WINDING PRACTICES, OUTPUT TRANSFORMERS

Over the last 4 years many people have emailed me for my advice about how
they could wind their own OPT for their own DIY amplifiers. I have always give
what advice I can, and a few actually succeeded although many just gave up
because transformer winding is a trade needing learned and practiced skills, and
only knowing all about what makes a good transformer as described in text books
or elsewhere at this website will never lead to any really good transformers unless
someone is at ease with basic workshop practices, has the workspace for production,
ie, the workshop, and has a good tool set, and has the time and patience to devote
to a time consuming activity.

But to keep my repeated advising time to a minimum, I now devote some time to
exactly how to wind a transformer.

The first thing required is a powered lathe. For those without electricity, a pedal
powered job will do since the amount of power is similar to a treadle powered sewing
machine that grandma may have used in 1920.

The lathe speed need only rotate at a maximum of 5 turns per second, or 300rpm,
and one does not need a huge amount of torque hence the pedal powered lathe
will suffice.

However, I have built my own powered lathe. This will then give you both hands
free to handle the wire.
My lathe uses some 100mm x 50mm and 100mm x 100mm timber offcuts from
building work used to make a chassis on which to mount a motor box and lathe
shaft with pulley belt drives.

Pictures below show some idea of the lathe with a bobbin being wound.

I bought a cheap electric drill for the motor, and clamp mounted it in a box for
more silent operation.
It couples with a flex drive to a 40mm dia fan belt pulley wheel mounted on a
12mm dia DRIVE shaft running in two 12mm ball bearings and trunnions bolted
to the woodwork frame.

The shaft is plain bright steel bar of 1/2" dia and was very cheap from an
engineering supplier. The ball bearings were also quite cheap from a tractor
spare parts supplier. A second 12mm LATHE shaft about 200mm long also
mounted in two firmly bolted bearings has a  200mm dia pulley for auto fan belt
at one end.

The timber chassis frame is like a giant U shaped frame, turned 90degrees so
each leg runs east-west in front of you when you look at it. The motor and drive
shaft is mounted on the rear furthermost chassis leg, the lathe shaft mounted
on the chassis leg closest.

The lathe shaft end away from the pulley has a 150mm x 40mm plate about 5mm
thick welded on to the shaft and exactly square to the shaft. The pulley belt
isn't too tight, and allows the belt to be sprung off the pulleys if needed when much
unwinding from a bobbin is attempted because there is no reverse direction possible
on the motor. There is a speed reduction between the drill motor and lathe shaft of
about 5 to1. I have a manual turn handle bolted to the 200mm pulley to allow slow
hand turning which is sometimes required to force wires to be where I want them at
ends of layers, because, as you will find, wire tends to lay up irregularly at the beginning
and ends of built up concentric layers.
 
A 100mm long x 10mm dia threaded rod is also welded to the shaft beyond the plate
so that bobbins can be clamped in place between specially cut out plates of plywood
to allow wires to be brought out of the bobbin as the layers are wound on.
I use a few screws in the ply to hold the long free ends of wires to prevent them being
tangled and yanked which would ruin a wound bobbin.
When I began I counted the turns at the end of a layer to confirm I had the right
number wound on. This is quite tedious, difficult and prone to mistake, especially
when wire sizes are less than 0.5mm dia so I made a mechanical turn counter using
an old automotive odometer mechanism powered by a copper clad wooden wheel
attached that is exactly 10 times the shaft dia so that for 1 turn of the bobbin shaft I
get a reading of 1 turn, with a turn being equal to what was 1/10 of a mile.
By carefully filing down the thick copper wire outside cladding of the wooden odometer
drive wheel I was able to ensure that for 300 turns of the lathe shaft the number 300.0
appears with error less than a fraction of a turn.

Before commencing the layered winding with accurately counted turns, you will have
got your turn counter arranged and checked for accuracy by counting the turns from
one to 300, and reading what the turn counter says. It should say 300.0 If you get 295.8,
the wheel diameter against the shaft is too large, so you may have to file it down;
hence I made mine with 1.5mm wire glued with epoxy around around a plywood wheel
of 124mm dia so that I could file it down to be 126.5mm, which is 10 times the dia of
the clean bright 1/2 inch dia lathe shaft. I filed the wheel diameter down until I got
very accurate turn counting.

The wheel and counter is mounted on a spring loaded wooden block so it can be
slightly swung away from the lathe shaft but is otherwise held against the shaft
with the old spring I found among the many old bits of junk in one of my junk boxes.

This allows me to swing out the counter and spin it up or down to a whole number
of thousands at the beginning of a new wind up it make recording the turns easier.

