at reduces
life and what you can do about it.
Article prepared for www.audioXpress.com
I
n a recent issue of
audioXpress,
Edwin G. Pettis wrote an article in
which he describes how the lack of
a sufficient space charge in a power
tube can promote the possibility of arc-
ing within it (Why Power Tubes Arc,
p. 28, Nov. '04). His description of the
emission process and how peak currents
cause it to deteriorate was excellent.
Modern design practice can also con-
tribute to this process, and significantly
shorten the life of a power tube if it is
not properly applied. When older equip-
ment is updated, and new designs are
contemplated, paying attention to a few
basic rules will help ensure maximum
life from your tubes.
MODERN POWER SUPPLY DESIGN
One area of amplifier design thats un-
dergone some of the biggest change
over time has been the B+ supply.
Years ago, an economical power sup-
ply for a fixed bias class AB1 amplifier
might consist of one transformer for all
the power requirements, with a center
tapped full wave rectifier (tube) and
capacitor input filter. With only mod-
est
µF and maybe a small choke for
filtering, this design produced only
fair regulation. A more substantial ap-
proach typically used a separate large
high voltage transformer with the same
rectifier design, but used (possibly)
solid-state rectifiers and a choke input
filter instead. With (usually) multiple
chokes, oil caps, and a large bleeder re-
sistor, this supply had good regulation
under load.
With modern design however, the
same performance is usually met with
a transformer of much lower voltage
and higher current, in (typically) a
solid-state voltage doubler design. This
supply often uses a separate trans-
former, high
µF photo flash caps and
usually just a small dropping resistor
(if any) for filtering. The regulation of
this supply is very good, usually sur-
passing that of the choke input design
above.
This new supply has two major dif-
ferences from the old designs. (1) The
effective output impedance of the sup-
ply has been lowered significantly by
the large photo flash caps used. They
now provide a very large reservoir for
the amplifier to operate from. (2) The
internal resistance of the supply has
been similarly lowered by the reduced
winding and filter losses possible. This
allows significant current flow to main-
tain the low output impedance under
very heavy loads. It is these two quali-
ties that produce the excellent dynamic
regulation at the output of this supply.
But the tight regulation can also dam-
age the power tubes if it is not properly
accounted for in the overall design.
Ironically, what enables it to do the
most damage is a common feature that
was supposed to help prevent damage
in the first place.
(
Figs. 1A-1C)
DELAYED B+
Most of the power supplies from yes-
teryear applied the B+ to the amplifier
tubes in one of two ways. It was either
there quicklywell before the audio
tubes started to conductor applied
gradually by a cathode rectifier tube
after the other tubes were almost fully
heated. Either way, this allowed the
tubes to start conducting in a smooth
and uneventful process.
A few of those supplies used a delay
relay and possibly a separate trans-
former to apply the B+ after the am-
plifier tubes were fully heated. With
this approach, the tubes are usually
turned on more abruptly. However, the
internal resistance of those supplies
limited their peak current capabilities,
so any damage to the tubes was lim-
ited as well.
When delayed B+ is combined with
modern supplies however, there is lit-
tle internal resistance to protect the
tubes. The peak current capabilities
of these supplies are such that the
potential damage to the tubes can be
significant when the B+ is turned on.
This damage can either take the form
of an arc, which will end a tubes life 2 audioXpress 2005
www.audioXpress.com
immediately, or be more gradual in
effect, damaging a tube over timeor
both. These are two separate problems,
requiring two separate solutions.
ARCING
Equipment from hifis golden age was
never particularly prone toward out-
put tube arcing. Most of this equip-
ment used very reliable high trans-
conductance suppressor grid or beam
power output pentodes because of
the efficiency and/or versatility they
afford in operation. But it was also
these tubes that became so problem-
atic, although only under certain con-
ditions.
When the arcing first began dur-
ing the '70s, it was initially blamed
on cheap foreign tubesand some re-
ally were. But when NOS examples
started exhibiting the same behavior,
it couldnt be so easily written off.
With little information available on
the matter, the sporadic arcing con-
tinued unchecked. It was also during
this time however, that the power sup-
plies of this equipment were starting
to be upgraded.
These supplies were easy to im-
prove. Silicon rectifiers typically re-
placed a higher drop tube rectifier for
lower internal resistance. Filter caps
in the basic supply were increased
by a factor of five or more to lower
the output impedance and add cur-
rent reserves, and a B+ delay switch
was usually added to do what the new
silicon units couldnt. Since most of
this equipment used a single power
transformer with (typically) a tap
on the HV winding for bias, the B+
switch was usually placed at the out-
put of the B+ filter so the bias supply
(if present) would always operate. All
the characteristics of modern design
were now coming into play.
The upgrades provided obvious
performance improvements, but also
changed the relationship between the
power supply and the output tubes in
two important ways: (1) After preheat,
the full B+ (or more) was now applied
from a running B+ supply with fully
charged high
µF filter capacitors.
This new arrangement allowed for
high peak current capabilities when
the B+ is turned on. (2) The output
FIGURE 1A: Classic economical power supply. Even with a low drop rectifier
tube such as a GZ34, the internal resistance of this supply is
165.
FIGURE 1B: Classic supply with better regulation. The internal resistance is
now
140, but the combined bleeder and choke action effectively reduces
this to
85.
FIGURE 1C: Modern supply design. The internal resistance of this supply is
only
28 with significant peak current reserve. audioXpress 2005 3
tubes were now operating from a
power source of much lower imped-
ance. After these changes, many out-
put tubesregardless of their age
began to arc.
(
Fig. 2)
TUBE/CIRCUIT INTERACTION
While the condition inside a tube
that produces an arc under these new
conditions can be debated, it is al-
most certainly a violent oscillation,
and therefore represents instability at
its worst. Long ago however, I deter-
mined these arcs were always to the
screen grid, and therefore represent
a screen stability issue. As it turns
out, controlling it is very easy. But
understanding what factors combine
to produce the instability would be
helpful for any contemplated design
or modification.
Due to the lack of any definitive
reference material for this issue (to
my knowledge), the information Ive
gathered has been derived empirical-
ly. However, the factors involved are
definable and the results repeatable,
which makes the information very
reliable. Specifically, there are four
elements of design that can combine
to produce screen instability:
(1) To obtain maximum power out-
put means running the screen at
or near its design maximum volt-
age rating. Although many de-
signs do this, it does not particu-
larly promote an unstable screen
by itself in practice. But it does
mean that the screen is operating
at the safe limits for the physical
design of the tube.
However, if
the screen is running at greater
than 80% of its design maximum
voltage rating
in combination
with the other three elements, it
is a factor in producing screen
instability.
(2) Many designs operate the plate
and screen at nearly identical DC
voltage levels. This includes tri-
ode configurations, and many ul-
tralinear and pentode designs as
well. Although this too is a very
common design feature, it does
not particularly promote screen
instability by itself in practice
eithereven with the screen
operating near its design maxi-
mum voltage limit.
However, if
the screen is operating at 85% or
more of the actual plate voltage
in combination with the other
three elements, it is a factor in
producing screen instability.
(3) Fixed bias is often used to maxi-
mize power output, reduce dis-
tortion, and provide greater bias
stability