Highly erratic temperature readings above certain temperature

  • Hi forum,

    first of all let me say thank you to all the help you've unknowingly provided to me just by running this forum, especially to dc42, you're doing an amazing job at supporting the Duet and everything around it.
    I've just about finished building a printer around the Duet Ethernet, and it actually earned me my master's degree in engineering just yesterday. It's a high temperature machine, thus i've had to account for insulation, heating of the build volume, cooling etc. in rather unusual ways. It's the second printer I've built, first was around a smoothieboard but Duet seemed to be the way to go for #2, and yes it was!
    I've also worked as a development engineer in additive manufacturing, so I more or less know my way around FDM/FFF and also DLP printers.

    My problem today is one that I've not been able to solve by browsing the forum. Unfortunately it comes at a highly unfavorable time as we need the printer to be up and running in the coming weeks.

    My setup is "E3D everything" -- standard v6, 40W heater, copper block, copper nozzle, silicone sock, type K sensor with matching daughterboard on the Duet. Additionally, there's a PT100 daughterboard which monitors the heatbed and enclosure heater. The enclosure will reach well above 100°C and is suited to do so, so nothing to worry about there. The leads from the print head are guided in glass fiber sleeving. I've set this up just this morning, so the parts aside from the heater were all new (I changed out a PT100 sensor from the hotend as its insulation started to get crispy at around 400 °C and just crumbled away - you shouldn't use these at this high a temperature, it seems).

    My problem is as follows; temperature will rise just fine on the hotend up to about 280-290°C. From then, readings will start to jitter and then flip out, giving me the occasional 2000°C which means a short to something else (iirc).

    0_1523959720112_heater fault 1.PNG

    so the heater goes into fault, from then the reading will slowly and jittery creep to the actual temperature and just be fine again (see above).
    So i cleared the fault, tried again and well, it did the same thing.

    1_1523959720112_heater fault 2.PNG
    (second and third time)
    I then tried to tune the heater to 350°C which did not fit the controller model as it seems;

    4_1523959720113_heater tuning 2 - fault.PNG

    tried again at S250

    5_1523959720113_heater tuning 3 - succ.PNG

    which worked okay, but i could see what i suspected until now, that the misreadings are somewhat in connection to the heater being on or off -- as you might be able to tell from the S350 run, as soon as the heater is turned off in the graph above readings become jittery for a few seconds and then smooth out.

    I decided to give it a go and preheated the machine, the chamber heater's reading is straight from the heater itself, don't be alarmed! Actual chamber temperature is dotted red line.

    2_1523959720112_heater fault 3.PNG

    What puzzled me is that temperature drops to "straight" values such as 300°C, 250°C while it's spazzing out (see above). Next try to no avail:

    3_1523959720113_heater fault 4.PNG

    And again what's highly curious is that the reading slowly creeps up to the actual temperature after it's been all over the place. I can't imagine that the sensor would change its resistance in this way.

    The heater faults to "short to other wiring" which I didn't check yet but find rather unlikely as I've just put this all together. I've checked that the leads of the heater and the sensor don't touch the cartridge because that would cleary mean a short of sorts...
    I'm just puzzled by the repeatability of the thing, that it would start to spazz out just below 300°C...

    any pointers?

    thanks in advance for your time!

    Best regards, Niklas

  • I was having vary similar issues as this with a E3D sensor, it was not always a spike but a 20-30c error at times then other times it would spike to the max. Replacing it solved all issues even though it was a new barely used sensor.

  • I had similar issues with my E3D PT100 sensor.
    Had to remove the molex connector and crimped a new one and now its working fine.

  • Hi all,

    thanks for your remarks and sorry I didn't get back to you earlier. I was not able to fix the issue until now, but have a strong lead regarding the actual problem.

    First thing I did was replacing the thermocouple with an identical new one. The same problem occured, at the same temperatures. I then tried and grounded the print head structure directly to the printer's frame and surely, the "fault" LED on the Type K Daughterboard would light up whenever the ground connection was established above 300°C. This is usually not the case as I don't yet have a ground wire and am also using polymer bushings which effectively cancels out all connection to ground. In short - above 300°C, the Type K thermocouples' cartridges insulation fails and shorts the sensor tip to the casing.

    I then contacted the reseller in Germany who supplied all of my Duet and e3d hardware; according to him, the cartridge type thermocouples aren't really made for such higher temperatures. He suggested that the insulation fails at higher temperatures leading to the behaviour described above. I felt a little misled as in my understanding from the description on the e3d wiki, the Type K thermocouples should be safe to use even after the PT100 ones fall short. That seems to be true for the non-cartridge types anyways.

    So, I'm waiting for both PT100 and Type K cartridge type sensors from a german company at this moment. As soon as I'll be able to (hopefully successfully) install these I will get back to you and share my findings.

    Best regards, Niklas

  • administrators

    Thanks for your report. I think the insulation between the thermocouple tip and the cartridge casing is indeed breaking down. Your test of grounding the metalwork appears to have proved that. Without that ground connection, the leakage current between the heater and the hot end metalwork will then feed into the thermocouple.

    I was unable to find a rated temperature for the cartridge thermocouple on E3D's web site.

  • @dc42 thanks for your reply. Same for me - i wasn't able to find any datasheet or temperature ratings. The reseller however seemed to have datasheets and told me to use PT100 at the temperatures I aim for, however since the e3d ones don't hold up above 400°C (and apparently, alread much more below) I now ordered industrial types.

    Lawson at e3d told me the PT100 sensors (cartridge type) "should be fine up to 450°C" but with greatly reduced service life. I'm not sure if I just had a bad batch or whatever the problem with my sensors was... but it should be solved in the future.

    "Without that ground connection, the leakage current between the heater and the hot end metalwork will then feed into the thermocouple." thanks for the insight - if this explains the weird behaviour before and after the excessive misreadings i'd be happy because it means i've not gone insane 😛

  • Hi all,

    as i mentioned before i orered new sensors to replace the e3d cartridge type ones. A german shop i found stocks both PT100 and Type K sensors in cartridge configuration:
    PT100 click
    Type K click

    the only drawback is that the Type K variant is only available as 4 mm diameter. They are not even expensive however and (to my experience) of much higher build quality than the e3d types. The lead length is variable and they feature ferrule ends matching the Duet Daughterboards perfectly (as opposed to the e3d types which have to be fitted). Last but not least they are specifically rated for 400 °C and 550 °C respectively. I am running the PT100 type right now and am finally able to print again.

    Best regards, Niklas