On 11/01/2016 07:06 AM, Steve wrote:
What effects/damage are you seeing?
Arcing somewhere which kills the zener diodes in the screen supply.
Does it always happen in the same place?
I don't know where it happens. That's the problem. I am asking
questions in an effort to improve my understanding of relevant
theory in hope that may lead to a reasonable guess as to what the
problem is. Really, I am grasping for straws here.
This amp has the screen grid connected directly to chassis, with the
cathode 'floating' for DC but heavily bypassed for RF. It is a grid
driven amp. The screen voltage is stabilized by a shunt zener diode
When the arc occurs, it is a frying sound. This only happens with RF
present. It kills the zeners within a second or so. I'm quite
certain it is an arc from plate circuit to chassis, which means
plate to screen. The fact that I hear a frying sound puzzles me
somewhat. My experience in the past has been when B+ is involved
there is a loud snap. In this amp, virtually everywhere there is HV
RF there is also B+. Yet, the arc is a frying sound with no big
snap. This has happened twice. The first was at a power level of
around 1500 watts, the second less than 900.
I have hi-pot tested the tube to 10,000 volts, at which potential it
shows just about one microamp leakage plate to screen. I think
Unfortunately I took the plate line apart to look for damage before
I had the hi-pot tester. I found some slight darkening of one screw
and the edge of the hole where the teflon shoulder washer was. Not a
lot. Not anywhere near what I would have expected to find since this
thing had arced twice, destroying the 50 watt zeners both times.
But, I know all the material was impeccably clean when it was
assembled, so I believe something must have happened at this
I should add the teflon is .010" and is coated both sides with Dow
Corning 4 Electrical Insulating Compound. In theory this excludes
practically all air from getting into the plate line sandwich. In
reality, small pockets of air can and do get trapped during
assembly. *Maybe* the fact there is very little air accounts for
very little visible evidence post-arc?
Can you run the amp with reduced plate voltage?
With a few kV of dc+rf, you need air leakage paths >0.2"/5mm to avoid
breakdown. If there's anywhere in the assembly where the teflon doesn't
overlap the brass far enough it might be a weak point.
The distance from edge of hole in top brass plate to screw which is
in contact with lower plate is no more than 0.15". In theory the DC
4 compound is supposed to fill any tiny voids here (under the
shoulder washer) and prevent an arc. This is a proven amplifier that
did work at one time - for several years. After not using it for a
long time I recently rebuilt and cleaned up the whole thing. But it
has this arcing issue now. Unfortunately I have no access to a lathe
and my teflon shoulder washers are far from perfect. Maybe when I
reassembled it I didn't get enough DC4 compound in this one spot to
fill all the tiny voids.
Teflon is very soft
I'll say. I hate working with the stuff. It's almost a semi-solid
material which will easily distort under pressure.
Air gaps can be a weak point too. For example, consider if the teflon is
.01"/.25mm thick and the brass is slightly out of flat so there's a
.002"/.05mm air gap somewhere when it's clamped together. In that area
the air and teflon are like two capacitors in series and the voltage
divides between them inversely with capacitance. In this example,
roughly 30% of the (rf + dc) voltage appears across the air part which
is likely to cause breakdown and lead to damage to the teflon.
That's something to think about.
Thanks for the reply.
At this point my 'blind' plan is to rebuild the plate line again,
using .015" teflon instead of .010" (shouldn't make any difference
in performance), making sure everything is very clean (again), using
fresh DC 4 compound (I used old stock before), and new shoulder
washers (such as they are). This time I will hi-pot test the line to
10 kV prior to trying the amp. It should be able to pass that test.
If it doesn't, I will know I have a problem somewhere. I don't know
what else I can do.
Amps mailing list