Maker Pro
Maker Pro

Ceramic capacitors for higher voltage switchers

J

Joerg

Jan 1, 1970
0
Gents,

Anyone know what sort of ripple the FK22 series from TDK can take, say a
0.47uF/250V? I need some ceramics as output caps of a SEPIC that is
variable up to almost 100V, with some gusto (>50 watts). Can't use
electrolytics because of a harsh environment.

http://www.tdk.co.jp/tefe02/e4942_fk.pdf

Stacked SMTs are great but boutiquish and expensive. Anything
 
T

Tim Williams

Jan 1, 1970
0
Joerg said:
Anyone know what sort of ripple the FK22 series from TDK can take, say a
0.47uF/250V? I need some ceramics as output caps of a SEPIC that is
variable up to almost 100V, with some gusto (>50 watts). Can't use
electrolytics because of a harsh environment.

Well, take a guess at power dissipation and find the damping factor (X7R is
what, 1.5% or something?). If you're running hot, low power is probably a
good idea, like 1/8W or something. And use lots of copper (ha, like you
ever have the space).
http://www.tdk.co.jp/tefe02/e4942_fk.pdf

Stacked SMTs are great but boutiquish and expensive.

So stack them horizontally instead? :)

Or if you're cramped for horizontal space, maybe you could make a capacitor
daughterboard so they're stacked horizontally, on a vertical board...

Tim
 
J

Joerg

Jan 1, 1970
0
Tim said:
Well, take a guess at power dissipation and find the damping factor (X7R is
what, 1.5% or something?). If you're running hot, low power is probably a
good idea, like 1/8W or something. And use lots of copper (ha, like you
ever have the space).

I wish they were physically a tad larger.

So stack them horizontally instead? :)

Or if you're cramped for horizontal space, maybe you could make a capacitor
daughterboard so they're stacked horizontally, on a vertical board...

The client won't like that :)
 
J

Joerg

Jan 1, 1970
0
Robert said:
Just solder the caps in so they are on edge (vertical); one may be
able to use 4 in parallel this way and use the same PCB board space.


Can't do that, it must remain machine-solderable.
 
A

Archimedes' Lever

Jan 1, 1970
0
Can't use

Use hermetically sealed military versions then.

The difference in performance is worth the difference in price.
 
A

Archimedes' Lever

Jan 1, 1970
0
Just solder the caps in so they are on edge (vertical); one may be
able to use 4 in parallel this way and use the same PCB board space.


Pretty much very labor intensive assembly work. Not very good for any
automated process, so not good for anything that has intentions of large
volume production.

For short run, hand assembly though, no problem.

Vertical daughter cards are the way we did it in the HV lab.

50 Watts is a lot of loading for a ceramic cap bank to manage the
ripple of.
 
J

Joerg

Jan 1, 1970
0
Archimedes' Lever said:
Use hermetically sealed military versions then.

The difference in performance is worth the difference in price.


I know, and I do that. But it's often not easy to find some for higher
voltages with low enough ESR _and_ being available without huge leadtimes.
 
J

Joerg

Jan 1, 1970
0
mook said:
Haven't used these particular caps but I have run some X7R 250V surfamce
mount 2220 caps at 1A p-p ripple current at 10Khz. No
noticable heating at all.

A also run a 500V 2225 .47uF caps on the output of a 150V rail that can put
out 75W (300Khz ripple) with no noticable heating.

I can't use these large SMT sizes in this case because of thermal stress
issues, cracks, and all that.

You really have to hog some current on the ceramics to heat them up.

That said I did have some x7r .1uF cap with teh yellow epoxy turn red when a
5A audio amp IC started oscillating into it. It was in an RC damper on the
output. Scratched my head for a while trying to figure out why that cap
got hot. I had thought before that ceramic caps were nearly ideal. :)

That's one reason why I asked. I've managed to turn capacitor ceramic
into bubbly green glass. But that was several MHz and several amps
through just one. The usual, had only one, did a quick experiment, "Come
on, hang on another 10 secs" ... *PHOOMP*
 
L

legg

Jan 1, 1970
0
Gents,

Anyone know what sort of ripple the FK22 series from TDK can take, say a
0.47uF/250V? I need some ceramics as output caps of a SEPIC that is
variable up to almost 100V, with some gusto (>50 watts). Can't use
electrolytics because of a harsh environment.

http://www.tdk.co.jp/tefe02/e4942_fk.pdf

Stacked SMTs are great but boutiquish and expensive. Anything

You can get capacitor characteristics from TDK.

http://www.tdk.co.jp/ccv/index.asp

The resistance, reactance and voltage/temperature dependencies are
illustrated graphically.

