electrolytic mystery

W

Walter Harley

Guest
Perhaps y'all can help me with a mystery. All electrolytic capacitors have
a "working voltage" rating. The question is, what are the consequences of
exceeding that voltage? Now, before you jump to the obvious answers, read
on:

Audio devices commonly have opamp-based output stages, followed by a DC
blocking cap. The rating of the cap is more often than not either 16V or
25V, probably because of the typical +/-15V supply rails; the polarity
orientation of the cap varies, or rarely it may be bipolar. A typical
example would be the circuit at http://www.rane.com/pdf/old/pe15sch.pdf
(chosen for no other reason than that I used one last night).

Now, balanced outputs from audio devices often get hooked up to mixer inputs
that are expecting to see microphones. And those inputs often are supplied
with phantom power, which is equivalent to a 48V DC source in series with
6.8k. (An example, again from the excellent and helpful Rane web site, is
http://www.rane.com/pdf/ms1bsch.pdf.)

So, if one plugs a typical balanced output into a typical phantom-supplied
mic input, as is very frequently done, we get this circuit, where I have
been optimistic about the polarity of the capacitor:

6.8k
___
.--|___|--.
| | +
--- ### 15uF
- 48v --- 25V
| |
'---------'


QUESTION: WHY DOESN'T THIS BLOW UP??

Just for grins, I've got this precise circuit sitting there on my test bench
right now. I measured the cap after about 2.5 hours and its ESR and
capacitance were unchanged. I'll leave it running for a few days and
measure again. But I already know that a gazillion musicians and audio
engineers get away with this on a daily basis.

So, why does this work, and for how long should it be expected to work,and
how should the eventual failure manifest? What does the working voltage
rating really mean?
 
On 22 Aug 2004 22:50:02 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> wrote:

|Perhaps y'all can help me with a mystery. All electrolytic capacitors have
|a "working voltage" rating. The question is, what are the consequences of
|exceeding that voltage? Now, before you jump to the obvious answers, read
|on:
|
|Audio devices commonly have opamp-based output stages, followed by a DC
|blocking cap. The rating of the cap is more often than not either 16V or
|25V, probably because of the typical +/-15V supply rails; the polarity
|orientation of the cap varies, or rarely it may be bipolar. A typical
|example would be the circuit at http://www.rane.com/pdf/old/pe15sch.pdf
|(chosen for no other reason than that I used one last night).
|
|Now, balanced outputs from audio devices often get hooked up to mixer inputs
|that are expecting to see microphones. And those inputs often are supplied
|with phantom power, which is equivalent to a 48V DC source in series with
|6.8k. (An example, again from the excellent and helpful Rane web site, is
|http://www.rane.com/pdf/ms1bsch.pdf.)
|
|So, if one plugs a typical balanced output into a typical phantom-supplied
|mic input, as is very frequently done, we get this circuit, where I have
|been optimistic about the polarity of the capacitor:
|
| 6.8k
| ___
| .--|___|--.
| | | +
| --- ### 15uF
| - 48v --- 25V
| | |
| '---------'
|
|
|QUESTION: WHY DOESN'T THIS BLOW UP??
|
|Just for grins, I've got this precise circuit sitting there on my test bench
|right now. I measured the cap after about 2.5 hours and its ESR and
|capacitance were unchanged. I'll leave it running for a few days and
|measure again. But I already know that a gazillion musicians and audio
|engineers get away with this on a daily basis.
|
|So, why does this work, and for how long should it be expected to work,and
|how should the eventual failure manifest? What does the working voltage
|rating really mean?
|

Here's my guess;

It is probably easier to analyse if you look at asimpler mic amp
circuit such as
http://www.siliconchip.com.au/cms/A_102159/article.html
In this design the blocking caps are 10uF 16V NP.

Imagine that the balanced output is plugged into a mixer input
sourcing 48V phantom power.

