Capacitors Used in Power Supplies

[Jonathan Mohn]

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Thanks!

-Jonathan

 

[John Larkin]

On Tue, 03 Feb 2004 18:43:31 GMT, "Jonathan Mohn"
<j.c.mohn@worldnet.att.net> wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Thanks!

-Jonathan

Polarized 'lytics are just denser than other cap types by pretty big
factors: better than 2:1 over n.p. electrolytics, and roughly
thousands denser than film caps.

Something like 10,000 uF at 25 volts, a modest electrolytic, would be
outrageously big and expensive any other way. Scan a Digikey or Mouser
catalog for some comparisons. Film or ceramic caps *would* be more
reliable.

John

 

[tweak]

On Tue, 03 Feb 2004 18:43:31 GMT, "Jonathan Mohn"
<j.c.mohn@worldnet.att.net> wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Yes replacing with better quality capacitors is always a good idea.

I use epoxy sealed with the highest thermal rating I can get.
Some of the more knowledgeable techs here can tell you how to change
up to a better rated capacitor as well. This can eliminate blown caps
completely

Thanks!

-Jonathan

 

[John Popelish]

Jonathan Mohn wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

The power supply would be twice as big and cost ten times as much.

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

The two I already mentioned.

A capacitor must contain two conductive surfaces and an insulation
layer that separates them and stores energy as an electric field is
applied to it. Most non electrolytic capacitors have flat plates of
metal with flat a layers of some dielectric between them. In film
capacitors, this structure is often rolled up to save space.
Electrolytic capacitors have one metal plate and one conductive liquid
plate with only the oxide layer on the metal as a dielectric
separator. Even with a flat plate, this way of coating a metal
surface with a uniform layer of insulation with the other plate
intimately enclosing the metal plate is pretty space efficient. But
the metal plate in the electrolytic is electrochemically etched so
that it is essentially a metal sponge with a hugh surface area
compared to the original metal foil. The nit is oxidized
electrolytically and covered with the conductive liquid that forms the
second plate. This process multiplies the capacitance per volume by a
huge factor. Unfortunately, you cannot reverse the voltage or the
oxide will be removed as atomic hydrogen is produced at the surface of
the metal, which combines with the oxygen to form water. And you have
to live with the difficulties of a component that contains a liquid
component. The liquid also is not as conductive as metal, so
capacitive current heats it quite a bit. But if you can keep it cool,
and not expose it to reverse voltage or excess voltage it can have a
very useful life. Something like 10 years for good quality examples.
The short life given in the specs assumes the internal temperature is
constantly at the rated maximum.

--
John Popelish

 

[Jonathan Mohn]

Thanks for all the answers. I didn't realize that they were THAT much
smaller. Wow!

-Jonathan


"Jonathan Mohn" <j.c.mohn@worldnet.att.net> wrote in message
news:nZRTb.51144$6O4.1525090@bgtnsc04-news.ops.worldnet.att.net...

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Thanks!

-Jonathan

 

[Jonathan Mohn]

Thanks for the detailed response, John. That is really fascinating. I'm
continually astonished at the sophistication of these components. There
sure are a lot of clever folks out there!

-Jonathan


"John Popelish" <jpopelish@rica.net> wrote in message
news:402020E1.755F140C@rica.net...

Jonathan Mohn wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a
polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

The power supply would be twice as big and cost ten times as much.

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

The two I already mentioned.

A capacitor must contain two conductive surfaces and an insulation
layer that separates them and stores energy as an electric field is
applied to it. Most non electrolytic capacitors have flat plates of
metal with flat a layers of some dielectric between them. In film
capacitors, this structure is often rolled up to save space.
Electrolytic capacitors have one metal plate and one conductive liquid
plate with only the oxide layer on the metal as a dielectric
separator. Even with a flat plate, this way of coating a metal
surface with a uniform layer of insulation with the other plate
intimately enclosing the metal plate is pretty space efficient. But
the metal plate in the electrolytic is electrochemically etched so
that it is essentially a metal sponge with a hugh surface area
compared to the original metal foil. The nit is oxidized
electrolytically and covered with the conductive liquid that forms the
second plate. This process multiplies the capacitance per volume by a
huge factor. Unfortunately, you cannot reverse the voltage or the
oxide will be removed as atomic hydrogen is produced at the surface of
the metal, which combines with the oxygen to form water. And you have
to live with the difficulties of a component that contains a liquid
component. The liquid also is not as conductive as metal, so
capacitive current heats it quite a bit. But if you can keep it cool,
and not expose it to reverse voltage or excess voltage it can have a
very useful life. Something like 10 years for good quality examples.
The short life given in the specs assumes the internal temperature is
constantly at the rated maximum.

--
John Popelish

 

[John Popelish]

Jonathan Mohn wrote:

Thanks for the detailed response, John. That is really fascinating. I'm
continually astonished at the sophistication of these components. There
sure are a lot of clever folks out there!

You are welcome. There is a lot to know. If you want to learn more
about various kinds of capacitors, this is a good site.
http://my.execpc.com/~endlr/index.html

--
John Popelish

 

[DarkMatter]

On Tue, 03 Feb 2004 11:02:54 -0800, John Larkin
<jjlarkin@highSNIPlandTHIStechPLEASEnology.com> Gave us:

On Tue, 03 Feb 2004 18:43:31 GMT, "Jonathan Mohn"
j.c.mohn@worldnet.att.net> wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Thanks!

-Jonathan


Polarized 'lytics are just denser than other cap types by pretty big
factors: better than 2:1 over n.p. electrolytics, and roughly
thousands denser than film caps.

Not really. Same density, it just happens that an NP cap is really

two caps back to back inside the can. Always was... always will be.

Something like 10,000 uF at 25 volts, a modest electrolytic, would be
outrageously big and expensive any other way. Scan a Digikey or Mouser
catalog for some comparisons. Film or ceramic caps *would* be more
reliable.

Where are those 10,000 uF ceramic caps when you need one!
Oh that's right! I do NOT have a space on my board the size of a
commercial BRICK! DOh!

You could have at least mentioned the difference between the
different forms.

 

[Jonathan Mohn]

Thanks for the great link! I had found several sites on capacitors and
capacitance, but I had not come across this one, yet.

-Jonathan


"John Popelish" <jpopelish@rica.net> wrote in message
news:40203824.7D8020F5@rica.net...

Jonathan Mohn wrote:

Thanks for the detailed response, John. That is really fascinating.
I'm
continually astonished at the sophistication of these components. There
sure are a lot of clever folks out there!

You are welcome. There is a lot to know. If you want to learn more
about various kinds of capacitors, this is a good site.
http://my.execpc.com/~endlr/index.html

--
John Popelish

 

[John Larkin]

On Tue, 03 Feb 2004 18:40:04 -0800, DarkMatter
<DarkMatter@thebarattheendoftheuniverse.org> wrote:

On Tue, 03 Feb 2004 11:02:54 -0800, John Larkin
jjlarkin@highSNIPlandTHIStechPLEASEnology.com> Gave us:

On Tue, 03 Feb 2004 18:43:31 GMT, "Jonathan Mohn"
j.c.mohn@worldnet.att.net> wrote:

In virtually every schematic I've seen for regulated power supplies, the
storage capacitor that is used after rectifying the current is a polarized
capacitor. Why is a polarized capacitor always specified? What would
happen if I used a non-polarized electrolytic capacitor?

Also, I've read that electrolytic capacitors are fairly short lived. Is
there any downside to using a longer lived type of capacitor in this
function, aside from the size advantages of the electrolytic?

Thanks!

-Jonathan


Polarized 'lytics are just denser than other cap types by pretty big
factors: better than 2:1 over n.p. electrolytics, and roughly
thousands denser than film caps.

Not really. Same density, it just happens that an NP cap is really
two caps back to back inside the can. Always was... always will be.

