruded plastic filament looks good, as use that 'temperature' as you default. Likely, it will also be in error, but repeatable, and that is really all you need for an extruder controller. Don't sweat the small stuff, just make it good enough and tweak till it works good.

And yes, borosillicate glass is such a poor thermal conductor that 1000C across 5 cm is no problem.


rocket_scientist

Well, I took an accurate reading on the resistor (R6) on the extruder board and it has a resistance of 4510ohms which is inside 5% tolerance. I tried sticking this more accurate resistance into the python script and re-flashed the extruder with the new lookuptable. To my surprise I am now getting more accurate readings at room temperature, it gave me a reading of 20 degrees (although replicator displays the temp to 1 decimal place its always rounded to a whole number, is this correct?) whilst my accurate thermometer was reading 20.4 which I don't think I will be able to improve on. It remains to be seen if the thermistor will maintain this level of accuracy when I get the nozzle up and running and crank the heat up, but I guess as rocket_scientist has suggested, getting the PLA flowing nicely will give me a figure which is then repeatable and its not so important that it accurately reflects the temperature.

I was surprised at the difference 190 ohms made though!

Milly.

brnrd,
IR thermometers are no good for measuring the nozzle temperature because the apperture is too large compared to the nozzle. Also the emissivity they assume is too high for shiny metallic objects.

I put a thermocouple inside the barrel before any plastic and calibrate at room temp and 240C. It will then track the thermocouple within a couple of degrees throughout all the band of interest.

Milly,
Yes you will get a slightly different beta value if you calibrate over different ranges because the standard model with beta and r0 is an approximation to the actual curve. You will get the best value by calibrating at the highest and lowest temperatures of interest but that might make room temperature a long way out. It really doesn't matter that much though. Plastics from different sources all vary a bit so you have to play around with the temperature and you have to change it by about 5C before you notice any difference, so it doesn't need to be much more accurate than that.

---

[www.hydraraptor.blogspot.com]

nophead,

Thanks. I thought so but I ordered it anyway so that I can measure the bed temperature and other things like the motors, the part being printed and the room. When I get bored, I figure I can point it at people and measure their skin temperature.