The 48V phantom supply in the mixer desk will have a 6k8 resistor to
each of the XLR input pins (pins 2 & 3) thus its output impedance is
effectively 3k4 ohms. When fed down a cable to the mic amp output this
voltage is recombined via the two 1K resistors then via a blocking
diode and the 12V zener regulator to ground. We now effectively have a
series connection of 3k4 ohms + 500 ohms + 12V zener with 48V behind
it. Thus total current from the 48V supply into the mic amp output
will be (48 - 12)/3900 = approx 9mA. Thus 4.5mA will flow in each of
the 1K resistors meaning a volt drop of 4.5V across each of them. If
we add Zd = 12V + 4.5V we get 16.5V wrt grnd appearing at the
capacitor terminals connected to pins 2 and 3 of the output connector.
Since the op-amp output is referenced to Zd/2 = 6V the maximum DC
voltage which can appear across the output caps is around 11V which is
safely within their 16V rating.
 
On 23 Aug 2004 16:42:04 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> wrote:

|"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
|news:jlhji0l7mjq8efpk17fdo29tuivmp1t9e3@4ax.com...
|> Here's my guess;
|>
|> It is probably easier to analyse if you look at asimpler mic amp
|> circuit such as
|> http://www.siliconchip.com.au/cms/A_102159/article.html
|> In this design the blocking caps are 10uF 16V NP.
|>
|> Imagine that the balanced output is plugged into a mixer input
|> sourcing 48V phantom power.
|>
|> The 48V phantom supply in the mixer desk will have a 6k8 resistor to
|> each of the XLR input pins (pins 2 & 3) thus its output impedance is
|> effectively 3k4 ohms. When fed down a cable to the mic amp output this
|> voltage is recombined via the two 1K resistors then via a blocking
|> diode and the 12V zener regulator to ground. [...]
|
|
|Thanks, but I think that analysis does not apply to the situation I'm
|describing. The reason is that the circuit you chose appears to in fact be
|phantom powered, which means that at DC it has a steady current draw. A
|non-phantom-powered output would just have DC blocking capacitors on the
|output lines, which means that no DC current would flow (after startup)
|except for leakage through the caps; thus the caps see all 48V.
|
|If the circuit you showed weren't phantom-powerable, then the 1k resistors
|would not exist and there would be nothing to draw down the voltage.
|

The SC mic amp is designed to run from either a plugpack, batteries OR
phantom power, so it would be similar to your situation. I'm still
willing to bet that if you plugged the SC mic amp output into your
mixer input (or any other commercial unit supplying phantom power) the
caps wouldn't blow up even if it was powered by the battery or
plugpack. The 48V from the mixer input phantom supply would not cause
any problems in this situation. Thus the SC mic amp would conform to
the requirements for all other units which were also capable of
similar connection methods.
 
"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
news:j5hli0l2cmq1av5ctsg727qa1nu7g8if7n@4ax.com...
The SC mic amp is designed to run from either a plugpack, batteries OR
phantom power, so it would be similar to your situation. I'm still
willing to bet that if you plugged the SC mic amp output into your
mixer input (or any other commercial unit supplying phantom power) the
caps wouldn't blow up even if it was powered by the battery or
plugpack. The 48V from the mixer input phantom supply would not cause
any problems in this situation. Thus the SC mic amp would conform to
the requirements for all other units which were also capable of
similar connection methods.
Click to expand...

You would win that bet, if I were foolish enough to take it. But would you
make the same bet if you removed the 1k resistors that allow the unit to be
phantom powered?

The point here is that, regardless of how the unit is being powered at any
particular time, the 1k resistors pull down the voltage. But in the case of
a unit that is not *capable* of being phantom powered, those resistors do
not exist, and the output capacitors see the full 48V, despite being rated
for less.
 
