PI Regulation - some findings

Discussion in 'Discussions' started by k3rb3r05, Aug 15, 2019.

  1. k3rb3r05

    k3rb3r05 Member

    Joined:
    Aug 6, 2019
    Messages:
    32
    Likes Received:
    1
    Yes, I realise there is already a guide on this "feature" but, truth be told, it seemed greek to me.

    My current setup
    Single SS 316L spaced coil, 30 awg, 1.08 ohms
    TCR 118
    Temperature @ 210degC

    After some testing, my "sweet spot" is at 75W, P=380, I=50

    PI regulation is based on Celsius readings. Careful when changing regional settings from NToolBox. When going from Fahrenheit to Celsius, it will default to 315 Celsius. If you fire at that, you will get a nasty hit. Weirdly enough, if you adjust it from the mod, you will get proper C to F (and back) "translation"

    The formula to get a starting point (which you can adjust to your liking) is:
    P = (power * 1000) / the temperature you are trying to reach (in Celsius)

    So, using the numbers from above, (75W, 210C)
    P = (75 * 1000) / 210 => P = 75,000 / 210 => P = 357 = 360

    (Had to increase P from 360 to 380 because I was getting some spikes.)

    For I, the formula is
    I = (power * 1000) / (temperature you are trying to reach * average puff time)

    Note: Even if you have proper P configured for your usual temperature and for some reason it changes (like above), PI regulator will overshoot initial temperature. When it happened to me, it reached maybe 250 - 260degC. So, maybe the formula isn't entirely correct, but it has worked just fine when temperature was set properly. (Started sampling from 7W all the way to 75W).

    Formula(s) may not be applicable to exotic coils given AF defaults to P=1850 (I don't have any such coils to test against with)

    Any input is welcome

    Extra notes
    PI regulator isn't a "set it and forget it" type of setting. It is, if nothing changes in your configuration. But even rewicking, your coil can increase its resistance. If you are in the higher ohms group (like myself) , depending on your mod (its max V), increasing coil resistance can affect your power setting

    For example, earlier today I had to rewick. My mod read it .02 ohms higher and mod lowered my power by 1.3W. I had to readjust P, but it still was all over the place. Had to calibrate by raising my ohms by 0.04 (which lowered my power to 71.7W). I recalibrated P for 210degC at that power, but it still was all over the place. When I dropped my temperature to 200degC, I got 99% perfect vape
     
    #1 k3rb3r05, Aug 15, 2019
    Last edited: Aug 15, 2019
    Fladder72 likes this.
  2. k3rb3r05

    k3rb3r05 Member

    Joined:
    Aug 6, 2019
    Messages:
    32
    Likes Received:
    1
    While the formulas above "seem" to work, they aren't the correct ones. Heck, half of my 1st post (if not more) is probably gibberish

    For example, I got my PI controller set up with P=140 / I=10 and works better than yesterday's settings

    From what I have gathered so far, set temperature plays a role, but different that what I initially thought

    From https://en.wikipedia.org/wiki/PID_controller
    (Note: AF's controller is only a PI one , so disregard the "derivative" portion)

    In simple english,
    SP = my set temperature
    PV = coil temperature at end of loop
    If SP > PV, there is an error (e) , which PI controller tries to rectify next time it goes through the loop

    Is there an engineer in our mist that can give us some pointers and maybe an example, that is applicable to a vape device, on how to make PI work without too much trial and error?

    EDIT
    PID simplified
    1) http://www2.metso.com/MetsoTutor.aspx
    2) https://www.crossco.com/blog/basics-tuning-pid-loops
     
    #2 k3rb3r05, Aug 16, 2019
    Last edited: Aug 17, 2019
  3. k3rb3r05

    k3rb3r05 Member

    Joined:
    Aug 6, 2019
    Messages:
    32
    Likes Received:
    1
    PI should be set on a fresh or dry-burnt coil

    Gank built up diminishes coil's ability to heat up. If you set up PI on a coil that has gank on it and then you do a dry burn, you will see your previously set PI values are out of wack.