XPost: sci.electronics.design   
   From: bill.sloman@ieee.org   
      
   On 4/03/2024 5:03 am, Cursitor Doom wrote:   
   > Hi all,   
   >   
   > Using my Peak ESR/Capacitance meter, I was carrying out in-circuit   
   > checks on large electros in the linear PSU I've previously mentioned   
   > here. I was getting a lot of "in-circuit/leaky" warnings for two of   
   > them, so I pulled them out completely and am getting the same warning   
   > when they're checked out of circuit, which surprised me as it's   
   > unusual IME. Anyway, the leak would have to be very bad to result in   
   > ripple, would it not? AFAIK, the leading culprit for ripple is caps   
   > which have lost a significant amount of capacitance or else developed   
   > a very large ESR.  Do I have that right?   
      
   What you have wrong is the imagined association between leakage and ripple.   
      
   The "ripple" on the voltage across a capacitor reflects the charge that   
   is being taken out of and fed into the capacitor over the mains cycle.   
   More capacitance means smaller ripple.   
      
   Leakage is just the current flowing through the oxide layer on top of   
   metal conductor surfaces inside the electrolytic capacitor. If the   
   capacitor has been un-used for a long time, some of the oxide layer may   
   have diffused away, making the capacitance and the leakage current   
   higher. Applying the working voltage will re-form the oxide layer by   
   making it a bit thicker.   
      
   > I'd like a "second opinion" as it were on the leakiness of these caps.   
   > What's the best old-school method for testing for this? I just want to   
   > ensure the ESR meter isn't faulty (highly unlikely but the possibility   
   > must be eliminated to be sure).   
      
   If an electrolytic capacitor hasn't been used for some time, it's going   
   to be  leaky. If you want to measure how leaky, measure the direct   
   current flowing through the capacitor as you increase the bias voltage   
   across it. It should drop as the oxide layer re-forms, perhaps over hours.   
      
   --   
   Bill Sloman, Sydney   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)