COG caps and Dielectric Absorption (DA)

It's the dielectric that matters, not the construction. Classically, COG ceramics had excellent properties, but got too big to be useful at much over 100pF. Dielectrics offering a higher dielectric constant tended to have compensating defects.

COG just means a zero temperature coefficient with a tolerance G of +/-30ppm/K

suggests that this might mean that (ZrSn)TiO4 could be being used as the dielectric.

Hi Bill, you haven't been keeping up with advances in ceramic caps. (Troll digikey and see what's there.) Here's one NPO/COG, 0.1 uF, 50 V for ~$0.30

I've been using NPO's to replace films in high Q circuits. But the older NPO ceramics have much worse ESR.. not so good.

Did you look at fig 5.4B (page 301 AoE3) There was a similar (the same?) plot from B Pease. (also with some magic low DA ceramics.)

George H.

Me too. Surface-mount film caps tend to fall apart, and get wrecked by a water-wash process. The big NPOs do have far less DI than other ceramics, and are better than some films.

I was using 0.33uF COG's in the early '80's... as big as the tip of your pinky, but good quality ;-) ...Jim Thompson

G ceramics had excellent properties, but got too big to be useful at much o ver 100pF. Dielectrics offering a higher dielectric constant tended to have compensating defects.

e dielectric.

/3845702

It may be COG around 25C but the data sheet has the capacitance moving fast er with temperature once you get away from there, and it's not spelled out all that clearly.

Brave of you.

NPO isn't COG.

I did. It's utterly useless, because "brand X" and "brand y" haven't been i dentified.

If somebody could magically tell us where to buy them, this might be a usef ul observation.

Hand-waving claims that certain parts are better than others are less than helpful. You need a part number you can stick on a data sheet that will alw ays call up the same part - the buyers always want to buy something almost the same that the rep swears is just as good ...

I'm guess you paid more than $1 for one.

George H.

I don't remember now what they cost, but they were going into active filters for the "muff" modem that was in the GenRad (nee OmniComp) portable PCB tester... volume was low, tester sale price was exorbitant >:-} so price was no object. ...Jim Thompson

COG ceramics had excellent properties, but got too big to be useful at much over 100pF. Dielectrics offering a higher dielectric constant tended to ha ve compensating defects.

/-30ppm/K

the dielectric.

ND/3845702

ster with temperature once you get away from there, and it's not spelled ou t all that clearly.

Courage had nothing to do with it. All the good film caps were going away. (EOL) I assume that the reason why is that everyone else had already switch ed to NPO ceramics.

Huh? Well for digikey and me they are the same...check out the Temp Co. column.

identified.

eful observation.

n helpful. You need a part number you can stick on a data sheet that will a lways call up the same part - the buyers always want to buy something almos t the same that the rep swears is just as good ...

Hand-waving? That is data either from B. Pease or H & H. Perhaps Win will chime in and tell us (me).

George H.

Don't waste your time feeding the Sloman troll. Have you gained any insight from any of his posts? No! And his oscillator still doesn't work >:-} ...Jim Thompson

The "terrible loss" i believe, is on purpose, to increase filering.

There's an easy way around dissipation factor that I discovered in the early '70's when I was designing filters at low audio frequencies...

...Jim Thompson

All the manufacturers think it is.

pdf

y, COG ceramics had excellent properties, but got too big to be useful at m uch over 100pF. Dielectrics offering a higher dielectric constant tended to have compensating defects.

f +/-30ppm/K

as the dielectric.

-1-ND/3845702

faster with temperature once you get away from there, and it's not spelled out all that clearly.

ay.

tched

een identified.

useful observation.

than helpful. You need a part number you can stick on a data sheet that wil l always call up the same part - the buyers always want to buy something al most the same that the rep swears is just as good ...

The simulations work fine. Building a real example is going to be a pest. I 've just installed SuSE 13.2 on a second hard disk, and I'm trying to downl oad Kicad. The circuit is simple enough that a PCB layout is going to be fa ster (and a lot more reliable) than hand-wiring a board.

Yast2 in SuSE13.2 won't open repositories that Firefox can open in the same session. The web claims that disabling IPv6 can solve the problem - which it doesn't - and finding the right mirror can solve the problem - which it does until you try and download Kicad (which looked as if it was going to w ork fine until Yast2 started finding its dependencies).

Typical bloody Linux. It will probably take a couple more days to find a wo rk-around - all advice gratefully received.

DI? I presume he meant dielectric absorbtion - maybe he sees it as Dielectric Injection, since it is charge that slowly gets into the dielectric, and comes out again equally slowly.

I'm reminded of Warburg impedance, which you get with current flow through conducting liquids, where the conducting ions don't move all that fast.

on.

s.pdf

lly, COG ceramics had excellent properties, but got too big to be useful at much over 100pF. Dielectrics offering a higher dielectric constant tended to have compensating defects.

of +/-30ppm/K

d as the dielectric.

ng faster with temperature once you get away from there, and it's not spell ed out all that clearly.

away.

witched

o.

been identified.

a useful observation.

s than helpful. You need a part number you can stick on a data sheet that w ill always call up the same part - the buyers always want to buy something almost the same that the rep swears is just as good ...

I've just installed SuSE 13.2 on a second hard disk, and I'm trying to dow nload Kicad. The circuit is simple enough that a PCB layout is going to be faster (and a lot more reliable) than hand-wiring a board.

me session. The web claims that disabling IPv6 can solve the problem - whic h it doesn't - and finding the right mirror can solve the problem - which i t does until you try and download Kicad (which looked as if it was going to work fine until Yast2 started finding its dependencies).

work-around - all advice gratefully received.

You have been talking about this oscillator for years now and claim it's qu ite simple. Why don't you spend one day building it instead of continuously talking about it? Why is it a pest to build one?

Cheers

Klaus

e:

te:

flon.

ors.pdf

cally, COG ceramics had excellent properties, but got too big to be useful at much over 100pF. Dielectrics offering a higher dielectric constant tende d to have compensating defects.

G of +/-30ppm/K

sed as the dielectric.

ps.

ving faster with temperature once you get away from there, and it's not spe lled out all that clearly.

g away.

switched

Co.

't been identified.

be a useful observation.

ess than helpful. You need a part number you can stick on a data sheet that will always call up the same part - the buyers always want to buy somethin g almost the same that the rep swears is just as good ...

t. I've just installed SuSE 13.2 on a second hard disk, and I'm trying to d ownload Kicad. The circuit is simple enough that a PCB layout is going to b e faster (and a lot more reliable) than hand-wiring a board.

same session. The web claims that disabling IPv6 can solve the problem - wh ich it doesn't - and finding the right mirror can solve the problem - which it does until you try and download Kicad (which looked as if it was going to work fine until Yast2 started finding its dependencies).

a work-around - all advice gratefully received.

quite simple. Why don't you spend one day building it instead of continuous ly talking about it? Why is it a pest to build one?

It's not simple enough that you could hand-wire it in a day - my guess is s omething like a week or more (bearing in mind the surface-mount components) .

I read it as DF, dissipation factor, which is what you care about for a high Q.

It's a shame to have a nice inductor and then have the Q limited by the cap. I've never tried any high Q gyrator type circuits. But I'd guess the cap ESR would be a limit also.

George H.

In most resonant circuits, the L is worse than the C. You can parallel small Cs to reduce ESR.

Above about 400 MHz or so, coaxial ceramic resonators rock.

Here's a q-meter, just a series resonant circuit snooped by a sampling scope:

On Wed, 12 Aug 2015 06:55:34 -0700, John Larkin Gave us:

snip

Notice the coax usage on these units from russia: