[Top] [All Lists]

[Amps] More HV feed through capacitor

Subject: [Amps] More HV feed through capacitor
From: Larry <>
Date: Sat, 11 Feb 2012 12:52:02 -0800
List-post: <">>
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. It is 80x150 mm and is 
"microstrip" type. I.E. one side is ground and the other (hot) side is 
open. It is mounted on one side of the anode enclosure near the anode 
stripline resonator. Since I have had poor luck with PCB material and 
high voltage, I made sure that the edges were etched back and lots of 
clearance around the "hot" foil. It hi-potted with no leakage at 6kv DC. 
Even so the thing jumped more than 1/4 inch and melted a large part of 
the copper foil.

After insuring I had no suitable feed through caps in the junque box, my 
next thought was a 1/4 wave coax stub. Several people suggested coax 
stubs of one sort or another but this is one of those concepts that is 
great in theory but in reality you can't quite get there from here. 
Making a 1/4 wave stub from RG-302 coax was easy. But when hi-potted, it 
went arcy-sparky at the open end at less than 3 kv. By the time I got 
the open end prepped and dressed well enough to handle 6kv, the other 
end wasn't much of a RF short. It is really hard to build a true "open" 
at these frequencies!

Several people suggested the RF Parts feed through cap. It is a bit 
large physically for this project but I could probably make it work. But 
before I mess with ordering it and dealing with the minimum order 
requirements and the resulting wait for delivery, i though I'd explore 
other options.

A single disc ceramic cap at these frequencies is worthless due to built 
in stray inductance. However, there are things a guy can do to mitigate 
these effects. With a combination of several 270 pF/6kv disc ceramic 
caps, some copper foil and a near 1/4 wave piece of RG-302, I was able 
to come up with a network that easily hi-potted to 6kv and still 
provided a bit more than 40 dB isolation between the tank on the HV lead 
to the power supply. This will put a little less than +20 dBm on the HV 
line, not great but at least it won't hurt anything. Still probably 
better than most HF amps with a small choke and a single bypass cap.

The amp was reassembled and tested this morning. No arcy-sparky but 
power out was lower than before, efficiency down to 38% and eventually 
the telltale odor of Allen-Bradly farts caused me to cease testing. Odd 
because there are no carbon comp resistors at all in the amp. 
Dis-assembly showed the anode choke to have burned off the enamel 
insulation exactly the same as a 12 meter resonance in old HF amp 
chokes! Now this is the same choke that has gone through previous 
testing with no problem.

This choke is supposedly connected to the half wave anode resonator at a 
current node so there really shouldn't be much RF to deal with. Given 
that the PCB cap arced 1/4 inch when it shouldn't have and this choke 
burned up with the new bypass assembly, my guess is that there is so 
much RF running around the anode enclosure that the bypass/choke 
assembly is "grounded" at a point on the side wall that is at a RF 
voltage node. If anyone has a better explanation, I'd sure like to hear 
about it.

What next? I will fiddle with the choke but don't hold out much hope for 
success. My present thinking is that even a feed through cap at that 
point will have problems. It might be necessary to run the cold end of 
the choke out of the box with a ceramic feed through insulator and put a 
small "doghouse" on the outside with the necessary decoupling inside.

Another option would be to make a "stripline" style capacitor from 
teflon sheets and aluminum so that the "hot" plate of the cap is not 
exposed to the RF field. I've used this method in the past with success. 
Even so, the choke would probably need to be tweaked.

73, Larry W7IUV
Amps mailing list

<Prev in Thread] Current Thread [Next in Thread>