On 11/1/2015 12:05 PM, wrote:
rickman wrote:
On 10/31/2015 7:39 PM, wrote:
rickman wrote:
On 10/31/2015 6:01 AM, Jeff wrote:

Glass is used as a dielectric in high quality low loss RF capacitors so
I suspect that it would be usable in a home-made one.

Doesn't necessarily follow. The loss tangent of glass is low to very
low so it won't heat up much in use. But the important part is the
change in Er with temperature as I explain. In fixed value caps changes
in capacitance of a few percent are usually not a problem. But in this
application tuning of the circuit may be very critical and require a
much higher degree of stability.

I am also looking at alumina ceramics. The properties vary with
composition, but there are composites with very high stability numbers.
They usually are in a materials data sheet rather than in a product
offered for sale. Seems a lot of ceramics are custom items.


I think the change of Er with temperature is the least of your worries
when talking about a tuning capacitor for a magnetic loop. The change of
capacitance due to mechanical changes in the plates is likely to be at
least an order of magnitude greater than the dielectric changes.

I wonder about that. I know metals have a high tempco of expansion and
this will make changes in the capacitance. I haven't analyzed it to see
how significant that would be. So let me give it a try.

Expansion of the air gapped metal capacitor will have two opposing
effects. Enlargement of the plate surface area will increase the area
and so increase capacitance by the square of the tempco of the metal.
Enlargement of the spacing will decrease capacitance directly by the
tempco of the metal assuming the spacers are the same material. The net
effect will be to increase the capacitance in direct proportion to the
tempco of the metal.

Using a dielectric would reduce the effect of the plate spacing on
capacitance to a very small value since most of the capacitance will be
due to the material and much less to any air gap remaining. So it would
return the overall effect on the capacitance to the square of the
expansion tempco. Does that sound right?

In other words, it would be good if the dielectric had an effect that
was opposite to the effect of the metal tempco. I'll meed to consider
that as I search for materials.

An aluminum plate 6 inches square at 75 F heated to 200 F changes
dimensions by 0.0092 inches.

For plate glass the change is 0.0031 inches.

I will leave it to you to calculate how much that will change capacitance.


http://www.engineeringtoolbox.com/li...on-d_1379.html

Did you have a point?

It should be obvious, but since it is not, any change in capacitance due
to thermal expansion is going to be miniscule for a capacitor large
enough to withstand kilovolts.

I don't know what you consider to be "miniscule". I also don't see how
the voltage matters. I believe I have already posted that without
considering fringe effects, but only the first order effects of plate
area and spacing, the end result is a linear change in capacitance with
temperature according to the temperature coefficient. When I run the
numbers I get around 370 PPM for a 30°F rise with aluminum or closer to
277 PPM for the same rise with copper (I haven't seen a copper tuning
capacitor though).

It appears the temperature effect on the loop inductance is larger at
413 PPM for aluminum and 309 PPM for copper. These numbers may not be
spot on because I used a handy calculator for the inductance which may
not have considered the diameter of the conductor, most loops are wide
material.

These two effects augment to feed the equation for resonant frequency
which uses the square root of the product resulting in 391 PPM for an
all aluminum system or 340 PPM for a system with a copper loop and
aluminum cap. This is enough to impact the tuning of a high Q antenna
to give more than a 3 dB drop over the course of a day. Over the course
of a year some locations will see a change of 50°C or a three fold
greater change. That would easily be enough to disrupt an auto-tuner
and require recalibration. Do you get significantly different numbers?

If a dielectric were chosen with a slight negative temperature
coefficient, it could offset the natural drift of the antenna tuning
bringing it closer to zero.

--

Rick