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Something like 10,000 uF at 25 volts, a modest electrolytic, would be
outrageously big and expensive any other way. Scan a Digikey or Mouser
catalog for some comparisons. Film or ceramic caps *would* be more
reliable.

Where are those 10,000 uF ceramic caps when you need one!
Oh that's right! I do NOT have a space on my board the size of a
commercial BRICK! DOh!

0805 size ceramics are up to 100 uF now, but the voltage is low and
they're still expensive.

You could have at least mentioned the difference between the
different forms.

Oh, please tell us! Please!

John

 

[DarkMatter]

On Tue, 03 Feb 2004 19:54:37 -0800, John Larkin
<jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Exactly... so the density is the same. The size of the cap changes
for any given value because there are TWO... two... two caps in one!

It isn't half the CV product, it is two seriesed caps... period.
Anybody (any electrical idiot) knows that it would take twice the
capacitance to get the same value.

 

[DarkMatter]

On Tue, 03 Feb 2004 19:54:37 -0800, John Larkin
<jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

0805 size ceramics are up to 100 uF now, but the voltage is low and
they're still expensive.

Yes, and not always the choice to make. I have several times noted
that an 0402 0.1 uF cap, and an 0805 or 1206 0.1uF cap behave
differently in circuit, even with the same substrate.

We have a hard start cap for an oscillator that hates being a small
form factor.

I guess the answer could be that they perform differently with the
harmonics and such present in the signal, but there is no directly
observable reason. Still the fact remains that it will not work with
the smaller form factors in that location.

You could have at least mentioned the difference between the
different forms.


Oh, please tell us! Please!

Don't be a fucktard, boy.

 

[Bob Myers]

"DarkMatter" <DarkMatter@thebarattheendoftheuniverse.org> wrote in message
news:hsv0201h7kvdpu92u7vjvhi9go5hpgh8do@4ax.com...

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Exactly... so the density is the same. The size of the cap changes
for any given value because there are TWO... two... two caps in one!

Gee, exactly what is it you think "density" refers
to here, Dark, old man? The important "density"
when it comes to power supply "filter" (or "energy
storage") caps is how much charge you can put
into a given volume. For a specific dielectric material,
in this case AlO2, in a given volume you can either
cut the capacitance in half but keep the working
voltage the same (the two caps in series situation),
or double the capacitance at half the working voltage
(two in parallel, which doesn't do diddly for the
NP case which was the original question). But
since Q = CV, in either case the amount of stored
charge HAS to be cut in half. Half the charge in a
given volume = half the charge density.

It IS "half the CV" product that's important here, since
the job of these caps is to keep the current up
(i.e., supplying charge) to "smooth out the ripples"
that raw rectified-AC would otherwise present. Which
is, then, exactly why these supplies use polarized
caps - they provide the greatest capacitance, at the
necessary working voltage, in a given volume.

Bob M.

 

[DarkMatter]

On Thu, 5 Feb 2004 17:39:14 -0700, "Bob Myers"
<nospamplease@address.invalid> Gave us:

Gee, exactly what is it you think "density" refers
to here, Dark, old man? The important "density"
when it comes to power supply "filter" (or "energy
storage") caps is how much charge you can put
into a given volume.

I read what you (or whomever) wrote, and I know what was meant. I
am telling you that at the maximum density modern manufacturers are at
for a given voltage, a single cap of a given value back to back with
another to form a non-polarized cap at half that value, is going to
displace the same volume in a single can configuration. So 2 100uF
caps makes a single 100uF cap that takes up twice as much space, even
if made inside a single can construction.

The figure stated was "double" IIRC by whomever, and I am agreeing.
Sheesh... Don't get yer bloomers in a bunch.

That is a correct figure. The reason seems different though. I
simply stated that it takes twice as much capacitance to do the job
since at any given cycle period one is charging, and the other is
conducting like a short. It isn't calculated like two caps in series
as would true non polarized caps or polarized caps in series,
"properly polarized".

