Re: Power supply capacitors in general


Don Black <donald_black@...>
 

What sort of caps are they (electrolytic. paper, plastic, etc.). If not electrolytics how did you measure leakage? For coupling capacitors a hundred Meg can be significant and may not show until working (high) voltage is present. Also caps sometimes behave as batteries and produce a voltage which won't discharge, you need to test this with a voltmeter.

Don Black.

On 25-Nov-13 3:34 AM, yachadm@... wrote:
 

 Who was the member who has the Meter which tests dissipation? Many posts a few months ago.


I just took a bunch of caps off a high-end amplifier, and the caps tested OK (uF + ESR), but after replacement, the sound is much better.


So, I'd like to send him these, and when he has some "fun" time, measure the D, and see whether that is a factor as well, in caps which otherwise test OK.



---In TekScopes@..., wrote:

Some hands on study of piezoelectric effects in capacitors:

www.kyocera.co.jp/prdct/electro/pdf/technical/highcv.pdf‎

What bugs me is this statement:

"Tantalum capacitors have excellent reliability far
superior to aluminium electrolytics because the tantalum
dielectric has no known wear-out mechanism."

I've seen it in some other papers too (including manufacturers app notes), together with a "half bathtub" diagram suggesting that once they're past the initial infant mortality, they would work indefinitely with low rate of failure.

Yet we have older papers like this one:

downloads.hindawi.com/journals/apec/1976/349838.pdf‎

which suggests that there is a mechanism (crystal growth) that will bring an end of life condition and we may have a full bathtub lifetime diagram.

Which is pretty much supported by the experience of 4xx series owners noticing that after so many years of good service, the tantalums start failing in droves at about the same time.


--- In TekScopes@..., "vdonisa" wrote:
>
> Well it depends on the potential damage I could do... if little damage than I would try in place and measure, if I risk burning expensive or unobtainium parts, then I'll want to experiment off-circuit first...
>
> Basically since this is a "find a suitable replacement preferably more reliable" exercise, I guess the steps would be:
>
> - understand the role of the component and what characteristics it needs to have in order to correctly fulfill this role
>
> - read datasheets and write down the suitable candidates
>
> - pick the favorites, check availability, and order the first 2-3 in the list
>
> - if no previous experience with the particular part, do some experiments with it off-circuit as to avoid nasty surprises (i.e. it has some peculiarities that were missed when reading datasheet)
>
> - replace the original component, cross your fingers, power on, duck, if no pyrotechnic effects then measure to see if everything works within expected parameters
>
> I definitely wouldn't agonize over an aluminum power supply smoother. I would just check rated voltage, ripple current and dissipation factor / ESR and fire the soldering iron. My problems were with the tantalum .22u (HVPS) and 3.3u (A1). First and biggest problem was that I wasn't understanding what their role is.
>
> With David's help it turned out that .22u is in a RC circuit filtering role, I should be doing fine with C0G ceramic instead of tantalum, however it was also suggested that transient response could be interesting to check. I was thinking of somehow replicating the schematic fragment on the breadboard and watching off-circuit.
>
> The 3.3u - it's a shot in the dark, both David and me believe they're some form of smoothers/filters, but since they go directly into hybrid circuits of which no schematic is available, we may be wrong. So the safest bet would be "replace with similar type to what Tek has used". Or I could get brave. lol.
>
> Cheers,
> Valentin
>
>
>
>
>
> --- In TekScopes@..., Peter Gottlieb wrote:
> >
>
> > So, do you try different caps in place and measure how they perform, or do you
> > rig up a swept reactance arrangement and try and do it by calculation?
>




This email is free from viruses and malware because avast! Antivirus protection is active.


Join TekScopes@groups.io to automatically receive all group messages.