RESOLVED. WAS: thermistor bug in 3.2beta 2 on Duet Maestro
-
There is no chamber heater, just the sensor, the chamber is passively heated by the bed. Note that the corresponding M950 is commented out.
-
Are you sure the new heater cartridge matches your PSU voltage? 12v cartridge on 24v power?
What were the heater faults you experienced? Error messages?
Can you post the results of M122 and M98 P"config.g" and your full config.g?
-
The cartridge is not the issue, original e3d 40W 24V. Remember, the issues disappeared with the downgrade to RRF 3.1.1.
I attached the config.
config.g.txt -
@Phaedrux said in WARNING! thermistor bug in 3.2beta 2 on Duet Maestro:
What were the heater faults you experienced? Error messages?
Can you post the results of M122 and M98 P"config.g" -
The errors were too slow a heatup. I have to put the board back to 3.2 beta 2 for the commands first, give me a moment.
-
Too slow to heat would indicate PID tuning hasn't been applied.
-
Ok, now I see the same issue also on RRF 3.1.1
The thermistor readout becomes strange at around 200°C, until then it is fine and follows what my multimeter reads. Above 200 it saturated while the multimeter sees temperatures up to 280 where I then pulled the plug.
However, there is a chance that the high temperatures damaged my thermistors here.
-
The graph definitely looks odd.
The multimeter temp probe can be drawing heat away from it. How does the temp graph look when left alone?
Do you have any spurious M307 commands left over from before that could be feeding it incorrect PID values?
This is the only report of something like this happening so I'm doubtful that it's related to 3.2 beta 2. Correlation != causation situation.
When in doubt I like to use a default config from the config tool as a sanity check to eliminate any config errors. So if you could do some tests with a stock config from the tool on 3.1.1 to see if the behaviour remains that would be helpful. then we know it's hardware.
-
And now it’s working again, within 3°C of my multimeter, only change was B3950 in the thermistor definition. But I still get some funky business for a moment during heatup.
Some hardware fault? Noise?
-
@pixelpieper said in WARNING! thermistor bug in 3.2beta 2 on Duet Maestro:
Some hardware fault? Noise?
Definitely possible.
-
@Phaedrux said in WARNING! thermistor bug in 3.2beta 2 on Duet Maestro:
This is the only report of something like this happening so I'm doubtful that it's related to 3.2 beta 2. Correlation != causation situation.
Sure, but that was the biggest change in the recent days. Besides a) installing the display and b) adding a front switch for 230V. The latter could in theory increase the 50Hz noise, but I think the cable is not close to the thermistor.
When in doubt I like to use a default config from the config tool as a sanity check to eliminate any config errors. So if you could do some tests with a stock config from the tool on 3.1.1 to see if the behaviour remains that would be helpful. then we know it's hardware.
I will try tomorrow, even so I doubt it - used this config for like 4 months without problems. Have two more unused thermistors here and an other Maestro, so I have some more debugging options...
-
Well if there's still off thermistor behaviour happening with 3.1.1 then I think we can rule out 3.2 beta2.
If there were other wiring changes around the same time...
-
I have the habit of always taking a picture when changing things, nothing of the changes is close to the thermistor cable.
-
I meant from changing the heater cartridge. Could there have been some damage to the thermistor wiring?
-
Unlikely, I swapped the whole hotend and I have seen the issues with both now. Also, the Molex MicroFit connectors are very reliable. And in the drag chain I have Heluflon Teflon coated cable which is well suited for this application... - will anyway replace the drag chain cable tomorrow and twist the thermistor wire.
-
Sorry for the confusion, it seems I might have had a wiring or crimp issue which by chance developed in parallel to playing with the beta. Having it seamingly disappear while downgrading did not help either...
I will replace all my hotend thermistors with PT1000, a wire fault resulting in to low a temperature reading is to dangerous in my opinion, so the upgraded sensors will give me some peace of mind.
-
@pixelpieper thanks for confirming the cause in the end. And also thanks for taking the plunge and testing 3.2 beta!
