Hot end thermistor reading double in Pronterface?

Hi all,
I've had my Hophead kitted Mendel 90 a while now and I really like it. I thought it needed a sister and a friend had a load of dibond kicking about so I built another one using the first! Yay!

I have one small issue though that I hope y'all might be able to advise on.
I've used Nophead's version of Marlin and that all seems to work very well except for one thing, pronterface is reporting the hotend temperature at pretty much bang on double what it actually is. If pronterface reports the temp as 60c it would measure (using an IR thermometer) 32, at 90c it measured 46, at 200c it measured 98 or so. The hotend is a pukka J-head from hotends.com, the only thing I'm not certain about though is what thermistor was supplied with it. I think I recall when I bought the hotend a while back that they were talking about having to use a new thermistor since the previous one had been discontinued so I think it's probably a Semitec 104GT-2. Would anyone know how to identify it?
The Melzi I have is a cheap(ish) one from eblag and the (pull up is it?) resistor on the board next to the hot end temp terminals is 4.7k. I'm not sure if the Melzi on my "trueblood" Mendel 90 uses 10k ones and that might account for the difference. I can't find anywhere in the Nophead's version of Marlin to account for this.
I did try using a different version of Marlin where this was something I could account for but it made no difference. I also, just for the hell of it, tried using a 200k thermistor value and that made no difference either. I found a small section to do with temp adjustment (Gain and Offset IIRC?) and put in some random figures and that too made no difference. I'm beginning to wonder if I've not made the changes properly? I'm no software person by any means! I measured the value of the resistor and at 23c it's 113k.
I understand I may need to make a new thermistor table? but that seems like quite a bit of time when the issue just looks like some factor of 2 somewhere that I'm missing!
Oh, if it is of any relevance, the bed temp is bang on.

Anyone got any ideas?

Cheers
D


Edited 1 time(s). Last edit at 12/27/2014 12:31PM by maddog7.

The Melzis we supply have 4K7 pullups for the thermistors and our hotends still use the Epcos thermistor. Other sources including hotends.com have switched to Semitec but it is only 3%, so I don't consider it suitable.

You can't read the temperature with an IR thermometer. It will always read low because its aperture is too big, they are not calibrated for high emissivity of metal surfaces and the outside of the heat block is cooler than the inside of the barrel anyway.

The only way to get an accurate reading is with a thermocouple inside the barrel.

---

[www.hydraraptor.blogspot.com]

Cheers for that Chris,
I don't have a thermocouple so I'm a bit knackered there for now. Would you think it possible/likely that the emmisivity of the metal would be such that the measured temp would be pretty much bang on half what pronterface reports? I suppose I could start on the high side of pla extrusion temp and reduce it 10 deg at a time until the extrusion starts to spiral?
Do I presume by 3% that you mean it's accuracy?
Cheers D

As a very rough guide (to see if your thermistor is reporting in the right ball park) try heating the hotend to 200C, then try to feed some PLA by hand. If plastic comes out of your nozzle then I'd say Pronterface is closer to the mark than your IR thermometer.

It would seem (and I'm in not the least surprised! ) that Nophead was correct. Though It is a little odd that the IR gun would give bang on half.
Nonetheless It appears to be all sorted now, might get to try a test print in the morning.
Thanks to you both for your replies!

D