polymer alum cap leakage

Less leakage in the "-" direction. Did you have it wired correctly?

I'm going to pretend that you didn't ask that question.

Nice, thanks. So the obvious question is what happens at -30 V and beyond?

(I'd like a lab power supply that goes above 100V.) George H.

It's cooking at -10 now, for the weekend at least. I suppose I could crank it up after that test.

My HP6212A is older than my average engineer here. Beautiful little box. I got it on ebay.

Hey, John, are we data / measurement driven or not?

I didn't connect it backwards. (I wonder if it would stand 116 volts backwards.)

But note the third 10 volt leakage measurement: zero with 100 nA meter resolution. The 116 volts seems to have done some oxide forming that subsequently reduced leakage, which seems to have affected the following negative voltage measurements. Apparently polymer caps do have some amount of oxide forming, like the wet ones.

I'm running it long-term at -10 volts, and the leakage current seems to be slowly declining. I'm thinking these caps are usable at modest back bias.

I had one a couple years ago that got installed backwards in a test unit- it took a while to die- hundreds of hours. I think it was operating at about -13V.

--sp

Isn't the application issue one of frequent polarity reversal? Simple series RC at 60Hz requires no special supply to test.

RL

What was its rated voltage? I suppose I should be careful here.

No, it could park at +20 or at -5 for a long time.

I'll run this for a month or so and see what happens.

The total current, at -10 volts, started at 150 uA but is down below

"Anything worth doing is worth overdoing."

- Meat Loaf

leaky compared to a regular electrolytic. Could be cheaper (grumble).

--sp

e.jpg

If it cost less it might not be as good. JL, I'd test ten in parallel, beat on 'em a bit (electrically), bring out the heat gun and freeze spray.

If they all live... I'd give it a go.

Totally unrelated, but today I found an instrument we shipped to Cambridge (over the pond, shipping is expensive) had a star washer between the screw and a pretty thick plastic insulating barrier..(TO-3 pac.. probably from keystone.) It worked when it left, but shorted on the way over. (now replaced with flat washer and split lock... which I'm pretty sure is how it started..(it's been in productions hands for many years)) Thanks goodness they have smart grad students. A day of back and forth emails/ measurements and the 'mischief was managed'.

A new failure route for me.. though I've had plenty of run ins with burrs/ sharp metal and insulating pieces.

George H.

The original Oscon product from Sanyo indicated a 20% temporary or 10% continuous reverse voltage rating. I'm not sure what subsequent vendors claim.

Normal leakage currents have a two orders of magnitude response time, over hundred hour periods, for a stepped voltage (though this is speeded up at elevated temperatures). Rated leakage on first use is only specified after one hour at the application voltage.

While the normal failure mode in these parts is age/stress related to produce reduced C and higher ESR, a short circuit failure mode is predicted for forward or reverse overstress, surge charge or discharge current, or simple mechanical abuse.

Parts with 35 or 50 volts rating approach your requirement, without abuse. I'm not sure what your exercise is intended to prove, as qualification of even small numbers of parts outside of their specified limits is likely to be impractical.

RL

How about 6?

These are running at -10 volts; I'll need -5 in real life. I have maybe a month before I Gerber my board, so I can run them at least that long.

So far, the current is slowly declining. That pic was 46 uA for all six caps, and later in the day it was 30.

These are Panasonic 56uF 25V polymers. The ESR spec is 30m max, and they measure right around 30.

Yup. One bit of firefighting I did at IBM about a dozen years ago was an optical inspection system for that sort of damage.

Production of their then flagship server, internally called Regatta, was completely stopped because some bright spark installed flat washers burr-side down on hundreds of units. That's how you usually do it when assembling mechanical parts, of course, because it looks better.

However, on a 20-something-layer PCB, it caused short circuits between power and ground planes. (One part of the fix was a new rule: "relieve power planes under mounting pads." Well, duh.)

This gated some ridiculous number of millions of dollars in server sales, plus all the associated software and services revenue.

Cheers

Phil Hobbs

Update: six Panasonic 56uF 25V polymers in parallel, reverse biased at

-10 volts.

On Friday at noon, leakage current started at 150 uA total, and declined to 24 uA by 6PM. On Monday morning, total leakage is 2 uA.

The series LED is now invisible in room light.

What do they look like in the forward direction now?

I'd "saute" them a bit with a heat gun and cold spray. Don't Al elector's leak more at low temp? I don't know polymer. I should read about them... whose got a good app note?

George H.

kage.jpg

use

er

Screwing up and costing millions is not something I want to experience.

I had machined parts with a threaded hole and burrs piercing a sil pad.. electronics/ machining rule #27, deburr all holes. (and check) I sometime think it's amazing that electronics works as well as it does. (we should get more credit for our sucesses :^)*

George H.