I seem to have overcome the temperature problem. Once again I connected the sensor directly - the problem disappeared.
I connected it back in the usual way and everything worked - the error was within half a degree.
Not to sound brash but did you read the reviews for the T-D500 from Trianglelab? Last I checked the reviews on Ali, people had nothing but problems getting that thermister working properly with Klipper, so this isn’t just you more than likely. I don’t think that sensor is even in Klipper without defining a custom thermister
My printer was originally equipped with this thermistor. There were no problems on the marlin.
Judging by my measurements at temperatures above 100C, it shows quite accurately.
I also found the configuration on the Klipper without any problems
My Trianglelab T-D500 Klipper config
[thermistor D500]
temperature1: 100
resistance1: 147700
temperature2: 200
resistance2: 8000
temperature3: 450
resistance3: 146.9
Correction: “There were no reported problems on the marlin.”
I’m sure you’re feeling quite beat up and not that confident in Klipper but there’s a hardware issue here. It could be the thermistor (as @nathan22211 reported), it could be wires that are breaking internally or it could be something like a noisy or poorly output filtered power supply (which I was wondering about originally).
Question, when you were using Marlin, were you using the firmware that came with the printer? If it was, then I would be very suspicious if that was the case as the manufacturer’s version of the firmware may have been altered to mask the thermistor instability when the toolhead is moving.
I believe your printer has a Robin Nano V3 board (I’m very familiar with that). Could I ask you to find a cheap NTC 100k 3950 thermistor and mount it on your toolhead and wire it to the unused thermistor input (the connector is marked “TH2” on the board and is wired to pin PA2) and run the printer?
Doing that should help clear up this impass.
Now the problem has disappeared - I printed the part and throughout the printing process I did not see deviations of more than a couple of degrees when the airflow was turned on.
It was probably a problem with one of the contacts. Without blowing, deviations did not exceed 0.5C