Unstable temperature readings during heating



  • I am having issues with the bed temperature being unstable and erratic, primarily during heating - once up to temperature it's quite stable.

    0_1524135098144_Screenshot 2018-04-19 20.41.51.png

    It varies by a significant amount - 5 degress or more, seemingly randomly (ignore the hotend temp it's just cooling down in this instance, but is stable and smooth - and is on a PT100 daughterboard).

    This happens on two different beds with similar heaters using "Typical Chinese bed heater thermistor" - one I've had for several years, the other is brand new but has the same issue, so I am presuming the fault lies elsewhere. The heaters are silicone 24V and 250W / 225W respectively, and both attached to aluminium plates. The thermistors are embedded in the silicone plate so not easy to replace. I do have a third bed that is borosilicate glass with the same 250W heater as the order aluminium bed but I have not tried that in some time.

    While it's reporting these unstable temperatures, the bed heater LED on the Duet Ethernet is on constantly - it's configured as PWM but I have had it as bang-bang with no apparent difference. If the bed is cold, it's stable, and if it's at 60C or 100C it's stable, but even changing it from 60C to 65C will create a random 'blip' where the temp is reading erratically before it stabilises.

    The unstable temperature readings are causing other issues - I frequently get 'heater fault' errors when trying to reach 100C, as the 225W heater is barely strong enough to reach it (15m warmup), and with the temperature jumping all over the place I think the Duet just gives up.

    I had previously tried wrapping the thermistor cable a couple of times through a ferrite core, but it didn't seem to make any difference. I don't have it on any more.

    Any suggestions on what might be causing the issue, as I've run out of ideas?


  • administrators

    It sounds to me that either there is a bad connection in the thermistor wiring, or there is leakage between the thermistor and the heating element.



  • @dc42 I had thought the same thing regarding wiring initially - I re-crimped the original bed, and it made no difference. I crimped the second bed's thermistor wiring myself as well - so maybe I'm REALLY bad at crimping (but I don't think I am - I'm using an Engineer PA-09 and have done a fair amount of crimping).

    How can I test for leakage? I would have thought if that were the case then it would be a problem all the time, (including once it's reached temperature) - this is really only an issue during heatup. As an example, here is the temp graph from the end of a print just completed and everything is cooling down.

    0_1524145082820_Screenshot 2018-04-19 23.37.14.png

    Once it's reached 65C it's very stable, but you can even see during cool-down that there are small blips - at a time when there is no current going to either the bed or the hotend or anything else for that matter. Incidentally the original graph is from the heatup phase of this same print. There were further temperature blips after I took this screenshot.



  • Could it be a power supply issue? The heater may draw too much power?


  • administrators

    Yes it could be a power issue. Does DWC show a stable and sufficient VIN voltage during heating up?



  • My power supply is 17A (400W) and I have a power meter on it showing draw during heating as ~10.5A. Vin looks stable at 24-24.1V during the unstable temperature readings (confirmed on both the power meter and the DWC/Paneldue). Incidentally I think the DWC "hover over" text is incorrect for the voltage - it is displaying oC instead of V for min/max.

    If it was a power supply adequacy issue, then I would also expect unstable temperatures once at 100C - since the heater is underpowered, it is on almost constantly to try to keep it at 100C.


  • administrators

    @zylantha If you set it to (say) 30C once it reaches that is it stable?



  • This isn't my field of expertise at all but I doubt if the bed temperature is actually changing that fast due to the thermal inertia that it must have. It looks more like a noisy signal to my untrained eye.



  • @deckingman I agree - it's not a correct reading; this bed is very thick, 7mm cast aluminium plate (Geckotek 310mm round magnetic), however my other aluminium plate with the same issue is only 3mm.

    If it's a noisy signal (and I had drawn that conclusion too - hence my experimenting with a ferrite core), the first weird thing is that it's not noisy once it has reached temperature - why would that suddenly stop it from being noisy?

    Where is the noise coming from? And even then, what can be done about it ... I tried the ferrite core - wrapping the thermistor's leads around it twice, but it didn't seem to do anything; should I try putting a ferrite core around the bed leads? One bed lead? Both together? Or just one of the thermistor leads? I'm not an electrical engineer so am really just poking in the dark here ...

    I am eventually going to be replacing the heater with 240v (since it's so underpowered with 225w) - I don't know if that's going to help but it will eliminate some variables. It's dependent on Aliexpress lead times at the moment, so could be a long way away still.

    @T3P3Tony here is it getting up to 30C from cold (dead stable), lots of spurious readings during heating up, and then stabilising at ~31.5C steady, similar to once it gets up to 60C or 75C or 80C and remains stable - but only once it's there.

    0_1524309371361_Screenshot 2018-04-21 21.14.22.png


  • administrators

    @zylantha one more thing to try is setting it to a temperature that your bed cannot reach (as long as that is safe) as dc42 said:

    Yes it could be a power issue. Does DWC show a stable and sufficient VIN voltage during heating up?

