NPO ceramic vs polypropylene

[Spehro Pefhany]

Hi guys,

Spehro Pefhany Inscribed thus:


How do glass capacitors fare in relation ?
Thanks:

Interesting.. I think they're better again, I've not used them.
Perhaps better than anything else with a reasonable epsilon.

AVX shows them up to 2400pF.

http://www.avx.com/pt.pdf

See the DA curve on the 3rd-last page.

Judging by all the radiation and MIL spec stuff they don't look very
inexpensive.

--sp

 

[Anthony Stewart]

I had always avoided ceramics in S&H circuits due to bad DA memory effects.

If you have high reactive current, plastic is far superior to ceramic for high Q, insensitivity to DC bias, loss tangent, tan Delta vs f, piezo sensitivity.

For low current LC tuning in RF , NP0 or P250 was my preference using TDK or Murata.

 

[Spehro Pefhany]

I had always avoided ceramics in S&H circuits due to bad DA memory effects.

If you have high reactive current, plastic is far superior to ceramic for high Q, insensitivity to DC bias, loss tangent, tan Delta vs f, piezo sensitivity.

For low current LC tuning in RF , NP0 or P250 was my preference using TDK or Murata.

You can now get quite high value NP0 caps.. not cheap at the highest
values, but 0.1uF/50V for 37 cents in 100's isn't too terrible for a
all that capacitance squeezed into a 1206 (and 10x better tempco than
a PP cap).

--sp

 

[Baron]

Hi Spehro,

Spehro Pefhany Inscribed thus:

Interesting.. I think they're better again, I've not used them.
Perhaps better than anything else with a reasonable epsilon.

AVX shows them up to 2400pF.

http://www.avx.com/pt.pdf

See the DA curve on the 3rd-last page.

Judging by all the radiation and MIL spec stuff they don't look very
inexpensive.

--sp

Thanks for the link.

The reason I asked is that I have a lot of "Dow Corning" glass
capacitors of various values and have not been able to find any info
about them. Some are quite high capacitance values and close tolerance
ie 1%.

 

[[email protected]]

You can now get quite high value NP0 caps.. not cheap at the highest
values, but 0.1uF/50V for 37 cents in 100's isn't too terrible for a
all that capacitance squeezed into a 1206 (and 10x better tempco than
a PP cap).

I was just looking at .22uF C0Gs at DigiKey. They're not small (1210s
are the smallest) and NOT cheap ($1.34 ea. if you buy a whole reel
-2000 pcs.). Murata doesn't seem to have anything that big.

http://www.digikey.com/product-detail/en/C1210C224J3GACTU/399-11188-2-ND/2216674

 

[miso]

C0G has always been C0G (except when it was called NP0, but you know..).
It's basically straight... titanium dioxide, or strontium titanate or
something? I forget. Once it was discovered, the material properties
remain constant, there's just variations in formulation and manufacturing
process. Most notably the multilayer process.

In my junk box, I have some monster NP0s from '60s TV sets. 100pF is a
one inch disc! Purity and, more importantly, density probably weren't so
great back then, so they had to use excess thickness to guarantee
dielectric strength (they might actually handle 5kV.. who knows?).

MLCCs must be using high density formulations, or finer grained material,
or something, which allows them to make micron thick layers. Then they
just make a lot of layers. Then they make a lot of caps. Millions of
caps. Billions. High speed production does its thing, the market
responds, and prices get where they are now.

Tim

Trying to figure out the cost of external components for a chip design
can be tricky since the volume purchase prices are not published. But
when you do the arm twisting and find the numbers, the pricing of small
ceramic capacitors requires 4 to 5 significant figures if evaluated as a
unit price.

For sample and hold applications, dielectric absorption is the key
parameter.

 

[Robert Baer]

That's why they came up with COG, so we don't have to worry about that
anymore.

Oh...as in gears...

 

[Robert Baer]

Is that Charlie Oscar Golf, or Charlie Zero Golf?

Maters not,it is STILL a gear...

 

[George Herold]

I was just looking at .22uF C0Gs at DigiKey. They're not small (1210s

are the smallest) and NOT cheap ($1.34 ea. if you buy a whole reel

-2000 pcs.). Murata doesn't seem to have anything that big.



http://www.digikey.com/product-detail/en/C1210C224J3GACTU/399-11188-2-ND/2216674

Maybe two 0.1 uF in parallel?
http://www.digikey.com/product-detail/en/CGA5L2C0G1H104J160AA/445-6984-1-ND/2673002
Or is this a 'Joregesque' (sp) type of design that has to fit on a postage stamp?

George H.

 

[[email protected]]

Maybe two 0.1 uF in parallel?
http://www.digikey.com/product-detail/en/CGA5L2C0G1H104J160AA/445-6984-1-ND/2673002
Or is this a 'Joregesque' (sp) type of design that has to fit on a postage stamp?

Mine? Not particularly but eight 1210s would blow the space given
*and* 32 on the board would completely blow the $$ budget. I'll stick
with higher voltage X7Rs and eat the distortion. I was just sorta
curious about what I could do better next time. That ain't it. ;-)

 

[George Herold]

Mine? Not particularly but eight 1210s would blow the space given
*and* 32 on the board would completely blow the $$ budget. I'll stick
with higher voltage X7Rs and eat the distortion. I was just sorta
curious about what I could do better next time. That ain't it. ;-)

If there's vertical room you could piggy back 'em. But I guess not practical for a large number of units. Maybe wait a few years and ceramics will keep getting better? A 1uF (50V) NP0 for a buck or two would be nice.

George h.

 

[[email protected]]

If there's vertical room you could piggy back 'em. But I guess not practical for a large number of units. Maybe wait a few years and ceramics will keep getting better? A 1uF (50V) NP0 for a buck or two would be nice.

Not very production friendly, either. A buck or two???? Yikes!
Pennys hurt.

 

[legg]

NPO ceramic vs polypropylene

There was a little side discussion on JL’s LC tuning thread. When I looked at this ~2 years ago I found that ceramics were a tad better than the ‘good’ poly-pro’s.
(I used them in a low freq LCR circuit.. Q~35 so the cap type didn’t make that much of a difference.)
(The poly pro’s I have are from Panasonic. low dissipation 2% caps. I can dig up a part number if it’s important, I’m not sure they are made anymore.)
So these are both 10nF caps. I first measured them on an SRS720. at 10 and 100kHz.

10kHz 100kHz
type C R D R D
NPO 10.17 0.6 0.0004 0.006 0.00003
poly 10.16 0.23 0.00013 0.086 0.0056

The R/D numbers are so low I don’t really believe them.
So I stuck the C’s in the (original) LCR circuit. So I list peak frequency, Vin and Vout. (As measured on a ‘scope (digital w/average))

type freq. Vin Vout
NPO 6.08kHz 159mV 6.40V
poly 6.13kHz 159mV 6.36V

(again no real difference to talk about.) I pulled out a different L...

type freq. Vin Vout
NPO 157kHz 91.2mV 7.36V
poly 158.5kHz 92.8mV 7.24V

Finally a bit of difference I can believe. Vin dropped because the series LC was loading down the function gen.

So for those who think film are better, you might want to look again at NPO ceramics. They've improved a lot*. Of course this says nothing about the DA of the caps. (How does one go about measuring that?)

George H.

*you need to get the new little ones.. the old larger npo ceramics with the dull brown coating are not nearly as good.

You may be missing the point if tuned circuits are actually the topic.

Polystyrene tempcos are expected to cancel out those of ferrite.
An NPO or other film dielectric couldn't do that, unless paralleled
with a nother part with a suitable tempco. Caps aren't the weak point
in this area.
RL

 

[[email protected]]

The trick of using two in series and biasing the middle will reduce the
distortion there too.

That's an interesting idea, however it costs some seriously quiet
references and only helps 3dB.

 

[Phil Hobbs]

That's an interesting idea, however it costs some seriously quiet
references and only helps 3dB.

Well, in a single-supply circuit, you can probably bias the midpoint at
ground. And the even-order distortion should cancel quite a bit better
than 3 dB, unless the caps are +-100% tolerance.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[George Herold]

On Tue, 19 Nov 2013 08:52:20 -0800 (PST), George Herold

You may be missing the point if tuned circuits are actually the topic.

Polystyrene tempcos are expected to cancel out those of ferrite.
An NPO or other film dielectric couldn't do that, unless paralleled
with a nother part with a suitable tempco. Caps aren't the weak point
in this area.

RL

Interesting, Thanks. I'm doing mostly 'air' core inductors, or caps with opamps. A tuned circuit seemed like the easiest way to compare the dissipation in the caps. (at that frequency)

George H.

 

[[email protected]]

Well, in a single-supply circuit, you can probably bias the midpoint at
ground. And the even-order distortion should cancel quite a bit better
than 3 dB, unless the caps are +-100% tolerance.

I guess I don't see it. The DC bias is even worse, in this case. The
voltage across a single cap is only significant at the corners, often
outside of the interesting band. If the center is grounded, the bias
is not only the average voltage but is asymmetrical.

 

[Phil Hobbs]

I guess I don't see it. The DC bias is even worse, in this case. The
voltage across a single cap is only significant at the corners, often
outside of the interesting band. If the center is grounded, the bias
is not only the average voltage but is asymmetrical.

It depends on the situation. If the midpoint is biased halfway between
the two ends, and the capacitors are identical, all the even-order
distortion products cancel out exactly. Departing from that condition
makes things worse fairly fast, but it won't be as bad as 3 dB unless
the situation is very asymmetrical.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[[email protected]]

It depends on the situation. If the midpoint is biased halfway between
the two ends, and the capacitors are identical, all the even-order
distortion products cancel out exactly. Departing from that condition
makes things worse fairly fast, but it won't be as bad as 3 dB unless
the situation is very asymmetrical.

But that's exactly what you have with a single supply and a grounded
bias in the middle of the capacitors. With a single supply, the
capacitor's midpoint would have to be at Vcc/2 (or some such). For
capacitors between supply domains, there would have to be an
additional vref, half way between those references.

 

[[email protected]]

Polystyrene tempcos are expected to cancel out those of ferrite.
An NPO or other film dielectric couldn't do that, unless paralleled
with a nother part with a suitable tempco. Caps aren't the weak point
in this area.

I tried *hard* to find a source for a polystyrene cap a couple
years ago, for use in a picoamp integrate-and-dump.

The answer came back "No mas." Several sources said the
film was no longer being made; and only a few cap-makers were
still producing from life-time buys of the raw material.
Digikey shows none, today.

That led to comparing dielectrics. It was tempting to try
NP0. The dielectric absorption and leakage looked low, but
there wasn't time. We found a PS supplier, ultimately, and
a polypropylene part for back up.

Cheers,
James Arthur