Re: heating a cap

On Sat, 05 Oct 2024 11:38:35 -0700, john larkin <JL@gct.com> wrote:

On Sat, 05 Oct 2024 19:18:15 +0100, Cursitor Doom <cd@notformail.com>
wrote:
>

On Sat, 05 Oct 2024 08:03:14 -0700, john larkin <JL@gct.com> wrote:
>

On Sat, 05 Oct 2024 06:45:40 GMT, Jan Panteltje <alien@comet.invalid>
wrote:
>

On a sunny day (Fri, 04 Oct 2024 22:15:50 -0700) it happened john larkin
<JL@gct.com> wrote in <8qi1gj5d27uqdkudv6vfql0fcv273mjcve@4ax.com>:
>

On Fri, 04 Oct 2024 23:50:36 +0100, Cursitor Doom <cd@notformail.com>
wrote:
>

On Fri, 04 Oct 2024 10:31:13 -0700, john larkin <jl@glen--canyon.com>
wrote:
>

On Fri, 04 Oct 2024 12:10:38 -0400, legg <legg@nospam.magma.ca> wrote:
>

On Thu, 03 Oct 2024 16:36:24 -0700, john larkin <jl@glen--canyon.com>
wrote:
>

I got a small (under 1" long) aluminum electro cap, 220 uF 63v, and
cranked up the voltage. It started drawing a bunch of current at 105
volts, got hot, and folded back to 80 mA at 87v.
>
It got too hot to touch in a couple of minutes, after roughly 500
joules. Freeze spray let it go back up to 100 volts or so.
>
None of that seemed to damage it, so an electrolytic cap sort of has a
built-in MOV.

>
You're not 'heating a cap'.

>
Felt hot to me.
>

>
You're applying voltage overstress to failure, using a
current limited source.

>
But it didn't fail.
>

>
This tells you precisely nothing.

>
Told me a lot. Why elect to not learn things?
>

>
Were you earing safety glasses?

>
No earrings, and my normal glasses.
>

>
Are you sure you want to advertize this increasingly
erratic behavior?

>
Experimenting with parts is admittedly a bizarre thing for an engineer
to do. Sorry.

>
Legg seems to have a problem on the groups with anyone who isn't an
out-and-out Commie. Just ignore him.

>
There's no reason to not destroy parts. They don't have feelings.

>
How do you know, have you been once?
>
BTW I did test some parts out of spec..
But many problems come from aging with electrolytic caps.
My old Samsung TV lasted 20 years... on many hours a day.
So good electrolytics do exist.
Just designing a bit below maximum specs may help.

>
I know people who are terrified of running parts anywhere near abs
max. I run some parts at 2x abs max voltage or power. Pushing parts is
the way to get performance, especially speed. Some work fine at 4x.
>
Test them, blow up a few, define your own abs max. But only when there
is a big payoff.
>
I'm pushing a lot of HMC659's (15 GHz distributed amplifiers) to about
2x voltage. It's intimidating to test them to destruction because they
cost over $300 each.

>
Surely you are shortening the service life of those components by
doing this. Just because they survive 24 hours or whatever at 4x
voltage doesn't tell you everything. They might go 'pop' after a week
or they might last a year. Either way, it's not good for repeat
business?

>
We have over a century of unit time so far and no problems.
>
RF parts have terrible time-domain specs. The datasheets assume that
the supply voltage is coupled into the drain through an inductor or a
tank. So at max swing the actual drain voltage goes to 2xVcc. When the
RF data sheets say "abs max" they mean the supply voltage.
>
And they assume (without saying so) that the signal is RF or telecom,
namely AC coupled and DC balanced. 8b10b or some such.
>
So for pulse work, one throws away all that silly S-parameter and dBm
nonsense. I have a possibly new way to bias the HMC parts for
electro-optical use; park high, pulse low.
>
I wonder what sorts of philosophies various companies have when
writing data sheet abs max specs.

My impression is that use cases are the main driver.

Joe