High dielectric constan caps, some weirdness

[Joerg]

Folks,

Check out this datasheet which is for X7R caps that are called "high
dielectric constant". I need a 22uF/16V to store nearly all its charge
for about 40sec, which is afterwards used up by some electronics.

To be exact, this cap:

http://search.digikey.com/us/en/products/GCM32ER71C226KE19L/490-5242-6-ND/1979255

No (!) datasheet is provided so I searched Murata and found it:

http://www.murata.com/products/catalog/pdf/c03e.pdf

The 1st weirdness is on the bottom of page 17, insulation resistance
"More than 100,000Mohm or 1000ohms * F, whichever is smaller". Now a
22uF ceramic can hardly be 1000ohms times 0.000022, right? Or did some
characters not render correctly in Foxit Reader?

The 2nd weirdness is the graph on the top right of page 25. These are
supposed to be X7R caps and the graph looks more like Z5U.

Did anyone use these caps and could shed some light on this?

 

[Joerg]

I read that as R > 1000 ohm farads/0.000022 farads = 45.4 megohms. IOW
they guarantee that the self-discharge time constant is greater than
1000 seconds unless the capacitance is less than 0.01 uF.

Hmm, interesting, Foxit reader shows a dot instead of a slash.

Blech. X7R should be more like -10% at rated voltage.

That's why this graph surprised me. I've sent an inquiry but not sure
whether it gets answered in due time and I am not related to the emporor
or any of that

[...]

 

[tm]

Folks,

Check out this datasheet which is for X7R caps that are called "high
dielectric constant". I need a 22uF/16V to store nearly all its charge
for about 40sec, which is afterwards used up by some electronics.

To be exact, this cap:

http://search.digikey.com/us/en/products/GCM32ER71C226KE19L/490-5242-6-ND/1979255

No (!) datasheet is provided so I searched Murata and found it:

http://www.murata.com/products/catalog/pdf/c03e.pdf

The 1st weirdness is on the bottom of page 17, insulation resistance
"More than 100,000Mohm or 1000ohms * F, whichever is smaller". Now a
22uF ceramic can hardly be 1000ohms times 0.000022, right? Or did some
characters not render correctly in Foxit Reader?

The 2nd weirdness is the graph on the top right of page 25. These are
supposed to be X7R caps and the graph looks more like Z5U.

Did anyone use these caps and could shed some light on this?

--

Sure it's not 1000/C in Farads? That would be about 50 megs.

 

[tm]

An ohm-farad is the same as a second. They're specifying the
self-discharge time constant.

Cheers

Phil Hobbs
--

Potato - Potatoe!~

Still 50 megs. Well, 45.45...

Thanks Dr. Hobbs.

 

[Joerg]

X7R capacitance does not fall off 65% @ 5V regardless of rated
voltage. Someone slipped in the wrong graph, a 6.3V Z5U or some such.

Ok, but in a datasheet that is not supposed to happen.

 

[Joerg]

I don't think you'll have any problems in 40 seconds.

I don't think so either. But it would be nice to know to be able to put
some rationale about this function into the module spec. It's kind of an
unusual circuit that keeps doing some stuff while there no longer is any
power available.

Do you have room for something like a polymer aluminum cap?

Not really. I've got almost 250 mils of height to work with but it's a
high-temp environment. Ideally we don't want to have any electrolytics,
and most definitely not tantalum. Normally an X7R cap should do this
nicely but I can't put one in there when the datasheet says otherwise,
even in case the information in there should be wrong.

What irks me is the fact that Digikey and others are starting to have
bland links, instead of a datasheet all you get is meaningless sutff
like photos. I shouldn't have to scour the web for the datasheet.

 

[[email protected]]

This is typical: do a parts search on the Digikey site and get, say,
150 parts. Now qualify it by "stocked." It may drop to 30 parts.

Mouser is about as bad lately. What's the point of listing parts that
aren't stocked and apparently can't be ordered?

Just because they're not stocked, doesn't mean the parts can't be ordered. I
don't know why you'd order non-stocked parts from DigiKey, but it can be done.

 

[Tim Williams]

Just because they're not stocked, doesn't mean the parts can't be ordered.
I
don't know why you'd order non-stocked parts from DigiKey, but it can be
done.

Ok, so why buy anything if you simply can't have it same-day? "Non stocked"
just means you can't get it instantly. It then becomes the same purchase as
a purchase from literally anyone else on Earth: submit a quote with price
and ETA, decide if you want it or not, then sit and wait.

Indeed, Digikey is even better than the usual process, because they specify
price on almost all "non stocked" items, and you can usually establish a
rough ETA by adding to your cart and seeing what it says.

My one "non stock" story, I ordered a Vishay/IR diode which was out of
stock; ETA was something like 6 weeks, actually took something like 8 or 10,
give or take a call about it. Without a deadline on that particular
project, I was fine waiting.

At work, every day, we buy transformers and capacitors that physically don't
exist before they are ordered. But they arrive as scheduled. Why should I
be afraid that something doesn't physically exist before buying it?*

*And yes, I fully realize such a statement demands tongue at least half in
cheek. I also appreciate the reason you don't normally want to order
something non-stocked anyway, so don't reply like I'm trolling you about it.

Tim

 

[Joerg]

Hmm, that's true. Maybe an alien life-form from Earth's future
slipped it in there, as some sort of cryptic message/warning.

Yesterday they actually confirmed that the graph is correct (!). So, I
went with another manufacturer, I don't have much time left to dwell on
such things for this design project.

What's the stock market been doing? ;-)

Slid down 63 points yesterday but I stay out of it for the most part
right now. Until I know what shakes out in Europe and (hopefully soon)
get a feel on how the elections will go.

