Lucifer17:

I don't have a direct suggestion, but you may want to poke around this site, it is quite comprehensive list of material properties, including dielectric constants.

http://www.matweb.com/search/SearchProperty.asp?e=1

-Pete

Have no experience with these, but they are listed in following brochure. This brochure is old (about 1990) and these materials may not be availiable.

Dow Corning, Materials for High Technology Applications,
FLUOROGEL, Dielectric Constant 6.88
SILICONE ELASTOMER Q3-6605, Dielectric Constant 4.78

You might also try getting powdered ceramic and mixing it with something flexible like silicone sealant. Getting higher dielectric constant is much easier than getting lower dielectric constant. If half the mixed material is ceramic (having a dielectric constant of say 10) then the overall dielectric constant will be greater than 5.

Try filetr paper saturated with liquid e.g. water (DC=80)

<nbucska@pcperipherals.com>

Logbook:
Wrong.

If the diel. constants are e1 and e2, the 50% mixture will
have  e=(2*e1*e2)/(e1+e2) i.e. e.g. e1=2 and e2=100 will
result in e=3.92.  If e2 is infinity, e=2*e1.
Want me to E-mail proof?

Lucy:
Ba titanate is >> 1000 but not flexible.
What is the frequency range and what do you need?
Charge storage or time delay ?

If it is a transmission line, you can get the same result
by adding studs (capacitive loading) or inductors
( loops or thin sections ).

<nbucska@pcperipherals.com>

Rogers Corp has RO3010 that's a fairly flexible sheet if you only bend it around a cylinder. I've bent the 0.060" material around a 1" radius pretty easily. Dielectric constant is 10 and it's fairly cheap.
Emerson and Cuming has premixed ceramic powder dielectric in values from 3 to 15, maybe their pure ceramic powder is for sale and the earlier suggestion of embedding it in silicone would work. This E&C product is made to fill "complex cavities" so you don't have to machine anything in your prototyping stage. I've used the dielectric value of 6 to load antennas for imaging of concrete whose value is 5 to 7 in the 2-6 ghz range, it works pretty well.
kch.

nbucska

I read your posting with the sinking feeling that I had messed up and couldn’t edit my response. However, on further consideration I don’t think I am as wrong as I had at first thought.

My original thinking was based on two systems. The first is the coplanar waveguide, where the dielectric constant is half way between air and the dielectric. The second is iron dust magnetic cores which are iron particles in an insulating binder. These achieve a relative permeability well above 2.

I thought further about your comment. The coplanar waveguide has two capacitances in parallel so the partial capacitances add. If the capacitances were in series then the smaller one would dominate. Thus if the system had a cylindrical geometry and concentric cylindrical shells of the two dielectrics the overall dielectric constant would tend to a value closer to the smaller one.

But the system we are talking about has a well dispersed good dielectric mixed with the less good dielectric. Thinking in terms of spheres of the good dielectric, at a 50% mixture the spheres would be quite close on average. I therefore think the overall dielectric constant would be considerably better than the pessimistic value you suggested.

I would be pleased to see your argument posted here so we can think about it some more.

Log:
Lucy wants hi DC-- i.e. small % of binder, large % of HiDC.
It means that you can approximate it by assuming that the
E-field is normal to the electrodes.

A 50% e1=2 and 50% e2=10 can be modeled as a cap with half
of the gap  filled  with e1 and the rest with e2.

If the air gap gives C capacitance, the e1 will be 2*e1*C
the HiDC half 2*e2*C .
Q.E.D.

If you want to model it, put insulating or metal particles in an electrolite and measure the conductance which is
analogous with the capacitance. ( The capacitance
is proportional with the [reactive] conductance ]

<nbucska@pcperipherals.com>

I know of some long range radars that used domes of graded artificial dielectric. These incorporated metalic spheres on varing concentrations in a styrofoam to make a lens.

Nbucska,
    I am afraid I can’t agree with your model. I have a little picture which shows a parallel plate capacitor with about 60% high dielectric material as parallel plates. The other half shows the dispersed high dielectric, again with about 60% high dielectric.

http://www.logbook.freeserve.co.uk/image/dielectric.gif>

The cylinders (this is a 2D model) nearly touch so the gaps are small and the field is much more like a parallel connection rather than a series connection. Modelling it as parallel plates would seem to be inappropriate.

I have written my own 2D field solver (commercially available) so I tried your idea of putting insulating cylinders in the field. The result was a 4:1 Increase in the resistance for perhaps 60% of insulating cylinders. As I said, the evidence for this type of effect is seen for RF iron dust cores.

I hope you find this argument convincing.

Lucy:
Still waiting for answer.
Log:
No.

<nbucska@pcperipherals.com>

Log:
If you want a better model,try :
http://www.korozja.pl/2_01_01.pdf or search <www.dogpile.com>

Lucy:
What is your frequency range? Do you need energy
storage, time delay or Z0 matching?

<nbucska@pcperipherals.com>

Nbucska,
Thank you for the reference. I don’t think this paper is relevant to the question at hand, however, since it is talking about surface effects, whereas we are talking about effects in the bulk material. With a surface effect, such as on the surface of a cylinder, I would agree that the capacitances are in series.

I would think that the finite element analysis results which I mentioned would have been the conclusive proof. It is hard to understand what objection there could be to this sort of numerical integration. If I were to mix iron filings with oil to get a ferro-fluid with high permeability then the result might be contentious on the basis of the particles potentially coagulating into chains. The FEA avoids this sort of experimental/practical uncertainty.

Log:
I wouldn't use magnetic model -- it is non-linear and
you can't keep particle edges out of saturation.

<nbucska@pcperipherals.com>

Nbucska,
Yes, that is a good point … which we can both agree on.

Lucy
I found a source for e=40 composite:
http://www.tplinc.com/nano-composite_dielectrics.htm

Log:
Search <www.dogpile.com> for "composite dielectric" --
there are many ref-s.

<nbucska@pcperipherals.com>

To all:
Get USP 6,608,760 from <www.uspto.gov>. It has litt.reference and zillions of patents.

<nbucska@pcperipherals.com>