If I reverse wind to undo a mistake, the turn counter winds backwards automatically
and accurately.

I think I spent a week building the lathe and it cost less than $200.

I could not find a second hand coil winding lathe and the new ones were way too
expensive. Unless I employed someone to use it all the time to make transformers
the cost of a commercial winding lathe wouldn't pay for itself.
Nobody I know wants to learn to wind transformers accurately for a living unless
they gain employment at one of the dwindling number of commercial winding
workshops where mainly mains transformers are wound repeatedly, and as
competition to Chinese imports, which BTW, are usually always inferior compared
to best quality that can be achieved using best materials and neat layer by layer
winding methods.

There is no automatic traverse or wire tensioning device on  my lathe. Devising an
auto traverse mechanism where the wire is guided slowly across the width of the
bobbin as turns are laid on was beyond my capability when i made my lathe,
but anyone with a better mechanical ability would succeed at constructing a guide,
instead of having to manually lead the wire on by hand and slowly bring one's
hand across a hand rest to ensure no crossed over turns or gaps between turns.
So I feed the wire on with the spool of wire from a bucket on the floor, sometimes
with a cloth taped over the wire at the edge of the bucket to prevent over eager
wire loops springing off the spool and tangling or forming a kink in the wire that is
then pulled tight when the wire feeds up through my hand onto the bobbin.
I have a wooden rail about 500mm away from the bobbin on which to rest my
hands. This hand rest needs to be about 500mm away from the bobbin for almost
all transformers and chokes, and the leading on of the wire is easiest with this
distance.

Speed control for the motor is primitive but it works. There is a foot switch for on
and off, and 8 mains lamp sockets are mounted on a nearby wood block so
various numbers of lamp globes or various wattage can be connected in series
with the drive motor, so that I can get from 1/2 a turn per second to about 5 turns
per second. The extra light from the lamps help me see what is happening while
its turning. One has to watch very closely so that the wire layers are neat, flat,
and have the right number of turns, and do not pull down past insulation
sheeting at the ends of layers, and that gaps don't appear..

Apart from the lathe, you will need :-
A Carefully Calculated design sheet with exactly what is to be wound with
the bobbin details set out on an A4 page beside the lathe so the turns can
be recorded at the start and finish of each layer. If you accidentally leave a
whole layer of wire out of the design, the work is quite useless  later.
This is likely to happen if you get interrupted in your processes by visitors,
phone calls, debt collectors, or over zealous lady pals.

Side cutters and scissors.
Masking tape and felt marking pen to identify the ends of windings.
20mm x 80mm hard plastic blade sharpened along one end and edge to
be an adjust tool when slightly adjusting wires closer when gaps occur or
undoing crossed over turns. Never ever use anything metal to adjust wire
positions on the bobbin.
Pre-cut insulation material.
Tight fitting wire sleeving, preferably fibreglass hi-temp woven auto
grade, and not just shrink wrap.
Fibreglass adhesive insulating tape, 25mm wide.
Pre-made bobbins or carefully made DIY bobbins.
Grade 2 winding wire which is double enameled winding wire with high
temperature polyester-imide enamel.
Electrical varnish.
Vacuum impregnation tank.
Temperature controlled oven to allow up to 300C.

Alternative to electrical varnish, vacuum tank, vacuum pump and
oven heating:-
Using Estapol 7008 polyurethane epoxy two pack floor coating.

Patience, diligence, willingness, fanatic belief in doing it right or not
at all.
Time, for the above, and a good work place.
( Smoke without music is the result of a fool's shortcomings ).

Wind a choke to begin with.

It is impossible for anyone to expect to master the trade of transformer
winding in 2 hours. Once the lathe is fully set up, the first thing to be
attempted is a simple one winding smoothing choke.

You will need a bobbin. I try to buy mine already made from high
temperature rated molded plastic which are available from the small
number of suppliers to the transformer winding trades.

But quite often I have made my own bobbins. I start my making a wooden
mandrel which is the same dimensions as the assembled core inner
tongue. This has a 15mm hole bored through its centre to allow the
lathe shaft to go through. It is vitally important to be accurate when
working with wood or the bobbin with its cheeks will not turn true.
Where the wire bends sharply at each of the 4 corners during the
first few layers, the sharpness is reduced by slightly rounding of the
wood at the corners which means the mandrel must have an internal
maximum height about 6mm more than the stack of laminations to be
used. To make the base of the bobbin an electrical grade cardboard
strip cut to just the right width is bent tightly around the mandrel and glued
with wood glue and held tight with scrap wire turns. The two layers of
0.8mm thick cardboard ( and 0.8mm is a common size ) will keep the
wires at least 1.6mm away from the core, and when saturated with
varnish the cardboard will resist at least 4,000V.
 
