thermistor

i've been searching up (cheap) thermistors on ebay/aliexpress etc, for the cheap ones a lot of very similar ads for mf52 and mf58 etc
[www.ebay.com]
the notable ones are mf52 types
[www.eaa.net.au]
these apparently have a max temperature of some 120 deg c
hence these are out

then the other very common ones commonly labelled mf58
so called 'glass encapsulated' ones which could stand temperatures up to some 200-300 deg C (the max temps seemed ok initially but ..)
[www.rdd-tech.com]
[vatronics.com]
the trouble is even with these 'glass encapsulated' ones, many of them shows a resistance falling from 100k ohm at 25 deg C to a mere 2km ohm (often less) at 120 deg C
the big question would then be how far can that go when temperatures rise from 120 deg to the operating temperatures of say 230 deg C? drop to only a few ohms?

and given the extreme variations of the specs how do one go about calibrating the thermistors to measure the correct temperatures?
i'm wondering if many of the temperature troubles could after all be caused by inappropriate / incorrectly calibrated thermistors after all

Edited 2 time(s). Last edit at 07/17/2018 08:45AM by ag123.

1) you don't read the ohms, in real life you measure a voltage though the resistor
2) you use a 12-bit DAC on an arduino so thats 4096 voltages between 0 and 5v
3) there is also a voltage divider in the circuit, cant exactly remember the details but Im fairly sure this magnifies the readings further, so you get 4096 reading over a smaller voltage

This is why we have tables to turn read voltages into temptures (can use math, but ardunio is to slow to do this, some 32 but systems use this)

Yes you very much need to set the correct thermistor or you can get horribly wrong reading in parts of the temperature curve

How do you calibrate? maths. beta values for the thermistor allow you to calculate the resistance at a temperature and create a graph.
If you dont know the beta, you have to use an already calibrated system to set the temperature and take readings, draw your own graph.

You then need to generate tables compatible with your firmware.
Its far easier to get a thermistor that tables already exist....

thanks for the reply!

i decided to do a thought experiment, i take one of those specs for MF58 100k ohm (25 deg C) beta: 3950
then i take the formulae from
[en.wikipedia.org]
[reprap.org]

i got:

and the list of points that made up the chart

T	R
Deg C	Ohms
15	158,446.1
25	100,000.0
35	65,027.6
45	43,446.1
55	29,749.7
65	20,832.8
75	14,890.3
85	10,844.5
95	8,035.2
105	6,048.9
115	4,620.7
125	3,577.8
135	2,805.3
145	2,225.3
155	1,784.4
165	1,445.4
175	1,181.9
185	974.9
195	810.8
205	679.6
215	573.7
225	487.7
235	417.2
245	359.0
255	310.7
265	270.4
275	236.5
285	207.8
295	183.5

well, i'm not too sure how well those (cheap) thermistors would conform to the graph
as the specs such as MF58 100k ohm (25 deg C) beta: 3950 seem to be given in just about every ad on ebay/aliexpress etc

the other thing that don't look too well is the high temperatures parts (150 deg C - 250 deg C) which happens to be the common temperatures for PLA, ABS 3d prints
the graph is pretty much 'flat' hence the resistance changes between one degree centigrade to the next would also be small

it wouldn't be easy to calibrate the thermistor if the beta value is somewhat different than that stated, one would need a good thermometer (at least rather accurate and sufficiently precise)
and one which can take the high temperatures (e.g. at around 180-250 deg C)

i've been thinking if one could after all take those 'cheaper' digital thermometers (which often means one is limited to the boiling point of water 100 deg C) to calibrate the 'curved' part of the graph, and hope that it would 'just work' for the high temperatures part. boiling water is 100 deg C, which seem to provide a benchmark to calibrate it at least for the low temperature segment.

i'm not sure if such an approach (i.e. just try to calibrate between room temp to boiling point of water) could still leave one off the intended operating temperature at the 3d printing operating temperatures

edit: as it turns out there are thermocouple available for rather low cost, these has little issues working at the elevated temperatures
and thermocouple being linear would i guess provide a way to calibrate the thermistors or even perhaps replace the thermistors
[en.wikipedia.org]
[www.ebay.com]

Edited 1 time(s). Last edit at 07/17/2018 10:42AM by ag123.

Theremistor B values are typically quoted over a range of temperatures such as 25 to 85C, occasionally 25 to 100C but rarely more. The beta model of a thermistor is OK for this limited range of temperatures, but it's inaccurate over a wider range of temperatures. So knowing only a B value for an extruder thermistor won't tell you with sufficient accuracy what the resistance is at typical extrusion temperatures. For that you need either the resistance-temperature table published by the manufacturer, or the Steinhart-Hart model coefficients for the thermistor.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].