 

[[email protected]]

Ok, so why buy anything if you simply can't have it same-day? "Non stocked"
just means you can't get it instantly. It then becomes the same purchase as
a purchase from literally anyone else on Earth: submit a quote with price
and ETA, decide if you want it or not, then sit and wait.

Right. I won't buy anything from DigiKey that I can't have immediately.
Instant delivery is the only reason I'd ever buy from DigiKey. Like many
here, I do use DigiKey as a gauge of availability.

Indeed, Digikey is even better than the usual process, because they specify
price on almost all "non stocked" items, and you can usually establish a
rough ETA by adding to your cart and seeing what it says.

If you like spending too much of your boss' money, sure. ;-)

My one "non stock" story, I ordered a Vishay/IR diode which was out of
stock; ETA was something like 6 weeks, actually took something like 8 or 10,
give or take a call about it. Without a deadline on that particular
project, I was fine waiting.

At work, every day, we buy transformers and capacitors that physically don't
exist before they are ordered. But they arrive as scheduled. Why should I
be afraid that something doesn't physically exist before buying it?*

Because the only reason I buy from DigiKey, in the first place, is because I
need it *now*. If I can wait, I don't need DigiKey.

*And yes, I fully realize such a statement demands tongue at least half in
cheek. I also appreciate the reason you don't normally want to order
something non-stocked anyway, so don't reply like I'm trolling you about it.

Everyone's needs are different, but if an item isn't stocked at DigiKey, the
chances are good that it's never stocked anywhere. Custom parts are a whole
different kettle.

 

[josephkk]

Folks,

Check out this datasheet which is for X7R caps that are called "high
dielectric constant". I need a 22uF/16V to store nearly all its charge
for about 40sec, which is afterwards used up by some electronics.

To be exact, this cap:

http://search.digikey.com/us/en/products/GCM32ER71C226KE19L/490-5242-6-ND/1979255

No (!) datasheet is provided so I searched Murata and found it:

http://www.murata.com/products/catalog/pdf/c03e.pdf

The 1st weirdness is on the bottom of page 17, insulation resistance
"More than 100,000Mohm or 1000ohms * F, whichever is smaller". Now a
22uF ceramic can hardly be 1000ohms times 0.000022, right? Or did some
characters not render correctly in Foxit Reader?

The 2nd weirdness is the graph on the top right of page 25. These are
supposed to be X7R caps and the graph looks more like Z5U.

Did anyone use these caps and could shed some light on this?

I can tell you right up front that the ceramics are too leaky for that.
Maybe polymer tantalums will work. Otherwise,you need the space for
polymer caps.

Test if you don't believe me. I so would like to have current test data.

?-)

 

[josephkk]

Ok, but in a datasheet that is not supposed to happen.

That is such a 1970s attitude. According to some, passe by 1990s. YMMV

?-(

 

[Joerg]

That is such a 1970s attitude. According to some, passe by 1990s. YMMV

For me that's simple. There are only three options and it's up to the
manufacturer:

a. It's in the datasheet. That would be the proper solution.

b. I pester the heck out of the support stuff. Just done that in order
to complete an aerospace project. This can easily swallow more than one
manhour at the manufacturer to dig out all the stuff that's missing in
the datasheet.

c. Worst solution, rarely happens but does: Manufacturer responds that I
won't get more data. Then I walk. The old American saying, "If you don't
take care of your customer, somebody else will".

 

[Joerg]

I can tell you right up front that the ceramics are too leaky for that.
Maybe polymer tantalums will work. Otherwise,you need the space for
polymer caps.

Au contraire. One of the mfgs did suggest a tantalum series where we'd
have gotten the same capacitance in the same size but without
capacitance drop at higher bias. One peek at the leakage data thwarted
that, IIRC it was 3uA. Now that is guaranteed not to work, plus I do not
like to use tantalums.

Test if you don't believe me. I so would like to have current test data.

In a proper datasheet they state insulation resistance. Now how could a
22uF cap possibly not work in this application if it's spec'd in the
Gohm range?

http://www.niccomp.com/Catalog/nmc2.pdf

So, since I don't believe you on this one I just took a 1uF X7R out of
the bin, charged it to 10V, disconnected it, left it on the bench,
waited 40 seconds, measured -> 9.9V. Good enough, I'd say

 

[tm]

Au contraire. One of the mfgs did suggest a tantalum series where we'd
have gotten the same capacitance in the same size but without
capacitance drop at higher bias. One peek at the leakage data thwarted
that, IIRC it was 3uA. Now that is guaranteed not to work, plus I do not
like to use tantalums.



In a proper datasheet they state insulation resistance. Now how could a
22uF cap possibly not work in this application if it's spec'd in the
Gohm range?

http://www.niccomp.com/Catalog/nmc2.pdf

So, since I don't believe you on this one I just took a 1uF X7R out of
the bin, charged it to 10V, disconnected it, left it on the bench,
waited 40 seconds, measured -> 9.9V. Good enough, I'd say

--

Did you check it at temperature?

 

[Joerg]

I miss the days of having our own "Purchasing Dept" WITH properly
WORDED purchase contracts:
"Shall meet the spec, else supplier is liable...
then there's a long diatribe of reimbursable damages!

Then you'd probably all lose your jobs because none of the big guys will
sign off on that

 

[Joerg]

Did you check it at temperature?

No, just at 75F. I could pre-bake it but got too much work for that
right now. IIRC the minimum insulation resistance drops by a factor of
5-10 when you go to 200F. I'd never run anything that hot and even that
kind of drop would not be a concern.

 

[Cydrome Leader]

An ohm-farad is the same as a second. They're specifying the

haha. I should used that instead of normal time units.