A 0.1uF 100V X7R part in 1206/3216 body size, 3216X7R2A104M, exhibits
an internal series resistance of 30mOhms, over the range of common
power conversion frequencies.

RL
 
F

Fred Abse

Jan 1, 1970
0
That's one reason why I asked. I've managed to turn capacitor ceramic into
bubbly green glass.

Oppenheimer wannabe?
 
L

legg

Jan 1, 1970
0
Sounds trivial but 30mohms times 5A squared is .75W on a 1206 cap. You now
have made an LED. :)
At 100KHz, that would be 80Vac across a 0.1uF capacitor - a little
more ripple than is anticipatd here.

The internal series resistance of most LEDs is some order of magnitude
higher.

Degrees of triviality are possibly more relevant electronic units of
measure than most appreciate. Especially in this news group.

RL
 
B

Baron

Jan 1, 1970
0
Fred Abse Inscribed thus:
Oppenheimer wannabe?
:)

I've had ceramic caps get hot enough to melt the solder and fall off the
PCB.
 
J

Joerg

Jan 1, 1970
0
Robert said:
I bet pick-and-place can do it..by using a small dot of adhesive to
hold a cap that is placed, another dot nearby for nest cap, etc (all
dots could be placed first and then the caps).
Solder paste mask can be same as if one cap laid down in conventional
form.
"Never" say "never".

Sure, everything can somehow be done but once you deviate from industry
standard practice -> $$$.
 
J

Joerg

Jan 1, 1970
0
Thanks! Much easier than trying to crank that out of a sales channel.


Well, you wouldn't do it with one cap :)

Did that as a kid, never again, it resulted in a major bang and ceramic
pieces flying across the room.

At 100KHz, that would be 80Vac across a 0.1uF capacitor - a little
more ripple than is anticipatd here.

The internal series resistance of most LEDs is some order of magnitude
higher.

Degrees of triviality are possibly more relevant electronic units of
measure than most appreciate. Especially in this news group.

Hey, we haven't gotten into politics and global warming in this here
thread. Yet.
 
E

E

Jan 1, 1970
0
Joerg said:
Can't do that, it must remain machine-solderable.

I remember one old TV-tuner where every chip resistor was sitting on
their edge. And it was really tight, never seen board so full of components.
They were about 0603 size. I think it was Alps but not 100% sure.

-E
 
J

Joerg

Jan 1, 1970
0
E said:
I remember one old TV-tuner where every chip resistor was sitting on
their edge. And it was really tight, never seen board so full of components.
They were about 0603 size. I think it was Alps but not 100% sure.

For a TV tuner you can do that because of the sheer production volume.
Any setup or NRE costs pale compared to the total machine runtime. I
once had a hybrid produced at a tuner factory. I thought our qty was
high, until I saw the place. A room the size of a soccer field, tons of
machines hissing and clanging 24/7. >90% of the output was always the
same TV tuner and our li'l runs of maybe 10k at a time were mere drops
in the bucket. At night there was almost nobody other than a night
guard, all fully automated.
 
J

Joerg

Jan 1, 1970
0
Michael said:
You have been reported to the 'Society For The Prevention Of Cruelty
To Components'.

Now, wait! Have you seen John Larkin's collection of mutilated TO220
MOSFETs? So if there is a tribunal he should have to go first ...

--
SCNR, Joerg :)

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
 
J

Joerg

Jan 1, 1970
0
Fred said:
Oppenheimer wannabe?

I was a teenager back then and couldn't believe it, stood there in awe.
If was the same sort of green as a Heineken bottle, with tiny bubbles
frozen inside.

Unfortunately some other stuff bit the dust when that happened and I
didn't have a shortwave power amp for a while. Among other things a
chunk from the plates of an air-variable capacitor went AWOL (I figured
that must have been where the ozone stench came from) and its shaft
would no longer turn. And the coax to the wattmeter was a goner.
 
F

Fred Abse

Jan 1, 1970
0
Agreed, when Xc is considered the ripple votlage would be HUGH for a
power supply even at 5X that frequency.

Lawrie?
 
Top