# Chosing a High Ripple Current Capacitor

Discussion in 'Electronic Design' started by D from BC, Apr 22, 2007.

1. ### D from BCGuest

In spice I have these waveforms across the capacitor:

Voltage: 5Vpp 100 Khz triangle wave on 270VDC
Current: 2App (swings -1A to +1A.) square wave at 100Khz.
Capacitance: 1uF
Max. Ambient temp: 30C

Would a polypropylene film or electrolytic be better? There's lots of
spare PCB space..I prefer picking the longer life component..

I've read that heat ages electrolytics and one way to heat up an
electrolytic is with lots of ripple current to make I^2R heat.

When I looked at the polypropylene data sheets
http://www.vishay.com/docs/28124/mmkp383.pdf
I found AC vs frequency derating graphs.
What is this derating based on?? Heat? Material breakdown?
My app only has 5Vpp of AC and so a 400VDC pp film cap falls into the
safety zone at 100Khz..
But will it outlast an electrolytic?
How do I figure out if this cap gets toasty from the pulse current?
Do I look at the tangent of loss angle spec?
D from BC

2. ### EeyoreGuest

Polypropylene every day.

You'd have to use back to back electrolytics or a non-polar one for an AC
circuit and I doubt you can get a 1uF electrolytic that'll withstand that ripple
current anyway.

I assume it's heat that's the limiting factor btw. They do get hot in use. A mix
of foil resistance and dielectric loss I imagine. Tangent of loss angle is just
another way of looking at it.

Graham

3. ### Paul Hovnanian P.E.Guest

If I understood the OP, its 5Vpp on top of 270Vdc, so a polarized cap
would be OK.

4. ### D from BCGuest

After some googling, I found

tan d = DF = 1/Q = ESR/Xc

Ok... The polypropylene I picked has a "tangent of loss angle"
=<50x10^-4 at 100Khz

tan (50x10^-4) = ESR/(1/(2*pi*f*C))

ESR = 0.00014 ohms... <<< That looks bogus..
D from BC

5. ### EeyoreGuest

Ooops ! I missed the obvious.

An amp or so of ripple is still a lot for a 1uF electrolytic.

Graham

6. ### D from BCGuest

I noticed that..
It's true the DC level is dominant and the voltage ripple is tiny in
comparison. Who's cares if the current reverses polarity as long as
the voltage doesn't..Therefore..polarized cap is ok.

I might try the math sometime on paralleling electrolytics to compete
with the poly film.
But I hate math...so I'm just going to use the film cap.

D from BC

7. ### WimpieGuest

Hello D from BC

For relative low frequencies, the maximum AC voltage is limited
because of internal corona discharge in air pockets, that may erode
the dielectric. At high frequencies, heat generation inside the
dielectric is the limiting factor. At even higher frequencies, the
limiting factor is current carrying capability of the contacts.
Mostly the graphs are based on 10 degrees internal temperature rise.

You also have to derate when the temperature will be higher (for
example because of heat generated by nearby components.

Electrolytic capacitors.
You need a big cap (several 100 uF to mF) to handle a ripple current
of about 600mA rms. The larger the size, the higher the ripple current
capability. Ripple current is mostly based on 5..10 degrees
temperature rise with respect to ambient. Same current at lower
temperature gives somewhat higher temperature rise because of higher
ESR at low temperature. The life time is specified at given
temperature, rated voltage and rated ripple current (I thought).

When you stay below the maximum ripple current, you gain a factor 2
for every 10 degrease lower temperature then specified in the data
sheet. Like with the foil capacitor, check the actual temperature of
the electrolytic capacitor. When you use them according to the
specification, they have a predictable lifetime. Life time can be
shortened significantly by wrong treatment (for example certain
chemicals for cleaning). Most manufacturers have good information on
how to calculate the life time based on ripple current, frequency,
ambient temperature and voltage and how to treat their capacitors (for
example ELNA, Nichicon, Panasonic).

situations? Standard or SMPS capacitors are not made for fast
discharging.

When the value of the capacitor is of importance, a foil capacitor is
a good option.

Best Regards,

Wim
PA3DJS

8. ### D from BCGuest

Cool ...lots of info for me to think about..

Thanks..
Especially the bit about film cap derating graphs are based on a 10
degree temp rise...I didn't know that....
If needed, I'll check out the capacitor manufacture sites for
D from BC

9. ### Paul Hovnanian P.E.Guest

I calculated : 50x10^-4 = ESR/(1/(2*pi*f*C))

tan d = 50x10^-4 so you don't need the tan (50x10^-4) term

Take a look at the V vs f curves for the caps. These appear to be
constant power curves (I goes up proportional to f). Pick a V for your C
and f, then calculate its Xc and I.

