ceramic caps in a power supply

[Adam S]

I designed a 12V 25A step down converter switching at 300kHz. Input
capacitor ripple current is around 11A RMS. In an attempt to reduce
board space and cost, for the input capacitors I chose 6 x 10uF 25V
ceramics over low ESR electrolytics.
Unfortunately I discovered the hard way that these large ceramics
experience horrifying drop in capacitance at their rated voltage.

Here are some measurements I made of various capacitors I bought from
Farnell.

capacitor Capacitance at DC bias
type 0 VDC 25 VDC %
-------------------------------------------------------
Kemet 10uF 25V X7R 1210 9.3u 4.0u %43
Taiyo 10uF 25V X5R 1210 11.2u 3.1u %27
Taiyo 10uF 25V X5R 1206 9.9u 1.7u %17
Kemet 10uF 10V X7R 1206 9.2u 2.2u %24
? 1uF 25V X7R 1206 0.97u 0.83u %85


Should this be expected ? It appears one needs to triple the capacitor
count to achieve the rated CV value.


Adam

 

[nospam]

Unfortunately I discovered the hard way that these large ceramics
experience horrifying drop in capacitance at their rated voltage.

Should this be expected ?

Yes you should have expected it. It is an inherent feature of the
dielectric and the more compact the case the more pronounced the effect is.

At half rated voltage a 1206 might give you 90% of the nominal capacitance
while an 0805 only 40% for example.
--

 

[Adam S]

Yes you should have expected it. It is an inherent feature of the
dielectric and the more compact the case the more pronounced the effect is.

At half rated voltage a 1206 might give you 90% of the nominal capacitance
while an 0805 only 40% for example.

Ok, thanks. I did expect some drop, maybe maximum %80 of nominal
capacitance, but not %40.

I since found the following article summarising reasons MLCCs have high
voltage sensitivity.

http://www.kemet.com/kemet/web/home...capacitor drops to 37 uF- 30 uF- or lower.pdf

 

[[email protected]]

How are you measuring this?

With a capacitance meter across a pre-charged 4700uf electrolytic in
series with the DUT.

You're using reasonably decent dielectrics there -- the variation you're
seeing is *much* greater than what I'd expect (...which would be no more than
~10% and generally a 5% or less drop). Can you get an actual data sheet from
the manufacturer?

Here's a web page with results that are in-line with what I've previously
seen:http://www.cliftonlaboratories.com/capacitor_voltage_change.htm

---Joel

It seems the modern high CV surface mount ceramics (especially smaller
packages) suffer a great deal more voltage sensitivity than the
through hole parts listed on that web page. Its just plain annoying
how ceramic capacitor marketing has purposely hushed the truth about
voltage coefficients. The X7R, X5R..et ratings only apply to
temperature coefficients and there is no such standards for voltage
coefficients. The link John Popelish posted described the situation
clearly. Manufactures only require capacitance ratings at 0V DC bias
conditions. Capacitance change at the specified voltage rating is
completely uncontrolled and unspecified by most manufactures. The most
a data sheet may provide is a graph of C vs V for one to two standard
capacitors, e.g 100nF 50V 0805.

Adam Seychell

 

[LVMarc]

Yep. That's what happens. You should read the datasheets. The cheaper
dielectrics are even worse.


Bob

funny whe you need dc/dv you pay extra for varators

Marc