Is that more clear? Old man? Or have I "screwed up" again?

right.

 

[John Larkin]

On Tue, 03 Feb 2004 21:16:56 -0800, DarkMatter
<DarkMatter@thebarattheendoftheuniverse.org> wrote:

On Tue, 03 Feb 2004 19:54:37 -0800, John Larkin
jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Exactly... so the density is the same. The size of the cap changes
for any given value because there are TWO... two... two caps in one!

It isn't half the CV product, it is two seriesed caps... period.
Anybody (any electrical idiot) knows that it would take twice the
capacitance to get the same value.

If you put two 100v-rated film caps back-to-back, you can put twice
rated voltage, 200v, across the series pair. So energy stored per unit
volume is the same as you'd get from a single film cap of the same
technology. The problem is that putting two 100 volt polarized caps
back-to-back only allows 100 volts DC to be applied, even though the
volume has doubled. You can't back-bias an electrolytic film
steady-state - that would destroy it - while you *can* bias a plastic
film in any direction as long as you like.

Nonpolar 'lytics are funny. A 100 uF nonpolar electrolytic is
essentially two 200uF caps back-to-back. For AC, it looks like a
(crappy) 100 uF cap. For DC, one of the caps is back-biased, so it
leaks and discharges, so the whole thing sort of looks like a very
lossy 200 uF. This is a *very* non-ideal part.

John

 

[John Larkin]

On Tue, 03 Feb 2004 21:21:39 -0800, DarkMatter
<DarkMatter@thebarattheendoftheuniverse.org> wrote:

Oh, please tell us! Please!

Don't be a fucktard, boy.

I'll take that for a 'no.'

John

 

[DarkMatter]

On Thu, 05 Feb 2004 20:52:57 -0800, John Larkin
<jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

On Tue, 03 Feb 2004 21:16:56 -0800, DarkMatter
DarkMatter@thebarattheendoftheuniverse.org> wrote:

On Tue, 03 Feb 2004 19:54:37 -0800, John Larkin
jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Exactly... so the density is the same. The size of the cap changes
for any given value because there are TWO... two... two caps in one!

It isn't half the CV product, it is two seriesed caps... period.
Anybody (any electrical idiot) knows that it would take twice the
capacitance to get the same value.

If you put two 100v-rated film caps back-to-back, you can put twice
rated voltage, 200v, across the series pair.

There is no "back to back" with np type caps. They are just plain
"series" configured. And yes, the voltage drops across them split, if
they are of an equal value.

So energy stored per unit
volume is the same as you'd get from a single film cap of the same
technology. The problem is that putting two 100 volt polarized caps
back-to-back only allows 100 volts DC to be applied, even though the
volume has doubled.

No shit, and the capacitance math isn't the same either, as one
appears as a short one way, and the other does the other way.

You can't back-bias an electrolytic film
steady-state - that would destroy it - while you *can* bias a plastic
film in any direction as long as you like.

When an electrolytic NP is made one can never "back bias" one, as
it will conduct as short above a certain voltage, and during that time
the other will charge. Once charged, there is no longer a current
flowing for the "back biased" cap to have any pressure on it.

When conducting the other way, the opposite occurs for both caps.

Nonpolar 'lytics are funny. A 100 uF nonpolar electrolytic is
essentially two 200uF caps back-to-back. For AC, it looks like a
(crappy) 100 uF cap. For DC, one of the caps is back-biased, so it
leaks and discharges, so the whole thing sort of looks like a very
lossy 200 uF. This is a *very* non-ideal part.

It's even more complex than that, but likely a bit safer than you
make it sound when implemented correctly.

 

[Keith R. Williams]

In article <bd86201do9cgkjsdr87grgjf3vdufvjiov@4ax.com>,
DarkMatter@thebarattheendoftheuniverse.org says...