Compare that VI to your current waveform power dissipation through the
ESR and make sure you don't exceed the curve value.

10. ### Boris MoharGuest

Check out http://www.eci-capacitors.com/

11. ### D from BCGuest

Doh!... Newbie goof..Thanks for the correction..
....I was thinking angle ..That's why I did tan ( )..
Good thing I don't make life support systems for a hobby..

So... it's really a DF number. I was wondering why the spec was
unitless. If it was an angle it would be in degrees or radians.

In my case the poly-cap dissipation would be around (1Arms^2)*0.008 =
8mW = insignificant
D from BC

12. ### D from BCGuest

Interesting..
Electronic Concepts Inc. makes polycarbonate capacitors..
DF does not exceed 0.3%..from 25C to 125C

Digikey distributes only these poly's
Polyester
Polyethylene
polypropylene
polyphenylene
D from BC

13. ### WimpieGuest

Hello, D from BC

Polyethylene and Polypropylene have the lowest maximum operating
temperature, but lowest dielectric losses and best capacitance versus
temperature stability.

Polyester can be used up to 125 degr (derate), and have higher loss.
Polyphenylene has highest operating temperature and moderate loss.
Polycabonate is not that popular anymore (but it is not a bad
dielectric).

You should check the datasheet. Also you should be aware of the
construction. Wounded capacitors have some higher inductance then
"stacked" capacitors. I had it once in a half bridged semi resonant
converter. Most of the ringing was caused by the (wounded technology)
capacitor.

Epcos has a general document on foil capacitors that may help you
choose the best capacitor.

Best regards,

Wim

14. ### John LarkinGuest

Yup, nothing works like doing the math!

John

15. ### Paul Hovnanian P.E.Guest

Oh, I don't know about that. Blowing up ever increasing rated components
until one holds can be fun too.

Remind me and some time I'll post my story about destructive tests with
a bunch of \$8000 relays.

16. ### John LarkinGuest

Actually, I do enjoy destructive testing. It can be, well, very
constructive.
Now?

Who makes \$8000 relays? Are they surface mount?

John

17. ### D from BCGuest

I have a 1/8W carbon resistor for sale... \$500.00 sound ok?

I wanna hear the relay story...
I know the end!
"And all the other relays lived happily ever after...The end.."

\$8000 relays....mmmm
Perhaps liquid cooled copper tube windings on a core the size of a car
that pulls down a contact the size of a telephone pole???
D from BC

18. ### YDGuest

Late at night, by candle light, "Paul Hovnanian P.E."
Do it *NOW*!

- YD.

19. ### Paul Hovnanian P.E.Guest

Years ago, I was working on the design certification for the 747-400
electrical power system. They had upgraded the system to 4 paralleled 90
kVA generators, so a lot of the circuit breaker coordination and
equipment maximum fault duty had to be reviewed.

One part of the system included a 3PDT relay, rated at 10A, with a
specified low voltage dropout coil. Its use was to select between a
primary source and, if that dropped below a threshold, drop to the NO
position, selecting an alternate 3 phase power bus.

Since this gizmo had a couple of equipment circuit breakers downstream,
we decided to test the system coordination by applying a fault at one of
them, downstream of this relay. With 4 fully loaded generators in
parallel, we applied a fault and 'Pop'. The relay belched fire and
smoke. Needless to say we reported our findings back to design
engineering, who forwarded it to the vendor.

When the new test sample arrived, we ran it down to the lab and repeated
the test. 'Poof!', a somewhat smaller ball of fire. After about half a
dozen tests, each with smaller pyrotechnics, we finally got a unit that
survived.

When we were all done, the purchasing department contacted engineering
management and asked if we were through blowing up relays at \$8000 each.
Due to the organizational philosophy of the company, engineering had no
visibility of component costs. Had we (or more accurately, design
engineering) known what these custom relays cost, we (or they) could
have easily designed a circuit with a standard (probably a couple of
hundred dollar unit being aviation spec.) part and a voltage sensing
circuit.

That's why airplanes cost a couple of hundred million each.

20. ### YDGuest

Late at night, by candle light, "Paul Hovnanian P.E."
Yabbut, what caused the fireworks, underdimensioned contacts, voltage
swing or something?

- YD.