On 24 Aug 2004 06:31:55 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> wrote:

|"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
|news:j5hli0l2cmq1av5ctsg727qa1nu7g8if7n@4ax.com...
|> The SC mic amp is designed to run from either a plugpack, batteries OR
|> phantom power, so it would be similar to your situation. I'm still
|> willing to bet that if you plugged the SC mic amp output into your
|> mixer input (or any other commercial unit supplying phantom power) the
|> caps wouldn't blow up even if it was powered by the battery or
|> plugpack. The 48V from the mixer input phantom supply would not cause
|> any problems in this situation. Thus the SC mic amp would conform to
|> the requirements for all other units which were also capable of
|> similar connection methods.
|
|
|You would win that bet, if I were foolish enough to take it. But would you
|make the same bet if you removed the 1k resistors that allow the unit to be
|phantom powered?
|
|The point here is that, regardless of how the unit is being powered at any
|particular time, the 1k resistors pull down the voltage. But in the case of
|a unit that is not *capable* of being phantom powered, those resistors do
|not exist, and the output capacitors see the full 48V, despite being rated
|for less.
|

Have you checked that there is a physical connection between the
analog ground (AGND) and CHASSIS GND? If there isn't then the 48V from
the mixer input will be floating at the output socket of the mic amp
and the caps won't have that dc voltage impressed across them. You say
you have the circuit on your bench and operating but is that in its
built up form in the chassis or just the boards connected up on the
bench.

Whatever is happening it can only be that either there is voltage
division (as in the SC circuit) or the 48V from the mixer input is
floating at the mic amp output. As far as I can see from the circuit
of the Rane that would only happen if the AGND and CHASSIS GND were
isolated (perhaps using a capacitor).

What voltage do you measure;
a) across the blocking caps
b) from either side of the caps to both AGND and CHASSIS GND
 
Not much mystery here.

Caps operated at or below their voltage ratings are guaranteed not to fail.

Caps operated somewhat beyond their voltage ratings MAY fail.

Caps operated far beyond their voltage ratings WILL fail.

See manufacturer for definitions of "somewhat" and "far", or just don't do
it because it's a BAD idea.

--

KC6ETE Dave's Engineering Page, www.dvanhorn.org
Microcontroller Consultant, specializing in Atmel AVR
 
"Dave VanHorn" <dvanhorn@cedar.net> wrote in message
news:TMydneyiLOxmPLbcRVn-tA@comcast.com...
Not much mystery here.

Caps operated at or below their voltage ratings are guaranteed not to
fail.

Caps operated somewhat beyond their voltage ratings MAY fail.

Caps operated far beyond their voltage ratings WILL fail.

See manufacturer for definitions of "somewhat" and "far", or just don't do
it because it's a BAD idea.
Click to expand...

You may need to read the original posting to understand why there's a
mystery involved. Your compellingly simple argument seems unfortunately to
not match well with the observed facts.
 
"Roger Gt" <not@here.net> wrote in message
news:iHPWc.11229$XM.10000@newssvr27.news.prodigy.com...
Do you mean empirical data from a single test invalidates several
centuries of use and experience of hundreds of thousands of people
using capacitors?

Gee, maybe there is a tooth fairy after all! There was a coin
under my pillow and the tooth was gone!

No mystery, just an anomalous test!
Click to expand...

No, I certainly do not mean that! Was there something confusing about my
original post? I thought it was quite clear.

The mystery is *not* that my single test didn't show damage. The mystery is
that many, many pieces of audio equipment out there were designed and
manufactured in such a way that they endure the exact same circumstances my
test replicates, and by and large they are still working just fine.

I can point you to many schematics of audio equipment with "DI" (balanced
mic-level) outputs, containing 16V or 25V DC blocking capacitors. It is
commonplace throughout the audio profession to plug these outputs into mixer
inputs that provide 48V phantom power. It is undesirable and wrong, and
some people know better, but it is commonplace nonetheless. When this
happens, in many cases, the caps are exposed to nearly twice their rated
voltage, for hours, days, or months on end. And yet, the repair shops are
not full of blown DI output capacitors, and the manufacturers continue to
make their devices with 25V rated, rather than 50V rated, capacitors.

Obviously, capacitors do fail. Overheated and dried-out caps, for instance,
seem to be a primary mode of failure in TV's, computer monitors, and
switch-mode power supplies. But in audio DI outputs, they seem to be
failing much less often than I would expect.

*That* is the mystery.

