Art of thermistor tuning...?

....Background....

Yesterday when I was printing, I found that the temperature reading jumps from 258 to 270, with nothing in between. (I know 258 is very inaccurate, it couldn't be that hot)

Later I found that the table generated by
./createTemperatureLookup.py --r0=6360 --t0=100 --r1=0 --r2=4700 --beta=3990 --max-adc=1023

Just do not give enough resolution at the first band.

Now I solder a 3k ohm in parallel with the 4.7k, making r2=1818.

Much better resolution.

Now I fire up my heater, no matter what it could only reach 215! Heat is losing way too fast. Previously the upper limit is 285.

I didn't change any other beta value. (The beta is just comes from the datasheet).

....Ask....

Oh well so now I fully understand what it means "...Although the properties of ABS are fixed, it’s hard to get a dead-on calibration of the thermistor..." [blog.thingiverse.com]

Perhaps I should go and buy a IR thermometer gun, I just saw one selling at ~USD20 (RMB 120). Though I just wonder, without that, what does everything one use to tune their thermistor?

Or people just go ahead and use thermocouple instead?

I'm going with thermocouple on mine, as thermistors aren't really meant to be used at the temperatures that being used in the RepRap project.

There is a simple kit, which you can pretty much free form anyways on veroboard, from MakerBot.
I haven't got my machine up and running yet so its just an idea.

I bought a cheap IR thermometer, but it's useless for checking the nozzle temperature, as the nozzle is small enough that it never illuminates the entire IR sensor, giving me highly variable readings that are always well below the real temp.

I've got a thermocouple for my multimeter that I use to check things, but I also just use thermocouples on my extruder. I didn't want to mess with temperature calibrations, especially since extruders are fairly sensitive to small changes in temperature.

One thing to check though, is noise. I find that when I have grounding problems anywhere between the thermocouple and the PC, my temperature readings start jumping around by 20 or 40 deg C. When everything is good, the noise is less than 1 deg C.

Wade

Oh. I didn't know that thermocouple output is linear. (or I actually read that but I forgot)

Now I am going to buy a thermocouple-K and AD595 and see how it goes.

For the thermocouple chip,
What is the difference between AD595CQ (Cerdip) and AD595CD (TO-116)?

They are just the same footprint, with one of them have a metal plate over it. Functionally what are the differences? Which one is better?

While I am waiting for the thermocouple and chip, I milled my first usable PCB!
[picasaweb.google.com]#
(Surely it could be faster and easier with wire-wrap board, but I don't have one now)

Edited 2 time(s). Last edit at 09/06/2009 01:51PM by sam0737.

Nevermind about packaging. I got it running...oh my, it's much easier than thermistor except that it requires one more little piece (and pretty pricey) PCB and a few wires.

I spent 15 minutes on the noise though. I forgot the motor is grounded to my toolhead, and the thermalcouple is not isolated. Then when I spins up the motor, the measured temperature jumps -50 to +50 around the real temperature.
It's solved by putting a slice of Kapton tape over the measuring point.

With a thermistor running side by side, I can see how much different there were...

Nice milled PCB! Yes, the thermocouple gives you one less thing to worry about. Another plus, I don't use welded thermocouples; I just pinch the two wires in between the nozzle and barrel. That way, if something comes loose, it gives me an out of bounds temperature reading, which triggers my firmware to shut down the extruder, preventing this from happening:

[builders.reprap.org]

Your all-metal extruder won't fail like that, but a malfunctioning heater could possibly burn some feedstock.

Wade

And I changed my heater to 3.9ohm wirewound resistor, down from 6.8ohm. The 6.8ohm can't give enough output to combat with the heat loss.

I think there are something we can do in the firmware to combat with loss thermocouple. Say if the temperature wasn't raised for X degrees C in Y seconds, something is going wrong (a loose thermistor/thermocouple) and hence e-stop should be triggered.

The thermistors specified are rated for 300C, so are perfectly good for extruding plastic around 240C. I don't use a lookup table though so I get results which track a thermocouple within two or three degrees.

I think both work equally well unless you want to go above 300C.

---

[www.hydraraptor.blogspot.com]