On Thu, 05 Feb 2004 20:52:57 -0800, John Larkin
jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

On Tue, 03 Feb 2004 21:16:56 -0800, DarkMatter
DarkMatter@thebarattheendoftheuniverse.org> wrote:

On Tue, 03 Feb 2004 19:54:37 -0800, John Larkin
jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> Gave us:

Well, when you put the two AlO2 dielectric layers back-to-back, the
capacitance drops in half. That gives you half the CV product for a
given can size. Always was... always will be.

Exactly... so the density is the same. The size of the cap changes
for any given value because there are TWO... two... two caps in one!

It isn't half the CV product, it is two seriesed caps... period.
Anybody (any electrical idiot) knows that it would take twice the
capacitance to get the same value.

If you put two 100v-rated film caps back-to-back, you can put twice
rated voltage, 200v, across the series pair.

There is no "back to back" with np type caps. They are just plain
"series" configured. And yes, the voltage drops across them split, if
they are of an equal value.

You don't read so well do you DimBulb. Perhaps you should go
back to second grade, where you left off.

So energy stored per unit
volume is the same as you'd get from a single film cap of the same
technology. The problem is that putting two 100 volt polarized caps
back-to-back only allows 100 volts DC to be applied, even though the
volume has doubled.

No shit, and the capacitance math isn't the same either, as one
appears as a short one way, and the other does the other way.

You can't back-bias an electrolytic film
steady-state - that would destroy it - while you *can* bias a plastic
film in any direction as long as you like.

DimBUlb, do you have a *clue* what an NP electrolytic is? ...I
thought not. Yikes, you are *DENSE*.

When an electrolytic NP is made one can never "back bias" one, as
it will conduct as short above a certain voltage, and during that time
the other will charge. Once charged, there is no longer a current
flowing for the "back biased" cap to have any pressure on it.

Huh? You are a retard. Please read what you wrote in the
previous paragraph and tell us what it means! Jeezum!

When conducting the other way, the opposite occurs for both caps.

Nonpolar 'lytics are funny. A 100 uF nonpolar electrolytic is
essentially two 200uF caps back-to-back. For AC, it looks like a
(crappy) 100 uF cap. For DC, one of the caps is back-biased, so it
leaks and discharges, so the whole thing sort of looks like a very
lossy 200 uF. This is a *very* non-ideal part.

It's even more complex than that, but likely a bit safer than you
make it sound when implemented correctly.

It's certainly more complex than you're able to comprehend.

--
Keith

 

[DarkMatter]

On Fri, 6 Feb 2004 22:45:58 -0500, Keith R. Williams
<krw@attglobal.net> Gave us:

DimBUlb, do you have a *clue* what an NP electrolytic is? ...I
thought not. Yikes, you are *DENSE*.

You retarded bastard. We use them in our designs at work, and we

also configure our own NP electrolytic storage configurations in house
with two caps so configured.

Ever see a laser pulser bank? I thought not. Try RIFA, and
Maxwell.

 

[Keith R. Williams]

In article <5q0920h6am0rs4j7uidkulm7r04ndmglme@4ax.com>,
DarkMatter@thebarattheendoftheuniverse.org says...

On Fri, 6 Feb 2004 22:45:58 -0500, Keith R. Williams
krw@attglobal.net> Gave us:

DimBUlb, do you have a *clue* what an NP electrolytic is? ...I
thought not. Yikes, you are *DENSE*.

You retarded bastard. We use them in our designs at work, and we
also configure our own NP electrolytic storage configurations in house
with two caps so configured.

Then how do you explain these words (they're yours)?

There is no "back to back" with np type caps. They are just
plain "series" configured. And yes, the voltage drops across
them split, if they are of an equal value.

NP electrolytic capacitors *are* back-to-back electrolytics.
Putting them in series doesn't make them NP.

John Larkin was exactly right. You really should try to learn
from him, rather than constantly demonstrate your incompetence.

Ever see a laser pulser bank? I thought not. Try RIFA, and
Maxwell.

Again you bring up irrelevant issues to try to wiggle out of you
ignorance.

--
Keith