[Boatanchors] WTD: 3CX3000A7 Tube - UPDATE

[J. Forster]

A few weeks ago, several members of this list gave me helpful advice
regarding replacement of the Eimac 3CX3000A7 tube in my ham-radio HF power
amplifier.  These members' collective advice enabled me to find a good
replacement at a good price.  I've thanked them individually; now I wish
to express my gratitude by sharing what I learned from the experience.


On a fateful day in early October, I left a low-density polyethylene
(LDPE) box atop my amplifier, blocking the cooling-air exhaust port.  Of
course the amp was off at the time.  This amp, a Henry 8K Ultra, uses a
3CX3000A7 in grounded-grid.  The amp dissipates 1500 watts just idling,
and more when it's driven.  Because it generates so much heat and blower
noise, it is made to be located remotely.  Only its controls and meters
are on the operator's desk.

On Oct. 19th I switched my amp on and I'd been using it for about an hour
when I smelled hot plastic.  My nose led me to the next room, where I
found molten LDPE dripping through the amp's exhaust port onto the tube's
anode heat-exchanger.  After removing the remains of box, allowing the amp
to cool, and extracting LDPE stalactites, I was relieved to find that the
amp still worked normally.  I thought I'd dodged a bullet.

But, the next day, after the amp had run hard for a couple of hours in a
contest, it failed.  Its symptoms were (1) no RF output; (2) plate current
constant at 0.48 A DC regardless of RF input (drive); and (3) small
_negative_ grid current.  These symptoms were exactly those of a
filament-to-grid short inside the tube, or a filament-to-chassis short at
the tube socket.  (The grid is connected directly to the chassis.)

The DC plate current was just as expected for a 3CX3000A7 with zero bias
between grid and filament, and 4500 VDC on the plate.

No RF power can appear in the plate circuit when no RF AC voltage appears
between the filament and the grid.  For no RF voltage to appear between
filament and grid despite the application of normal RF drive power, the
short-circuit must have very low impedance.  The driving source impedance
is low, of the order of 12 ohms.  So the short must be within or very near
the tube.  It certainly cannot be on the other side of the bifilar-wound,
common-mode, RF choke through which filament heating power is applied.

Small negative (DC) grid current is expected if the grid is shorted to the
filament, because the filament current (in common-mode, the _cathode_
current) is always negative with nearly the same magnitude as the
(positive) plate current.  With the grid and filament connected together,
negative cathode current (of 0.48 A in this case) divides between the grid
circuit and the cathode circuit in proportion to their DC conductances,
which are essentially the conductances of the respective ammeter shunts. 
This ratio is equal to the ratio of the meter scales, in this case 1:1000.

After cutting power to my amp and letting it cool, I found that it worked
again; but then it failed again, and again....  The MTBF became
progressively shorter; and, after about a dozen cycles, the amp was
useless.

I opened the amp and removed the tube.  With a Triplett DVM I could detect
no conductance between the grid and filament of the tube, which did not
surprise me because now the tube was cold.  Nor could I detect any
conductance between the chassis and the filament contacts of the socket in
the amp.  I assumed that short had been occurring within the tube, when
the filament and/or grid became sufficiently deformed by thermal
expansion.  I knew that the grid and filament wires were tiny and that the
separation between them was tiny.

I wondered how to replace the tube, and my friend John posted my request
for advice to this list.  Based on the helpful suggestions I received from
list members, I decided not to buy a new Eimac tube, which would have cost
about $1,400.  Instead, for half this price, I bought a Chinese copy.  I
planned to spend the other half getting my damaged tube rebuilt by Econco.
 More about that below.

I was afraid of buying a Chinese-made 3CX3000A7 because I'd heard many
horror stories about Chinese tubes.  I'd heard horror stories about
Svetlana, too.  To be fair, I should mention, I heard that Eimac tubes
aren't what they used to be, either.  However, I have had good experience
with RF Parts, Inc.; RF Parts has been in business a long time and AFAIK
has a good reputation; RF Parts sells a lot of tubes, including tubes
bearing its own brand name of "Taylor," which are Chinese-made; and RF
Parts' warranty on its Taylor-brand tubes looked pretty good; so I ordered
a Taylor 3CX3000A7.