Thus far the best answer seems to be that the current limiting inherent in
phantom power means the capacitors are merely being re-formed to the higher
voltage. However, other postings, and some comments in the tech notes Legg
pointed to, suggest there may be subtler failure mechanisms happening; and I
wonder what the parameters of that process are, e.g., can any electrolytic
be re-formed to any arbitrarily higher voltage, and at what (if any) cost to
its eventual lifetime? If not, what's the breaking point? How much
capacitance will typically be lost, when re-forming to twice the rated
voltage? I also wonder, e.g., how much current limiting is necessary. I've
not seen any manufacturer information on that topic (e.g., max pulse power),
at least for small electrolytics.
 
The source is current limited, and I think you'll find that the majority of
the caps, in this situation either take it, or become slightly leaky.

It dosen't "blow up" because the current is limited by the 6.8k
The 48V is only twice the elevtrolytic's working rating.

Bench it, it's only a few components.
I've seen a lot of things in music circuits, that are pretty questionable
technically, but I guess they must sound good, and/or be blessed by the high
preists of audio.

--
KC6ETE Dave's Engineering Page, www.dvanhorn.org
Microcontroller Consultant, specializing in Atmel AVR
 
I can point you to many schematics of audio equipment with "DI" (balanced
mic-level) outputs, containing 16V or 25V DC blocking capacitors. It is
commonplace throughout the audio profession to plug these outputs into mixer
inputs that provide 48V phantom power. It is undesirable and wrong, and
some people know better, but it is commonplace nonetheless. When this
happens, in many cases, the caps are exposed to nearly twice their rated
voltage, for hours, days, or months on end. And yet, the repair shops are
not full of blown DI output capacitors, and the manufacturers continue to
make their devices with 25V rated, rather than 50V rated, capacitors.
Click to expand...
My guess is mostly luck.

One thing that might be interesting to look at... Typical Al electrolytics
come in standard size packages. For a given package size and given
number of microfarads, there is some max voltage that you can get.
If you ask for a lower voltage, you might get a smaller package,
or it is easier to manufacturer.

Maybe the caps that actually get used are low value for the package
but the manufacturers are actually only making higher voltage ones
and marking (and pricing) them as though they were a lower voltage.

--
The suespammers.org mail server is located in California. So are all my
other mailboxes. Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's. I hate spam.
 
"Dave VanHorn" <dvanhorn@cedar.net> wrote in message
news:Kf-dnWdvM6IgarbcRVn-rA@comcast.com...
The source is current limited, and I think you'll find that the majority
of the caps, in this situation either take it, or become slightly leaky.

It dosen't "blow up" because the current is limited by the 6.8k
The 48V is only twice the elevtrolytic's working rating.

Bench it, it's only a few components.
I've seen a lot of things in music circuits, that are pretty questionable
technically, but I guess they must sound good, and/or be blessed by the
high preists of audio.
Click to expand...

As I mentioned in my initial post, I have benched it. Leakage was a
fraction of a microamp.

My present idea on this is that the voltage rating must be taken in
conjunction with the temperature rating: since both temp and voltage
increase leakage, a cap rated 85C and 25V is good for more than 25V if you
keep the temp well under 85C. And probably vice versa, too.

If I get a chance, I'll try the experiment in a temp-controlled oven, and
see if the cap that fared well at 48V 25C fares as well at 48V and, say,
65C. Need to build a temp-controlled oven, first - I don't think my wife
would be happy with me trying the experiment in our kitchen oven.
 
On 24 Aug 2004 16:59:09 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> wrote:

|"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
|news:g1sli01sgevq6mg4pjst2ftdo19lllj45k@4ax.com...
|> Have you checked that there is a physical connection between the
|> analog ground (AGND) and CHASSIS GND? If there isn't then the 48V from
|> the mixer input will be floating at the output socket of the mic amp
|> and the caps won't have that dc voltage impressed across them. You say
|> you have the circuit on your bench and operating but is that in its
|> built up form in the chassis or just the boards connected up on the
|> bench.
|>
|> Whatever is happening it can only be that either there is voltage
|> division (as in the SC circuit) or the 48V from the mixer input is
|> floating at the mic amp output. As far as I can see from the circuit
|> of the Rane that would only happen if the AGND and CHASSIS GND were
|> isolated (perhaps using a capacitor).
|>
|> What voltage do you measure;
|> a) across the blocking caps
|> b) from either side of the caps to both AGND and CHASSIS GND
|
|
|These questions were answered in my original posting.

