[Amps] HV Feed Through Capacitor for UHF Amplifier
jtml at losalamos.com
Sun Feb 12 23:29:10 PST 2012
A reply follows, I apologize for the length. I have worked on this problem a lot over the years and its a tricky one to solve at high power coincident with high frequencies:
> From: Larry<larry at w7iuv.com>
> Subject: [Amps] More HV feed through capacitor
> To: amps at contesting.com
> FWIW here is an "update on my feed through capacitor problem:
> The amp in question is a GS-31B based on one of the YU1AW designs
shown on the ND2X web site. This design specifies a anode choke bypass
> capacitor constructed from FR4 PCB material...
I had similar problems with planar HV bypass capacitors made from laminated material. Many years ago, I stripped an ancient Amana Radarange microwave oven. It was one of the earliest, having a dial timer. The maggie had a little box around it with both heater leads coming through little feedthru caps made from machined aluminum disks and PET film, along with a Teflon insert to center the bolts through holes. That stimulated me to design HV feedthrough capacitors for broadcast FM amplifiers that way. I used Copper-clad Kapton polyimide film, called Pyralux, made by Dupont. The capacitance was very good for VHF and the self resonance was way up there. It was designed as a Pi filter, with the bolt through the middle being the L. I didn't realize how strong an effect sharp edges had with HV. The field enhancement (only 4100 VDC) slowly ate the edge of the Kapton on the inside capacitor, from ionization there. After over a year in the field, they began coming back from customers failed.
You could see a ring around the metal edge where the Kapton was 'etched'. Lesson learned - don't have sharp edges on the electrodes, even with wide spacing of nearly an inch. HV will do damage there, over time. It is amazing how far it can track over a surface, given some ionization from partial discharge (corona).
Using FR4 at 430 MHz is asking for trouble too. Even though the capacitor is supposed to be at a voltage node for the fundamental freq, based on the tank circuit design, it might be an antinode for the even harmonics. Not only is FR4/G10 a poor dielectric (loss tangent is marginal due to the epoxy resin) but it is also capable of thermal runaway as it heats and the loss worsens. Also it may absorb moisture. All bad things for a dielectric under HV stress.
As stated by another commenter, making it with Teflon-filled PCB material is probably better. Pure Teflon is difficult as it has low dielectric constant so you'd need a thinner piece or more area to get the same capacitance. Some of the standard Rogers microwave PCB materials with K of 6-10 would be better. Still, you run risk of flashing around the edges, if not right away, a little while later.
You are moving in the right direction, in my opinion, by putting a small choke in series, into a 'doghouse' as you called it. The choke should be perpendicular to the transmission line circuit, with as few turns as possible, perhaps just a few turns about a half-inch diameter, and inch long. Just enough to give some inductive reactance in series, but not dangerous from too much stray cap to cause it to resonate at the fundamental or second harmonic. If you can, measure it with a network analyzer, antenna analyzer, or other UHF instrument. At the cold end, put your capacitor to ground. Get rid of sharp edged copper and use radiused plates or even a radiused cylinder clamped inside another cylinder (see below). Where there is a radiused edge of metal, fill that space with silicone RTV, or better yet, 2 part silastic potting like Dow Sylgard 184 or GE/Momentum RTV12. These are messy, but very good materials with high power RF - improvements over ancient Gylptol paint. By sealing the
edges, you raise the dielectric constant over air, and lower the RF voltage across the tiny air gap where the metal rolls away from the film. [Look up "triple point" and "dielectric flashover" on Google to see explanations.]
Use Kapton, PET (Mylar) or Teflon film, a few mils thick will do. Its better to use a few layers of thinner material than one fatter sheet, although at amateur radio voltages it might be OK. Kapton only comes up to 5 mils (0.005) thick anyway. You'll want less than a nanofarad to make a decent bypass at UHF this way.
I have been relearning this in the past 2 years, as I had to build a big HV feedthru cap for my 200 MHz cavity amplifier at work. Nothing was commercially available. I am passing 125 amps of plate current through it, at 28 kV DC. I borrowed an idea from the bigger Broadcast Electronics FM transmitters that I remember. I made an aluminum cylinder and then a matching aluminum slug that fits into the center with a slight interference fit. By wrapping the slug with Kapton, several layers, then heating the outer cylinder, I pressed the center slug into place, and it sealed tight when it cooled off. Then I potted the edges. I did the potting under vacuum to ensure that there are no air pockets that will ionize and cause failure of the Kapton. It has been working for a year now, fingers crossed.....
I hope some of these ideas will help or inspire you to break out of the box and try something else. Forcing disk capacitors to work at UHF is risky. Good luck, let us know what works out.
More information about the Amps