Incorrect temperature reading from Vertex (K8400) thermistor

I did some temperature measurements on the Vertex since the temperature appeared to be to high when printing PLA. The temperature was measured using a K-type thermocouple and a multimeter. The sensing tip was placed in the same hole as the thermistor mounted from the top side. To ensure a correct measurement a heat conductive (silicon) paste was applied. Since I do this kind of measurements at work I am pretty sure that the readings I get from the thermocouple would be accurate within one or two degrees.

The next step was to set the machine temperature to different values using G-code. There was no printing or extrusion at the time for the measurements. For each step I waited until the temperature was stable, approximately 30s.

The results are listed in the table below. There are three columns:

  • The first column show the reading on the Vertex
  • The second shows the reading from the thermocouple
  • The third shows the difference between the two values

The results show:

  • that the temperature is approximately 50C too high at a set point of 200C
  • The error is increasing linearly. If you plot the values in excel it is shown clearly.

Possible causes for the error could be:

  • Defect thermistor
  • Incorrect interpolation table in Marlin
  • Resistance in the connection board (U-shaped PCB close to the extruder)
  • Resistance in the cables
  • Something on the motherboard

If anyone have the possibility to measure the resistance of the

  • Disconnected thermistor
  • Full chain from the thermistor down to the connection the motherboard

it would be helpful.

[ul] Machine_reading K-Thermocouple Difference
25 24 1
34 36 −2
36 39 −3
44 49 −5
55 63 −8
75 88 −13
94 111 −17
113 135 −22
132 158 −26
152 183 −31
170 209 −39
190 233 −43
200 249 −49
210 259 −49
230 285 −55[/ul]

Best Regards,
Urban

I did a measurement of the thermistor

  • lose and
  • at the connector at the motherboard when detached from the motherboard

In both cases the resistance was 94,6k Ohms. The thermistor was at room temperature.

Does anyone have a measurement to compare with?

The value of your thermistor at room temperature is within the specs. (100K @ 20°).

The inaccuracy or the component vary from brand to brand but normally doesn’t exceed 10%. From your readings, the difference is way more important (25%)!! I’m afraid all Vertex are affected but without a thermocouple it’s almost impossible to get an accurate measure (I have an IR thermometer with a laser pointer but the surface is too small and the reading is not reliable).

I take some mesurement

Around 200°c termocouple reading about 10°c higher than vertex annoucement. At room temperature it’s quite the same thing.

Dear Urban,

I was troubleshooting with PVA on the K8400, I didn*t got a reliable printing. Instead, there was a crackling noise when heating up the extruder; finally I tried a similar procedure like you did: I measured the temperature.

Using a thermocouple I checked the temperatures from the outside first, which were already too high.

Finally I placed the thermocouple directly in the nozzle head (where the filament is liquefied)

For a set-point of 210 °C I got a measurement of 265°C on the thermocouple with fan switched on.

A second set-point of 150°C gave the following readings:

180°C with fan switched off,
190°C with fan switched on.

This proves my impression that with running fan the extrusion temperature is higher.

My explanation for this behavior: a systematic construction error.

The temperature sensor in the K8400 design is protected by a temperature resistant isolator. Even with a proper alignment of the sensor within the mounting hole the heat flow through the isolation is too low to compensate the cooling of the connection wires. This produces a temperature drop across the isolation material, which will be impacted by the position of the temperature sensor in the assembly.
With the fan working the wiring is cooled, therefore the heater must set a higher temperature to provide sufficient heat to the temperature sensor to compensate the cooling of the wires.
Finally, if the temperature sensor is not mounted well in the hole, the temperature drop will increase dramatically leading to a damage of the PEEK parts too, as it was reported some times in the forum.

Actually I*m thinking of an improved heating block, I will come back when my ideas are running well.

Nevertheless, I believe, this is also a task for Vellemann to redesign this heating block.

Dear hoho61,
A late thank you for the answer:) I solved my problem by changing to two E3Dv6 and the temperatures are better now. I also added a heated platform and a new power supply, added a power switch and changed to SilentStepStick.
Best Regards,
Urban

Thanks for the very informative post that I unfortunately skipped while fighting against temperature problems. That I solved in a different manner: http://goo.gl/6hgKBL

I am afraid one day or an other we will have to get down to the code and find a way to calibrate these sensors in order to compensate their non-linear responses.

Nice reading about temperature measurements: http://goo.gl/45EIiK