The 10 kV suggestion sounds reasonable. Peter brings up a very important
point, however. Air pockets within your dielectric film will eventually EAT
the dielectric. More accurately, ionized air (ozone?) will eat it. I have
experienced this several times with my designs, and it seems to always slip
through. Failure is not immediate, but comes weeks, months later, when the
molecules of polymer are finally degraded to something weaker. This is not
just a high power RF concern, even a few kV of RF or DC will cause problems
if these points are ignored.
Dielectric theory is as simple as series capacitors in this 'flat plate'
example. Imagine a film of 0.01 inch dielectric with K=2 (rounded down from
2.1 for this simple explanation). The gradient is then 1 Megavolt/inch for
a 10 kV applied voltage across this film. It's the voltage divided by the
gap spacing, in this case, since the dielectric is homogenous. Readers in
the EC will have to convert to SI units yourselves. Now, imagine a small
air void in one place, also about 0.01 inch thick, which has K = 1.0. If
those were in series, and the applied voltage across the sandwich were 10
KV DC, then the gradient in each material is figured for each dielectric
constant. This has nothing to do with the loss tangent or dissipation
factor, strickly DC dielectric stuff here. It also applies for RF or AC,
the capacitors are capacitive reactances for AC.
The two dielectrics act as series capacitors, one with twice the capacity
of the other since K = 2 there. Then the voltage division is 1/3 in the
film and 2/3 in the air space. So 3.33 kV in the film, and 6.67 kV in the
air. The gradients are 333 kV/in and 667 kV/in respectively. Air will
breakdown quickly at this level. If the pocket is trapped, then it will be
only a matter of time.
Grease or potting the entire assembly in epoxy, with a vacuum exhaust, are
workable solutions. You have to get the air out with pressure, replacement
with silicone grease, etc. I gutted an old Amana (1 KW) microwave oven
once, and noticed that the cathode leads into the magnetron box were
bypassed with some polyester film (Mylar) and round aluminum disks that
were radiused. There was a lot of grease between the plate and the film,
and the film and the metal chassis. Nowadays, magnetrons have built in
ceramic capacitors for bypassing on the cathode leads. If you open one, you
will see it. Some even have RFC in series with the heater leads.
This brings up another more obsure failure mechanism in film dielectrics.
Around the edge of the electrodes, the plate you use for your capacitor,
there is an enhanced field. The E field lines are bunched, and the gradient
is enhanced, expecially if it is a sharp edge. Aha! The solution to this,
you might think, is to radius the edges. Guess what this does? It creates a
boundary where the electrode surface tapers away from the film, and in that
tiny first space as the taper begins, you have that same air void problem.
Except that here the air space is ever so tiny, that the gradient can be
In high voltage design, this 'triple point' effect is a problem, where the
ground leaves a HV cable, where connectors terminate, where capacitors are
designed. It's why terminating AC HV powerlines that are shielded require
those fancy kits from 3M and Raychem. It's also why manufacturers squirt
some silicone grease into that vold around the electrode edges. Silicone
RTV goop also works, but one must be careful to avoid bubbles.
Continental Electronics in Dallas makes a plate blocking capacitor for one
of our transmitters, that runs 22 KV plate voltage. Its a cylindrical
capacitor, like a barrel, that connects to the plate of a grounded-grid
Burle 7835 super-power triode. Heat shrinking Teflon film is used, and
around the edges RTV silastic has been applied (impregnated is more like
By the way, you can also use Mylar, or even Kapton tape or film. If it is
thin film, and the area is large enough, the capacitive reactance can be
kept small, and the RF voltage across it is low and safe. And keep all
those little metal filings out of the space (clean and deburr holes) when
you clamp down on the assembly. Teflon will cold flow and the capacitor may
become loose over time. The other materials are better in this respect.
Pyralux* is DuPont's tradename for copper-clad Kapton H film. It's the
stuff used in high performance flexable printed circuits. Using it for the
capacitor will guarantee the value of the cap is consistent, and there are
no voids underneath. But it has a real problem with the sharp edge of the
copper plating. So it is best used in things like the filament bypass
capacitors, Eimac and Thomson use it in a number of their newer sockets
with integral cathode bypassing. Such as the 4CX3500, 5000, 7500, 20000,
sockets (not the SK300).
>Does anyone know how much DC voltage
>0.25mm (0.01 inch) teflon sheet can handle
>used as a dielectricum in a decoupling cap (made of brass)
>Any help appreciated!
>Date: Mon, 15 Feb 1999 11:36:33 -0000
>From: Peter Chadwick <Peter_Chadwick@mitel.com>
>Subject: RE: [AMPS] isolation question
> >Does anyone know how much DC voltage
> >0.25mm (0.01 inch) teflon sheet can handle
>The ITT Reference Data for Radio Engineers lists teflon as 1000 to 2000
>volts per thosandth of an inch for thickesses between .005 and .012 inches.
>This suggests a minimum of 10kV for your thickness. However, at that level,
>you would need to watch out for air pockets which could ionise and degrade
>the insulation. I don't know if a thin film of silicon grease would help -
>it would depend on what the losses in the grease are at V/U HF.
FAQ on WWW: http://www.contesting.com/ampfaq.html
Administrative requests: amps-REQUEST@contesting.com