    If that continues to show the noisey profile when it is at a stable max temperature, limited by what the power supply can output then it could be that the power supply injecting noise when it is at its maximum current draw, but now when it is at < maximum



  • @t3p3tony said in Unstable temperature readings during heating:

    setting it to a temperature that your bed cannot reach

    The limiting factor is the Duet's propensity to return "heater fault" due to (what I presume is) the unstable readings messing with the dead time safety. I frequently get a heater fault when trying to reach 100C, when it is definitely capable of getting to this temp in 15 mins (which really isn't that awful). Sometimes I have to restart several times before it will get to 100C - but I have never had an issue with a heater fault error after actually reaching 100C.

    Some of the issues in getting to / maintaining 100C are caused by my stepper driver fan under the bed, which kicks in at 180C hot end temp to avoid the warnings about stepper drivers overheating - the extra air movement under the bed results in a significant heat loss and means the underpowered heater struggles to maintain the 100C (I have seen it drop to 92C) - in these circumstances, the temp readings are still stable, just lower than the target - during times when the heater would logically have full PWM.

    The noisy profile seems to only be during heat up / cool down (note that this includes cool down), but not while cold or while at temp.

    I will try to get it stable at 100C and then push it up to 110C and see what happens (this is something I haven't tried yet as I never thought it could get to 110C).



  • Try #1 to reach 100C .. fault after reaching only 50C ☹

    0_1524311104282_Screenshot 2018-04-21 21.44.32.png


  • administrators

    @zylantha you can use M570 to extend the time for heater fault detection so you can get to a higher temperature:

    https://duet3d.dozuki.com/Wiki/Gcode?revisionid=HEAD#Section_M570_Configure_heater_fault_detection

    However the noise on cool down means I am probably wrong with the idea that its power supply noise from being on full power. Still worth checking.



  • OK so in frustration at only being able to reach 50C before a heater fault, I re-installed the ferrite core, wrapping the thermistor lead around it three times (previously only did it two with the original bed heater - lead was too short to do it any more). I also rotated this new bed by 120o - since with the shortened thermistor lead it needed to be closer to where the Duet connectors are. This does result in a very different path for the 24V leads - away from the input leads, and the bed leads are in a completely different orientation as well (fwiw).

    Result ... a completely smooth and perfect heating profile
    0_1524312274072_Screenshot 2018-04-21 21.57.07.png
    0_1524312294391_Screenshot 2018-04-21 22.03.59.png

    I can't really accept that this was because I wrapped the thermistor leads through the ferrite core 3 times instead of 2, so looks like I have to do some isolation tests ... removing the ferrite core while leaving everything in its current orientation, looking at the power lead paths, etc.... it all seems very strange! 🙄

    Most obvious stupid cause is a loose thermistor connection that was solved by removing and re-connecting the bed temp sensor, but I will continue to refuse to accept that this is the cause as well (until all other less likely causes have been eliminated).


  • administrators

    @zylantha that does look a lot less noisy - what would be odd about it being a physical connector issue is that it was not noisily when at temperature, and was noisy when cooling down.

    That said you appeart o have fixed it which makes further fault finding more difficult 😉



  • @t3p3tony said in Unstable temperature readings during heating:

    what would be odd about it being a physical connector issue is that it was not noisily when at temperature, and was noisy when cooling down.

    That's what had me stumped ... it's been an issue for some time, and I thought I had eliminated the connector issue with all of the iterations of different thermistor cable connector that had exhibited the issue ... and I can't think of why it would only be unstable during heat-up/cool-down (but hey I'm a lawyer not an electrical engineer).

    I will continue to try to unroll these latest changes and reproduce the issue again and isolate the fix, if nothing else the thread may be useful for somebody else who may end up having the same issue, even though I seem to be the only one ever.


  • administrators

    Cheers for doing more work to isolate the issue. You are definitely not the only one to have noise interference with heater temperature readings, just the only one I know of to have such specific circumstances.


  • administrators

    @zylantha said in Unstable temperature readings during heating:

    Some of the issues in getting to / maintaining 100C are caused by my stepper driver fan under the bed, which kicks in at 180C hot end temp to avoid the warnings about stepper drivers overheating - the extra air movement under the bed results in a significant heat loss and means the underpowered heater struggles to maintain the 100C (I have seen it drop to 92C) - in these circumstances, the temp readings are still stable, just lower than the target - during times when the heater would logically have full PWM.

    It sounds that you have insufficient thermal insulation underneath the bed heater.



  • @dc42 said in Unstable temperature readings during heating:

    It sounds that you have insufficient thermal insulation underneath the bed heater.

    This is true. Unfortunately I haven't found the right case for the Duet to isolate the flow from the cooling fan, and have deferred the cork base until I receive the 240V heater - so resolution of this is in the works, however this isn't the cause of the erratic temp readings.


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.