Sorry Walter, those references to the Rane circuits actually confused
the issue...
|
|The circuit is a test circuit: a 48V bench supply, in series with a 6.8k
|resistor and a 15uF 25V capacitor. The voltage across the capacitor
|measures 48V. It's been running for about two days now; capacitor seems
|fine.

Now I understand... You haven't actually got the Rane circuit on your
bench but simply the series connection of 48V supply - 6k8 resistor -
15uF cap. It is regrettable that the ascii sketch does not show what
the test circuit looks like on my news reader (Agent). It is just a
jumble of characters.

here's what we have;



Rgds,

Ross Hpos (48V) ----- 6k8 -------15u/25V------- neg (48V)

Lets see what Rubycon has to say;

Q1: What consequences are expected when the voltage exceeding the
rated voltage is applied on an aluminum electrolytic capacitor?

A1: On the anode foil of aluminum electrolytic capacitor, an oxide
film capable of withstanding the rated voltage even if it is
continuously applied at the maximum operating temperature.

In the case when voltage higher than the withstand voltage of this
oxide film (overvoltage) is applied, the anode foil of the aluminum
electrolytic capacitor will form the oxide film equivalent to the
applied voltage. Owing to the reaction, gases will be generated, thus
leading to the pressure buildup in the capacitor. As the
characteristics of capacitor, decrease in electrostatic capacity and
increase in tangent of loss angle will be caused. The higher the
applied voltage is and the higher the ambient temperature is, the more
the gases are generated and the higher the internal pressure. This may
sometimes lead to the phenomena such as swelling of sealing material
(rubber packing) and further to activation of safety device (slipping
out of rubber packing in the products with no safety device).
Therefore, avoid the use of capacitor in the circuit where the voltage
exceeding the rated voltage may be applied to it.

The structural breakdown modes in case when overvoltage is applied are
as follows:


(1) Open
The safety device is activated (or rubber packing slips out), and
liquid electrolyte in the capacitor is flown out, thus leading to
dryup and finally to open condition.
(2) Short-circuiting
If the voltage higher than the withstand voltage of anode foil, that
of liquid electrolyte and that of separator paper is applied and it is
no longer possible to keep insulation, dielectric breakdown will be
caused, thus leading to short-circuiting.

Note the words "even if it is continuously applied at the maximum
operating temperature" in the first para of the answer.

Because the 6k8 resistor limits the maximum possible current to 7mA
there is no way that the capacitor can reach its maximum operating
temperature or even get warm. Therefore no gases are produced to
cause pressure buildup and without pressure buildup the cap will not
self destruct. If there were no current limiting then the excessive
voltage would be able to produce gases which would cause the
characteristic " bang" we associate with applying overvoltage to an
electrolytic cap.

This means that Watson A. Name was precisely correct in his initial
response.
|
Rgds,

Ross H
 
"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
news:pdroi0h37pjt9a677l1hh92htvtpmkbk5h@4ax.com...
Thanks, Ross. I just went up and read all the rest of their tech notes and
FAQ. That's a good reference - much more helpful than what I had found on
my own.

One comment; you'd said:
Because the 6k8 resistor limits the maximum possible current to 7mA
there is no way that the capacitor can reach its maximum operating
temperature or even get warm. Therefore no gases are produced to
cause pressure buildup and without pressure buildup the cap will not
self destruct.
Click to expand...
But the Rubycon tech notes suggest that the gas formation is an
electrochemical process, a consequence of the formation of oxide layer from
electrolyte and foil, and will happen at any temperature, albeit with more
gas at higher temps. My read was that the main thing that current limiting
does is that it limits the rate of that chemical reaction. Presumably below
a certain amount of production, the gases either are reabsorbed by the
electrolyte, or dissipate out of the capacitor at a safe speed.

