The epoxy issue with tektronix transformers is well
known, and
has been explored extensively. Many people have tried to
use
field expedient techniques to solve the problem, and all
have
failed in the long term. Basically, tektronix used an
unsuitable
epoxy that degrades over time, and takes on moisture. The
moisture
and degradation increases its already lossy nature, and is
the
proverbial "straw that breaks the camel's back" in an
already
touchy HV design.
The HV design is "touchy" because it is a Hartley
oscillator that
is being heavily loaded. It has about 5 watts of filament
load
in a 547 design, slightly less in the 647 with its 3
instead of
5 tube rectifiers. The oscillator needs sufficient
feedback to
run, and the ferrite core is of an ancient variety first
made by
Allen-Bradley back in the late 1940's. It is starting to
get
very lossy at 50-60KHz where the oscillator runs, and its
Curie
temperature is too close to room temperature for safe
operation
over the temperature range the specified for the 647
family.
Ok, what happens: Well simply, as the loss in the epoxy
rises,
the current drawn by the oscillator rises. The rise in
current
causes the I2R losses in the transformer primary to
increase,
further heating the core. As the core gets hot, it gets
closer
to the Curie point, and as such its internal losses rise,
and
the whole process quickly runs away.
This Hartley oscillator design first appeared in the 513D
scope,
as far as I can tell. There it had the entire HV section
soaked
in oil, and ran at about 400Hz... it used a somewhat
larger
core to handle the low frequency... a very conservative
design.
The next time I saw this design was in, well every 500
series
scope made. By the time the 535 came along, it had reached
its
"modern" configuration, with a beeswax impregnated core,
air
cooling, and 5, 5642 rectifiers acting as multipliers for
the
various anode voltages. Its frequency was about 60KHz.
Even though the core was right on the hairy edge of its
loss
curve, because it was in free air, and was mounted right
near
the fan, everything worked nicely. Tektronix used the
design
virtually untouched until the early 1960's when the 647
was
designed... and coincidently, the 545B and 547.
The 647 was an attempt to make the first ruggedized all
solid
state 50MHz scope. For it, they did two things, first
created
an enclosed plastic box to hold the entire HV section, and
second, changed the transformer to an epoxy varnished
coil.
Historical evidence embodied in the circuitry, and the
physical
embodiment of the circuit show that tektronix immediately
found
the HV section was too hot and too lossy for reliable use.
But
I am guessing they had spent their wad, and were
committed...
and being 20 some years since the original circuit was
designed,
by then nobody in tek really understood the original HV
section
design...so they prepared several expedient patches to the
problem.
One patch was to open a hole in the bottom of the plastic
box
to allow fitting a bar of aluminum, or beryllium oxide
ceramic,
to heat sink the transformer core to the chassis. This
appears
in production 647's, but not 545B/547's. Another was to
replace two 5642 tubes in the 647 supply, with silicon
rectifiers.
And another was to add a fuse to the HV oscillator supply
in
the 647, and add a shunt diode to the screen grid of the
6AU5
oscillator int the 545B/547 to limit the maximum drive to
a
safe 100V (120?).
Later, after production, fixes were to replace all the
5642's,
with silicon, and to change the potting to a black
epoxy...
I suspect that that transformer is triply potted, first in
epoxy varnish, next in silicone, and finally in the black
epoxy... Other scopes, like the 565 used epoxy varnish,
potted
in silicone RTV, in an aluminum can... I'm surprised they
didn't
toss some silica gel bags into the HV compartment as part
of
the fix.
Nothing worked! The root of the problem is the old formula
core, and they never changed it! It even shows up in the
5000
series scopes and beyond.
Anyway, after trudging through all of the field expedient
fix
attempts with my own 545B and 547, I started to study
winding
HV transformers, and produced a remanufactured transformer
for
that family that reuses the old core. The winding is
beeswax
impregnated, and all other parts are new manufacture. A
little
poking and prodding will reveal that the group is
saturated with
545B's and 547's that contain my remanufactured
transformers.
Since I am now the proud owner of a 647A scope with the
transformer disease (Thanks Chris!) I am now working on a
rewind
for the 647/647A transformer. I expect to have it
available
this winter... if I can keep the copper thieves from
making a
return visit to my machine shop...
Like the 545B/547 transformers I am currently making, I
plan
to use a beeswax impregnated winding, and there will be
enough
room for the 5642 filament windings. It will look a little
different, in that it won't be a plastic potted winding,
but
rather will be naked, like in the older scopes.
I still have some work to do on my winding machine, as
this
transformer is a little more challenging, and there will
be
some testing to be sure that the beeswax can handle the
environmental needs of normal use. We will probably have
to
lose some of the upper end of the temperature range,
though as
it is above the melting point of beeswax. If necessary, I
am
primed for the effort to use modern epoxies.. but I hope
it
won't be necessary for this scope.
Stay tuned!
-Chuck Harris
acuffe@... wrote:
> I bought a basket case of a 647 mostly to get the
11B2 for my 647A. After
> repairing the 11B2 with help from this group, I
decided to see if I could do
> anything with the 647. I found a wiring error,
several bad transistors, an NPN
> transistor in place of a PNP, and an open solder
connection on one of the heater
> pins on the CRT base.
>
> At that point it was working, but it had the common
HV transformer problem. After
> about 45 minutes, the 2A fuse feeding the HV supply
would blow. I monitored the
> current to the HV circuit, and found that it took 45
minutes for it to go from
> around 400mA to 1A. When it hit 1A it would get to 2A
within a few seconds.
> Keeping the transformer cool with a fan allowed it to
run all day, but it's not
> very practical. I tried replacing the HV rectifier
tubes with solid state diodes.
> This reduced the current demand by about 50 mA, but
didn't help with run time.
> Interestingly, the transformer didn't appear to be
overheating (the core only got
> to about 50C, and the outside of the windings were
cooler than the core.
>
> I noticed that when the transformer was hot, the
epoxy potting was quite soft and
> rubbery. I decided to try to remove as much of the
potting as possible. Using a
> hot air gun, I was able to get most of the epoxy off
the windings. I had to be
> very careful not to damage the windings (I used a
plastic tool to chip away at the
> epoxy). I reheated it every time the epoxy became
hard to work with. Be careful
> not to break the 3 floating secondary wires since the
epoxy was supporting them.
> Next, I gave it a 4 hour bath in 120C bee's wax to
try to seal things, and bake
> out any moisture. I did see a few bubbles, but they
may have just been air. If I
> did it again, I would probably skip that step, and
use a more modern HV sealant.
> I wanted to use something that was known to work as a
baseline. Finally, I
> injected some silicone rubber in places that looked
at risk for arcing.
>
> After installing the naked transformer I noticed an
immediate improvement. Cold
> supply current was down more than 50mA, proving that
removing the epoxy has
> reduced the losses in the transformer. I plotted the
current over time and it
> leveled off at 332mA. Compared to my 647A with a good
original transformer, it
> follows an almost identical curve. The 647A is higher
overall because it still
> has tube rectifiers. I should have plotted the
current before, but based on my
> observations, it would have shown a roughly
exponential rise in current.
>
> Andy
>
