COG/ (NPO) ceramic caps, I must be the last to know.

If you don't need really high values and the application is not cost-sensitive then I don't see much reason to use film caps.

I've recently used them in a protoype where low microphonics is important, but I've not tried to test NPO caps to see if there is actually any problem (I suspect not). I bet I could create a problem through pathological processing, but that's another story..

re

Hi Geo, Seph, I have used lots of the newer C0G/NP0 caps with great results. The only gotcha is that they are not 'self healing" as the better films are. Not a good idea to place at AC inputs where big spikes reside. I had a 540W LLC converter that improved efficiency from 93% to 96.5% when the X7Rs were changed to C0G. The X7R have terrible voltage coefficients, 50% loss of capacity at max volts. Cheers, Harry

I did the actual measurement of a distortion created by a voltage coefficient of an X7R capacitor. I made a basic RC lowpass circuit and tested it for THD at 1 Vpp at the cutoff frequency. The X7R was 0.1uF rated at 50V. The THD was 0.01% ballpark. Although a film cap performed better by two orders, the distortions because of the voltage coefficient in X7Rs aren't all that terrible.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Yup, try 7Vpk on a 10V X7R and see what you get. Harry

re

Yeah the 0.1uF COG's are a bit pricey. But I'm dying to try one in a high Q application.

George H.

here

Thanks Harry, that's good to know.

George H.

IIRC their soakage is a bit worse than polystyrene or polyprop, but they're excellent otherwise. Also nice and stable at high temperature, since the dielectric isn't ferroelectric.

Cheers

Phil Hobbs

I have a paper somewhere around here that uses the nonlinearity of ceramic caps to make a shock line, sharpening up the rise time of 2KV voltage steps.

John

Yes, they are pricey. But, if you need high Q in ceramic, don't use anything other than C0G.

Cheers, John

In my experience, NPO ceramic caps are great. I somewhat remember COG to be similar to NPO.

My experience here is impressively low tempco and availability in +/- 5% tolerance - maybe even sometimes tighter now.

My next-favorite of ceramics is X7R. Usual tolerance is +/- 10% and the temperature characteristics look good to me. These often look better to me than film types due to being much more compact, often also due to having lower effective series resistance and inductance at higher frequencies.

After that comes the Y and Z types, especially Y5 and Z5 types, such as the common Z5U. In my experience, Z5U usually has tolerance of +80/-20 % at 25 C, and severe decrease of capacitance when temperature gets away from 25 C (mainly outside the 5-45 or 0-50 C range).

They're OK in small values, like a penny each at roughly 330pF/50V and down. We pay 27 cents for 1.8 nF and 62 cents for 22nF.

John

Soakage?? Define please. Mikek

Dielectric absorption. I think soakage is a British term.

ere

I think I saw that somewhere. Do I bang on 'em with a square wave to measure soakage? Loook for a long tail.

George H.

(opps spelling mistake was not intended.)

there

?)

Yeah, There's a few high Q circuits where a COG may work. What's the effect of soakage? Say, it's an L/C parallel resonance with a Q of

George H.

Coil resistance is ~10 ohms,

s.

=A01k)

there

n?)

hy

Hmm, must be a cost per layer in there somewhere. Are the small ones (

s.

ere

n?)

xt -

I've got a graph of dielectric absorption vs frequency for different caps by B. Pease (RIP). He called it soakage too.

Looks like there are some ceramics that are good and some not. Or maybe it changes batch to batch? Are there any better vendors, soakage-wise?

George H.

=92s.

there

ion?)

re,

text -

I think if you are looking for a feature in a cap not known for that feature, you are basically picking parts for a "one-off" rather than production. So you find a ceramic with decent dielectric absorption and it works today. But formulations change, factories are relocated, etc. If there are no datasheet limits, there is no assurance the parameter will be controlled. This is real life. I've seen bean counters change the wafer manufacturer and THD (something not always tested) change by an order of magnitude.

Whatever the sample and hold guys use is what you want. They have beat dielectric absorption to death. Polystyrene comes to mind, but the caps are not very compact. Teflon is expensive.

BTW, a well known trick when testing caps is to use multiple units in a manner to make the parameter look worse. For dielectric absorption, parallel up 10 caps when you test them. This gets the test signal out of the noise.

I have tested them, they are definitely better than X7R.