It sounds to me like the thing to watch out for would be DI circuits in
power amplifiers, which can get quite warm. Otherwise, since an increase in
ESR wouldn't matter in this application, and a slight decrease in
capacitance also wouldn't matter, the consequences of re-forming the cap
would probably not be noticed by most users.

I'm interested to try and find out how much capacitance is lost. The one
capacitor I tried showed no change at all; in fact, a slight increase. I
assume that was an aberration.
 
On 25 Aug 2004 17:36:08 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> wrote:

|"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
|news:pdroi0h37pjt9a677l1hh92htvtpmkbk5h@4ax.com...
|> Lets see what Rubycon has to say [...]
|
|Thanks, Ross. I just went up and read all the rest of their tech notes and
|FAQ. That's a good reference - much more helpful than what I had found on
|my own.
|
|One comment; you'd said:
|> Because the 6k8 resistor limits the maximum possible current to 7mA
|> there is no way that the capacitor can reach its maximum operating
|> temperature or even get warm. Therefore no gases are produced to
|> cause pressure buildup and without pressure buildup the cap will not
|> self destruct.
|
|But the Rubycon tech notes suggest that the gas formation is an
|electrochemical process, a consequence of the formation of oxide layer from
|electrolyte and foil, and will happen at any temperature, albeit with more
|gas at higher temps. My read was that the main thing that current limiting
|does is that it limits the rate of that chemical reaction. Presumably below
|a certain amount of production, the gases either are reabsorbed by the
|electrolyte, or dissipate out of the capacitor at a safe speed.
|
|It sounds to me like the thing to watch out for would be DI circuits in
|power amplifiers, which can get quite warm. Otherwise, since an increase in
|ESR wouldn't matter in this application, and a slight decrease in
|capacitance also wouldn't matter, the consequences of re-forming the cap
|would probably not be noticed by most users.
|
|I'm interested to try and find out how much capacitance is lost. The one
|capacitor I tried showed no change at all; in fact, a slight increase. I
|assume that was an aberration.
|

I must admit that the theory behind electrolytic capacitors is still a
bit of a mystery to me. However, the explanation given here
http://www.faradnet.com/deeley/chapt_02.htm seems the most plausible
to me.

In the para on Leakage Current it is shown that the less conductive
the electrolyte is, the lower the leakage current and that the oxide
layer thickness is automatically matched to the potential difference.
It seems that as V increases, the thickness of the oxide layer also
increases, hence the ability to withstand a greater field strength, at
least that's how I read it. Perhaps this point is somewhat cryptic in
the Rubycon info.

Faradnet says:

Effect of Temperature on Breakdown Voltage

As the breakdown voltage or potential is a function of the anode film
thickness and conductivity of the electrolyte; and the conductivity of
the electrolyte varies with the temperature, it must hold true that
the voltage breakdown is also a function of temperature. Increases in
temperature cause increases in electrolyte ionization with resultant
increase in electronic emission from the electrolyte. This lowers the
potential required to rupture or puncture the dielectric or oxide
film. Thus, an increase in temperature results in a lowering of the
breakdown voltage and a decrease in temperature causes an increase in
the voltage breakdown of any specific dry electrolytic capacitor
structure. No graphic illustration of this effect of temperature
change is shown as there are other factors which also concern the
breakdown voltage. Reference is particularly made to the type of
separator material employed. These various other factors will be
mentioned again in later paragraphs.

When a static DC potential is impressed across the capacitor the
leakage current will be minimal and as long as the potential is below
the "breakdown" voltage, the capacitor should be able to withstand a
quite high voltage probably double its normal rated voltage. Where the
potential is varying the current through the cap will also vary and
the temperature will increase and thus lower the breakdown voltage.

Anyway, I leave the full reading of the information on Faradnet to you
and hope you can gain some further understanding of what is happening.

Ross H
 
On 25 Aug 2004 07:43:38 GMT, "Walter Harley"
<walterh@cafewalterNOSPAM.com> put finger to keyboard and composed:

"Dave VanHorn" <dvanhorn@cedar.net> wrote in message
news:Kf-dnWdvM6IgarbcRVn-rA@comcast.com...

