Hotend Thermistor temp vs. Nozzle temp



  • I recently converted to duet (both the printer and myself are now believers!), and noticed my B and C values for the thermistor were off.

    No big deal, right? I updated them and just thought to myself... hmmm... let's just verify the temperature is correct for a second.

    grabbed my thermocouple (Uni-T multimeter) and probed the nozzle of a hotend that has been soaking at 240c for a few minutes.... thermocouple reports ~185c !!

    I probed around and found that I could perhaps find a point with 190c but not more.

    Verified thermocouple is accurate using boiling water.

    I played around with the calibration and ended up manually increasing the B value until the nozzle temp was close enough to the reported temperature in DWC.

    Started with:

    M308 S1 P"e0temp" Y"thermistor" T100000 B4725 C7.060000e-8
    

    and ended up with:

    M308 S1 P"e0temp" Y"thermistor" T100000 B5400 C7.060000e-8
    

    Some questions:

    • How likely is it that this is just a weird thermistor?
    • Is there a list of some common B and C values for thermistors?
    • I have very long wires to the thermistor (~2m... big delta with "neat" wiring), could this affect the result? I measured ~99.5kohm at room temperature (~28c).
    • Would be extremely interesting if forum members would be willing to share their DWC temp vs. nozzle temp!

    Looking forward to learn from your collective experience!



  • @Leav do you actually know what type of thermistor you have?
    Is it stock to the machine? If so, what machine is it?



  • @jay_s_uk
    Originally it was an E3Dv6, but I can't remember if the thermistor has survived all the periodic maintenance sessions...
    I'll admit I was naive and thought all "3D printer thermistors" were more or less the same... 🤦



  • @Leav
    They are all wildly different... not just between different part numbers. Even supposedly identical thermistors can be different by 20 degrees at 200C.

    Add in variation in mounting...

    For example, are you just touching that thermocouple? or its it clamped in place? Heat transfer is stongly effected by pressure at the boundry, so bolting a sensor on will give a more accurate reading than touching or even taping a sensor.



  • @theruttmeister said in Hotend Thermistor temp vs. Nozzle temp:

    They are all wildly different... not just between different part numbers. Even supposedly identical thermistors can be different by 20 degrees at 200C.

    Amazing! then how are you supposed to use them?? I've never seen anyone mention a calibration procedure....

    For example, are you just touching that thermocouple? or its it clamped in place?

    I did just push the thermocouple against a corner (block and nozzle), but I think the 50 degC is too much to chalk up to contact difference....
    My thermistor is in the dedicated hole with some thermal paste to improve heat transfer somewhat (old e3d design with non-cartridge thermistor)

    [blah](http://Blahblahblah-lots-of-filler-text-so-the-spam-filter- let's-me-pass.com)


  • Moderator

    @Leav said in Hotend Thermistor temp vs. Nozzle temp:

    Amazing! then how are you supposed to use them?? I've never seen anyone mention a calibration procedure....

    This is why printing temperature towers is common as there is no way to guarantee the temperature displayed is accurate to the temperature at the thermistor, let alone at the nozzle several mm away.

    Even if you use a PT100 which should be very accurate to the requested temp, the different between the thermistor and the nozzle tip can vary substantially.

    @deckingman has done some impressive detective work with thermocouples stuck down inside of the nozzle during his development of his mixing hotend. Check out his blog for the results https://somei3deas.wordpress.com/



  • @Phaedrux said in Hotend Thermistor temp vs. Nozzle temp:

    This is why printing temperature towers is common

    Interesting, I always just assumed it was because of variations in filaments.

    I wonder if PT100/PT1000 or thermistors are more uniform... (I assume PT100/1000 have to be by definition?)


  • Moderator

    Yes PT100 and PT1000 are far more reliable. PT1000 is a good option since you don't need a daughterboard and for printing purposes are more than enough. PT100 is overkill and you do have to be a bit careful with wiring. Using a 4 wire lead is basically required for error rejection.



  • @Leav said in Hotend Thermistor temp vs. Nozzle temp:

    @theruttmeister said in Hotend Thermistor temp vs. Nozzle temp:

    They are all wildly different... not just between different part numbers. Even supposedly identical thermistors can be different by 20 degrees at 200C.

    Amazing! then how are you supposed to use them?? I've never seen anyone mention a calibration procedure....

    Industrially, if you want accuracy, you build a test rig, using something like a PT100 or a thermocouple as the reference. And you throw away the ones outside of your required tolerance.
    Or you buy more expensive thermistors that went though that at the factory (but that quickly gets expensive).

    Otherwise, you do some prints, see if they look a bit melty, or de-laminate... adjust the temp accordingly. Unless you are building a lot of printers, that's close enough.

    For example, are you just touching that thermocouple? or its it clamped in place?

    I did just push the thermocouple against a corner (block and nozzle), but I think the 50 degC is too much to chalk up to contact difference....
    My thermistor is in the dedicated hole with some thermal paste to improve heat transfer somewhat (old e3d design with non-cartridge thermistor)

    That could easily account for 50C difference. The thermocouple wires are a big heatsink (compared to the junction) and as a result you are going to get a reading somewhere between the actual temp of the target and air temp. This is why its ideal to stick it in a hole.

    If you want an accurate reading, stick the thermistor in boiling water (you don't need to for thermocouples, they are inherently accurate to +-3 degrees). You can use that to verify you are using the right settings, although thermistors get less accurate as they get closer to their max temp (assuming its an NTC), so your readings at 220C might still be off.

    Personally I just use PT100 sensors. Much more expensive than thermistors, but still not that expensive in the grand scheme of things and totally worth it.



  • @Leav said in Hotend Thermistor temp vs. Nozzle temp:

    I wonder if PT100/PT1000 or thermistors are more uniform... (I assume PT100/1000 have to be by definition?)

    PT100/1000 sensors are defined by DIN/ISO standards, so yes 🙂 . IIRC, the lowest grade will be accurate to +-0.5 degrees C at 200C, from device to device, regardless of manufacturer.

    The rule of thumb I was told is:
    Thermistors are the cheapest.
    Thermocouples have the biggest range.
    RTD's (PT100 type devices) are the most accurate.


Log in to reply