Thermocouple temperature stability

Discussion in 'General' started by brucehvn, Aug 4, 2019.

  1. brucehvn

    brucehvn Member

    Joined:
    Jul 14, 2014
    Messages:
    6
    Likes Received:
    0
    I was having big problems with the thermistor cartridges for my e3d, I was getting wild temperature swings during printing, bouts of max temp shutdowns, etc. So I bought the type k thermocouple and the board from Filastruder. I installed that and while the temperatures are usable and I'm able to print, the temperature swings are about +/- 2 degrees meaning a range of 4 degrees. Even when the printer is idle, and not printing anything, I'm still getting variations in temperature. I've changed out every piece of the electronics and the printer itself has been completely rebuilt and now uses a Titan extruder with the e3d.

    I guess my question is this normal behavior for the thermocouple. I was bundling the thermocouple wire with the other wires going to the extruder, but then I saw this Stack Exchange post and I then isolated the thermocouple wire so it doesn't travel anywhere near any of the other wire bundles. The first graph below is when the printer is idle, the 2nd is during printing. There was a time when my hotend was just as rock solid as the bed with only about a +/- .1 degree variance, but I haven't seen that for a long time now.

    [​IMG]

    [​IMG]
     
  2. Old_Tafr

    Old_Tafr Well-Known Member

    Joined:
    Mar 7, 2016
    Messages:
    711
    Likes Received:
    91
    Have you done the PID tuning ? Also is there a draft blowing on the printer from say an open door or window?

    The info you quote seems really good and informed.
     
  3. brucehvn

    brucehvn Member

    Joined:
    Jul 14, 2014
    Messages:
    6
    Likes Received:
    0
    Yes, I forgot to mention that. I've done the PID tuning several times. Also there are no drafts on the printer and I don't have a cooling fan on the hotend. I thought maybe the location of the thermocouple board might be getting some draft from a cooling fan on the electronics, so I slipped an anti-static bag over it so it wouldn't be adversely affected. That StackExchange post was really informative. I never had any idea that thermocouples were so sensitive to magnetic fields, etc. I always wondered if the same was true of thermistors when I was having issues, but I ran the printer for years with the older style thermistor assemblies on the e3d and always had rock solid hotend temps. It was only when I started using the thermistor cartridges in latest generations that I started having trouble.
     
  4. brucehvn

    brucehvn Member

    Joined:
    Jul 14, 2014
    Messages:
    6
    Likes Received:
    0
    I printed a small case to hold the thermocouple board and mounted it to the back of the frame out of any draft from any cooling fans. I'm still getting the variations in my original post. I guess my main question still stands whether this is normal fluctuation for the thermocouple, or should it be more stable?
     
  5. orcinus

    orcinus Well-Known Member

    Joined:
    Oct 20, 2015
    Messages:
    336
    Likes Received:
    111
    I’d say that’s fairly normal.
    You’ve got to remember that the voltages from a thermocouple are very very tiny. They’re subject to noise in the ADC as well as electrical noise from the surroundings (and it gets worse the longer the leads).

    And, to top it off, the lower the temperature, the more noise you will get.
     
  6. Old_Tafr

    Old_Tafr Well-Known Member

    Joined:
    Mar 7, 2016
    Messages:
    711
    Likes Received:
    91
    If it were possible I would suggest a test rig rather than the printer as a way to check what is theoretically possible, some web searches may yield more info on the accuracy possible and orcinus's comments are a good start plus of course the Stack Exchange info.

    Things to look at are...

    What is theoretically possible i.e. assuming no electrical noise and the components like the ADC (Analogue to Digital Converter) are perfect.
    • From the TC spec you can find out what voltage you should get at your operating temp
    • From the spec of the ADC you can see what its resolution is.
    • From the design of the circuitry you may be able to find out what one bit represents as a temp change (this alone may be the answer)
    Working through the above should show what resolution you can achieve and mean that a test setup isn't necessary if you can't resolve less than a one or two degree change
    If checking the above shows you should be able to resolve say 0.1 oC then follow the simple guidelines given in the Stack Exchange post.

    Although external interference would have to be constant to affect the temp all the time (i.e. not spikes from say a room thermostat) have a look at what is close-by, radio, TV, PC, DECT phone, anything with a motor, anything with a wall-wart power supply (all have very noisy switching mode supplies) Microwave etc. power supplies for LED strip lighting, a million suggestions on the web.

    Assuming no wires shorted, checks as suggested in the SE post, then separating different wiring, TC from motor wires etc. Twisting wires (different pitch of twist may be difficult; this is what is done in Ethernet cables to cut down cross-talk) and screening wires which will probably mean replacing wires with screened twisted pair, (just watch the power rating if you do this). If you screen wires, then ground the screen at one end only or you get an "earth-loop" and megga noise (used to be in your stereo system where you would get massive mains hum) . What you ground the screen to is a difficult one, mains earth would be a start, signal ground of the ADC circuitry would be another.

    If you are not sure about messing around with mains wires etc. DON'T


    But first, check the theoretical maximum resolution of the measuring system as this may have more bearing than noise.

    PS. I have posted in the past about the way stepper motors are controlled, although they may not be moving they rely on being powered on all the time and they receive a square wave, this square wave equals megga wide spectrum noise, look at simple Fourier Analysis which shows a square wave is actually made up of a mass of higher and higher frequency signals = noise.
     

Share This Page