Reforming capacitors (technical description, not politics)

On Fri, Jul 31, 2015 at 06:43:44PM +0100, Robert Jarratt wrote: The only automation was the data collection. For the reforming I set the lab supply to the current limit I think is safe and set the voltage to the final reform voltage and just let it go. If your reforming several capacitors at once if you put a resistor in series with each cap you can use to the measure the leakage. If you use a 100 ohm resistor you get 100 mV = 1 mA which makes it easy to convert the meter reading. Stuff I found wasn't consistent for reforming current. I though 5 mA was safe for a capacitor so used that for the entire bank. The most strict reference said to use the rated maximum leakage current for reforming. If one capacitor is worse it may take most of the current. The supply I had wasn't entirely stable regulating at that low of a current so you can see the steps where I readjusted it when it started to creep up. I used the maximum continuous voltage on the cap label as my reforming voltage. I just do the reforming run until the current stops dropping. You can see that in the graph. I then verify that the leakage on each capacitor is reasonable. For coke can capacitors I want them each to be under 1 mA. I had forgotten to put the series resistors but since my entire bank was .5 mA I didn't worry about not measuring each one though the variance would have been interesting. My capacitance meter didn't measure that high and my ESR meter didn't measure that low so I made the measurements using the other methods. The slides seem to have gotten a little confusing when I split some into backup. The capacitance was measure by setting the supply to a reasonably high current limit and timing how long it takes to charge. The automation was logging the data. I had a digital scope with USB to a laptop that I used to measure the voltage on the bank and the voltage across the series resistor. As others have said the capacitor and storage conditions are the driver. If your not in a hurry just let it go until its not changing much.