The source is current limited, and I think you'll find that the majority
of the caps, in this situation either take it, or become slightly leaky.

It dosen't "blow up" because the current is limited by the 6.8k
The 48V is only twice the elevtrolytic's working rating.

Bench it, it's only a few components.
I've seen a lot of things in music circuits, that are pretty questionable
technically, but I guess they must sound good, and/or be blessed by the
high preists of audio.


As I mentioned in my initial post, I have benched it. Leakage was a
fraction of a microamp.

My present idea on this is that the voltage rating must be taken in
conjunction with the temperature rating: since both temp and voltage
increase leakage, a cap rated 85C and 25V is good for more than 25V if you
keep the temp well under 85C. And probably vice versa, too.

If I get a chance, I'll try the experiment in a temp-controlled oven, and
see if the cap that fared well at 48V 25C fares as well at 48V and, say,
65C. Need to build a temp-controlled oven, first - I don't think my wife
would be happy with me trying the experiment in our kitchen oven.
Click to expand...
I believe this reverse-biased-cap scenario always (?) occurs when
connecting two pieces of AV equipment, eg TV and VCR, DVD and
amplifier, etc.

C C
--||--o --> o--||--
+ - - +
device #1 device #2
output input

Actually, this arrangement looks like the NP substitution trick I
alluded to elsewhere in this thread.


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.
 
"Ross Herbert" <rherber1SPAMEX@bigpond.net.au> wrote in message
news:hldqi0ho73rcmi9emih3odi3oa3jverj2h@4ax.com...
Thus, an increase in temperature results in a lowering of the
breakdown voltage [...]
Click to expand...
Seems pretty clear. What works on the bench is not promised to work inside
a hot amplifier :)

Thanks for finding and quoting that.
 
Jim Williams published a very nice equation to predict electrolytic failure.
Temperature, voltage, ripple current, many terms affect their lifetime.

--
KC6ETE Dave's Engineering Page, www.dvanhorn.org
Microcontroller Consultant, specializing in Atmel AVR
 
"Dave VanHorn" <dvanhorn@cedar.net> wrote in message
news:76KdnRS8wYuyobLcRVn-qA@comcast.com...
Jim Williams published a very nice equation to predict electrolytic
failure.
Temperature, voltage, ripple current, many terms affect their lifetime.
Click to expand...
Got a reference for where I might find that? Google didn't turn anything
up.
 
"Dave VanHorn" <dvanhorn@cedar.net> wrote in message
news:tfydnQS5i-cxKLLcRVn-vg@comcast.com...
Hmm. My copy is out on loan, and I don't remember the title offhand..

I'm pretty sure it's this one.
http://www.amazon.com/exec/obidos/tg/detail/-/0750670622/qid=1093645686/sr=1-4/ref=sr_1_4/103-2372900-8017423?v=glance&s=books
Click to expand...

Doesn't seem to be. I have the book; just took a peek and didn't see
anything in the index nor in any of Jim's articles in there. That book is a
compilation of articles by various authors. I didn't look through all the
articles, but I don't remember seeing it last time I read it.

-walter
 
It's definitely in one of his.
It's an expansion of the half life for +10C, and adds in voltage, ripple
current Vs ripple current rating, and a few more terms.

I used it in a printer design, where I was operating from a 40W wall-wart,
supplying 400W pulses to a printhead, at maximum energy transfer rate. It
operates like a photoflash, the boost switcher charges up the output caps,
and when it comes into regulation, it triggers an int on the processor that
starts the next burn pulse.

The interesting part, was that I ended up using three cheap caps, because
when I ran the calclations, they beat out all the expensive single caps, and
by far, were (in this case) the least BANG for the buck. :)

One case had a relatively expensive "low ESR" cap, predicted to fail in
about 5 minutes!

That design has been in production now since '94, and cap failures are
essentially zero, so I guess it was valid.


--
KC6ETE Dave's Engineering Page, www.dvanhorn.org
Microcontroller Consultant, specializing in Atmel AVR
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top