Hot capacitor in switcher

Discussion in 'Power Electronics' started by inwo, Jul 8, 2016.

  1. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    Looking for an easy fix for capacitor that runs hot on a cheap 5v 5a switching power supply.

    C1 gets over 105C at full load. C2 and C3 are not used. Adding them is not helpful.
    C1 and C2 = 1000/16v.

    I pulled a high quality 3300/16v from a computer PSU to swap. Runs < 50C.

    Is it the size 3300 vs 1000, or quality that makes the difference?

    Should I replace with caps rated for 100khz.? Same value? 1000/16 or 2 X 470/16? psufilter.gif
    No way of checking what came from the factory, except they overheat.
    I have 30 pcs. Need a fix.
     
  2. #12

    Expert

    Nov 30, 2010
    16,250
    6,746
    I believe you just proved that a high quality, low ESR, cap is the answer to the heating problem.
    3300 uf instead of 1000 uf would tend to increase the current through that capacitor, and that would cause MORE heat. Therefore, it must be the low ESR which is saving this cap from overheating.
     
  3. Veracohr

    Well-Known Member

    Jan 3, 2011
    549
    75
    I think higher capacitances generally have lower ESR in the same series.
     
  4. RichardO

    Well-Known Member

    May 4, 2013
    1,229
    382
    True but the observation may be misleading. If all the caps in a series are measured at the same frequency then the hihger value caps will have lower ac impedance -- which is good. But... the higher value caps could have higher series resistance -- which is bad.
     
  5. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    Thanks, I tend to agree. Before ordering caps I would like to understand the problem.
    These are new supplies. Cheap though they may be.
    Should have known better, the sample blew up.
    These are from another supplier.
    I do have an esr meter.
    Do I need to check at high frequency? I guess that's what I'm doing in circuit.
    I'll compare the good to the bad and see what that tells us.
    This is new to me. I've never seen a capacitor overheat. At least not a new one.
     
  6. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    The 1000/16 =.05 esr
    The 3300/16 = .00 esr

    Are these reasonable?
    My calibration is correct.
    Scope shows esr testing at ~500hz.
    Are my esr results helpful?
     
  7. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    DK esr.gif

    Good choice?
    Do high freq rated caps go by impedance rather than esr?
     
  8. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    No edit button...........
    After using meter a while it stabilized.
    1000 = .07
    3300 = .02

    One of my 1000/16 = .14
     
  9. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    First, your ESR meter barely has the resolution to be of use to you.
    Second, you are getting a good hands on education into the effects of ESR.

    The capacitor spec sheet in post #7 looks like a pretty good (low ESR) cap.

    If you post more information, maybe we can post some better answers.
     
  10. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    Capacitor impedance is a parameter that represents the combined resistive and inductive elements. Capacitor ESR represents only the series resistance of the capacitor. Some capacitor manufacturers use "Tan Delta" (Tangent Delta) to represent capacitor goodness. In other words, they are not making it easy on you to select capacitors with these properties.

    Stick with ESR, that is the "goodness" parameter you are looking for.
     
    #12 likes this.
  11. inwo

    Thread Starter Well-Known Member

    Nov 7, 2013
    2,433
    315
    The capacitors rated at 100khz do not list esr.
    Others are rated at 120hz.
    As esr = R - that is where heat is generated?... I*I*R

    General question.
    How will value affect temperature?
    I can tolerate ripple in my application. And this is first stage after rectifier. Capacitor input.
    So I can use lower value. This makes more heat? (more ripple) I can't notice on scope.

    Or is larger value preferred? Less ripple.
     
Loading...