When it arrived, the Taylor 3CX3000A7 _looked_ fine, but I was cautious. 
I applied rated filament voltage with no grid or plate bias and let the
tube bake for 24 hours, hoping that filament heat (around 400 watts) alone
would be sufficient to activate the getter and at least partially clean up
the vacuum, in case the tube had sat on a shelf for a while.  Then I
connected a Variac autotransformer ahead of the primary of the
high-voltage transformer of my HVDC supply; and, still with rated filament
voltage, I increased the plate voltage gradually, starting with just a few
hundred volts, while keeping the grid bias zero or small so the plate
would draw relatively high current and would dissipate relatively high
power at low voltage.  My reason for keeping plate voltage low was to
avoid avalanche breakdown in case the tube was gassy.  My reason for
keeping plate dissipation high was to get the getter to work.  Little by
little, I increased plate voltage and dissipation over the course of two
days, until the tube had spent a few hours with plate voltage near 3 kV
and dissipation near 1 kW (with air cooling, of course.  I monitored
external anode temperature with a thermocouple.)

Then, I began applying RF drive power, watching the grid current and
calculating the DC-to-RF efficiency.  Everything looked just fine.  So I
declared the tube operational and used it at normal voltage and power for
26 hours in an HF DX contest, as usual transmitting 1500 W on all bands
from 160 meters (1.8 MHz) through 10 meters (28+ MHz).  This tube's
performance was indistinguishable from that of my original Eimac tube
before that tube failed.  For the same RF output power at the same
frequency, it required the same RF input power; it drew the same (small)
grid current; and it drew at least approximately the same plate current. 
(I do not observe plate current as closely as I observe input and output
RF power and grid current.  I don't care about plate current unless it's
WAAaay out of line.  I always monitor grid current, for safety; and I
always monitor RF output power, for legality.)

For a given radio frequency, the plate-circuit (pi-L network) capacitor
and inductor settings for maximum output from the amp were also the same,
from which I infer that the Taylor tube's inter-electrode capacitances and
other "stray" reactances and electron ballistics are also
indistinguishable from that of my original Eimac tube.

Bottom line: I am very happy with the new tube, which cost half as much as
a new Eimac tube would have.  Only
time will tell how long it lasts.

I sent my original Eimac tube to Eimac's sister company Econco for
rebuilding.  A few days ago I got a telephone call from an Econco engineer
telling me the results of his "acceptance" or "qualification" testing of
my tube.  I had expected to receive such a telephone call telling me
whether the tube could be rebuilt, and if so, for what price.  However,
this guy told me that he could find nothing wrong with the tube!  He said
that the tube shows no sign of having been overheated or having been
damaged in any way.  In particular, there's no darkening or discoloration
of any exterior surface, metallic or otherwise; and the vacuum is
effectively perfect.  He had tested the tube with 15 kV plate-to-grid and
(more impressive IMO) 5 kV filament-to-grid.  Filament emission was just
fine, typical of a tube that's been used for relatively few hours (as mine
has been).

He agreed with my interpretation of the symptoms I'd observed, namely that
they implied a grid-to-filament short; but he thought the short must have
been external to the tube.  He said that, if the filament and/or grid of
my tube had been deformed enough that a short could occur when the tube
got hot, then the tube could not have passed his 5 kV grid-to-filament
test.  He said that, when a tube of this type is overheated, the first
thing to go is the vacuum, because a metal-to-ceramic seal fails -- NOT
the grid or the filament structure.

Bottom line:  Econco is returning the tube to me, at no charge except for
UPS Ground shipping.  Wow!  How many times have you paid someone more than
$100 just to tell you that he couldn't find anything wrong with your [fill
in the blank] ? !  Ya gotta love Econco.

I'm left wondering how a filament-to-grid short could have occurred
external to the tube.  I can only guess that a loose screw or other piece
of metal found its way into the space between the tube socket and the
chassis, began shorting one of the filament terminals to the chassis, and
then fell out when I pulled the old tube out or pushed the new tube in. 
Eventually I'll take the amp apart far enough to look for a culprit. 
Meanwhile, my amp is working like a champ.  Last night I spent another 12
hours or so using the amp in a contest, again with no sign of trouble.







======


A friend has a Henry Linear Amp and the tube has died. I'm posting for
him. He needs a replacement.

It's an Eimac 3CX3000A7. Preferably new, but would consider known good used.

Also what do folks know about the Svetlana 3CX3000A7 or the Chinese tubes?
Svetlana claims that it's an exact replacement for the Eimac tube.

Thanks,

-John

=========