Struggling with PT1000 calibration

zemlin

First disclaimer - my reference device is a $30 crap thing from Amazon which uses type-K thermocouples. That said, checking with ice water and boiling distilled water the readings were where they should be.

I have three machines running Duet2 Wifi boards. I have been running cheap thermistors for the extruder temps, but I'm trying to get my machines more in sync - and extruder temps were not reading consistent across them so I changed one machine to a PT1000 from Slice to see if I can get readings to match my thermocouple thermometer. My machines are Raise3D N2-series printers and they have a long heat block. With the short Slice sensor I'm able to slide a 3mm diameter Type K probe well into the heat block and get a good temp measurements. I also tested with a bare TypeK junction shoved into the melted plastic in the hot end and that reads the same as the probe in the heat block, so I feel confident about my measurement technique.

I looked up the calibration processes. I have a 2000 (.01%) ohm resistor, which should compare to 266.4C on a PT1000. I also have an 1096 ohm array of resistors that should compare to 25C on a PT1000 and 1000 (.01%) resistor which should compare to 0C.

The semi-auto isn't an option on the Duet2 Wifi board. Tried the Manual calibration with resistors. That isn't calibrating the sensor, but the circuit. Are the PT1000 sensors supposed to be consistent enough that sensor-to-sensor differences can be disregarded?

If I put the 1096 ohm resistor on the E0 connector on the print head I get a reading of 25C with H15. With the 2000 ohm resistor I get 266.4 (or so) with L-2.

If I then hook up the PT1000 and set the heater temp for 270, the thermocouple reads 280. L-97 brings the PT1000 temp into line with the thermocouple. I then set the temp to 50C. With H127 the thermocouple reads 40C. I can't get an H value to make the PT1000 and thermocouple match.

The numbers I get with the resistor tests aren't correlating to an accurate temperature reading, and I have been unable to tweak the H and L numbers to get correlation to the thermocouple readings.

What am I doing wrong here?

moth4017

@zemlin
hi im using a pt1000 had no issues but make sure your thermo couple is clamped to the heat block just placing it on the side is not good enough .

zemlin

@moth4017 it’s in a deep, close-fitting hole in the heat block.

deckingman

@zemlin Any noise on the signal? Are you using 2 wire or 4 wire? In general, thermocouples are less accurate, less stable and more inclined to drift over time compared to platinum resistance sensors. So maybe the PT is correct and the error is with the thermocouples?

zemlin

@deckingman said in Struggling with PT1000 calibration:

@zemlin Any noise on the signal? Are you using 2 wire or 4 wire? In general, thermocouples are less accurate, less stable and more inclined to drift over time compared to platinum resistance sensors. So maybe the PT is correct and the error is with the thermocouples?

Two wire, but all looks stable. No sign of noise. The thermocouple being off is a possibility and the reason for my noting the dubious nature of my temperature "standard" in the message - but as I said, I did confirm it at 0C and 100C, so it seems to be OK.

What do folks use a temperature standards for confirming extruder temps? Not keen on spending $400 for a Fluke meter.

deckingman

@zemlin said in Struggling with PT1000 calibration:

What do folks use a temperature standards for confirming extruder temps? Not keen on spending $400 for a Fluke meter.

TBH, I don't get too exited about trying to achieve the ultimate in temperature reading accuracy for several readings. Mainly that the reading is associated with the temperature sensor, the position of which my be some distance away from the temperature that we really care about, which is temperature of the filament inside the nozzle. This nozzle temperature can often be affected by things like deflected part cooling air which might not be replicated by the sensor in the heater block. The there is also usually quite a wide temperature range that filaments can be extruded at maybe +/_ 5 deg C or even +/_ 10 degrees. For these reasons, I always print temperature towers to determine the best temperature for the filament/hot end combination that I'm using. If I find that the best temperature is say 250 deg C +/_ 5 deg C, as indicated by whatever sensor I'm using, then that's the temperature that I'll use even though it might be a few degrees different from a lab grade certified standard.

dc42

@zemlin it's as @deckingman says.

@zemlin said in Struggling with PT1000 calibration:

If I put the 1096 ohm resistor on the E0 connector on the print head I get a reading of 25C with H15. With the 2000 ohm resistor I get 266.4 (or so) with L-2.

H15 and L-2 are within normal tolerances. L-97 is well outside the normal range.

If you are sure of the accuracy of those resistors, then RRF is working as it should. Either your PT1000 sensor is inaccurate, or your thermocouple sensor is inaccurate, or you have not succeeded in getting those two sensors at exactly the same temperature.

zemlin

@dc42 said in Struggling with PT1000 calibration:

If you are sure of the accuracy of those resistors, then RRF is working as it should. Either your PT1000 sensor is inaccurate, or your thermocouple sensor is inaccurate, or you have not succeeded in getting those two sensors at exactly the same temperature.

I have far more faith in the accuracy of the resistors than the accuracy of the thermocouple readout. The 2000 ohm is a pair of 1K .01% resistors in series. The 1096 is a 1K and 100 ohm resistor - but I sorted through a number of 5% resistors to find a pair that added up to 1096. Checking these with a good Fluke digital meter. I have a couple more Slice PT1000 sensors on the way. If they are in line with the other, I won't sweat it. Having them consistent across several machines is the goal.