When the glue has set, the wire clamp is removed, and the resulting
rectangular tube can be slid off the mandrel. I use 2mm thick fibreglass
sheeting for each bobbin cheek plate. It is very carefully cut to size with a
hole to take the end of the cardboard tube. Holes are drilled to allow wires
to come out wherever they need to be, but to ensure there is always an
available exit hole for a wire adjacent to a layer end or for a tapping.
I drill plenty of 3mm holes in the bobbin side to allow wires to exit.
I don't use slots because it weakens the bobbin cheek flange.
In addition to the bobbin cheeks, plates of at least 12mm plywood are
cut and drilled to provide support for the bobbin cheeks, and have
13mm holes for the lathe shaft. One plate is placed on the shaft,
then the mandrel with cardboard tube is placed on the shaft, then the
two cut out bobbin cheeks, with some epoxy glue to hold them to the
cardboard tube.

The bobbin cheeks are pressed outwards against the ply plates before
the glue sets. Don't use excessive glue or allow excessive glue to glue
the ply plates to the bobbin; you do want to be able to pull the assembly
apart later, but you want the cheeks to remain well glued to the bobbin base.
Applying glue so it **does** get only where you want it is very important.

So the surfaces to be glued both need to be wet lightly, and the easy
sliding fit will be filled with glue. I cannot stress how important it is to work
square level and to +/- 0.25mm tolerances so that when you are ready to
wind the assembly will turn true without wobble. Before starting to wind,
remove the the glued up bobbin from the shaft, remove the wood mandrel
carefully, and try some laminations, There should be 1mm clearance between
cardboard centre tube and core, and between tops of bobbin cheeks, and E
should be able to touch the I with 1/2 a mm clearance. Obviously, such
manufacture is the work of a practiced craftsman if it is to turn out right.
To do it well you will need a good drill, drill set, and jig saw, clamps, and few
other tools.

Winding the choke.
This need not be layer wound if the wire is less than 0.5mm dia but can
be wound on without making neat layers or insulation between each layer.
The aim is merely to fill a bobbin up with wire to get the hang of lathe working
on an initial project that isn't critical. The wire is wound on with little guidance
so the wires DO cross over each other often. On a choke the angle of crossing
of wires is very shallow, so pressure stresses on insulation remains low.
The wire is gradually  guided by a steady hand so it gradually traverses across
the layers to try to keep the surface of the wire built up fairly level and without
sudden wire direction changes. This will give you a useful article at the end
and get you used to not using too little tension and not too much tension and
after a couple of chokes you should be used to the lathe, and the way the wire
tends to tangle if you are not careful, and tends to get kinks from loops as it
unwinds off the spool on the floor. Arrange the wire and spool so kinking does
not occur. If it does with a choke, try not to pull the kink tight, but carefully bend
it out straight before proceeding further. Winding with 0.15mm dia wire is very
precarious since it will break so easily. If a wire break occurs, stop to make
a join by removing the enamel with a razor blade or fine sand paper and solder
the wires, tape around the join with thin adhesive fibre tape, then continue more
carefully.
At the end of the wind up approaches, the level of the wire in the bobbin will have
undulations of about +/- 1.5mm, so make sure none of the hills in the wire project
up past the cheeks of the bobbin or you won't be able to easily insert the
laminations in without having then grind against the wire. If you have wound too
many turns on, unwind them back onto the spool, which means winding the lathe
backwards and winding the wire in sections back onto the spool; this can be a
tangling experience, but eventually you will get used to what you have to do
or not do with wire to stop it tangling. At least you are not in a boat in the dark
while out fishing and dealing with fishing line! You won't catch a feed winding
chokes, but you can catch good music if you are persistent.

The layered winding requires more practice and persistence.
The next item one should wind is a layered choke.

For this you will need the insulation material wanted between each
layer of wire.
I like to use polyester at 0.05 to 0.1mm thick. I have bought a roll of umpteen
metres from a supplier, and when beginning a new item I cut maybe two metres
off the 600mm wide roll and fold it up into maybe ten layers thick and clamp
a wooden straight edge down over the layers of insulation. The straight edge
width is a sliding fit between the cheeks of the the bobbin. With a box cutter
knife I cut the strips of insulation about 10 at a time and a metre long.
Its a bit wasteful this method, but the polyester  is dirt cheap, and it is heck of
a lot better and easier than trying to mark the insulation and cut strips with
scissors!

It is very important to ensure the strip width of the insulation is no more than
0.1mm less than the distance between the bobbin cheeks, and never more
than the bobbin cheek distance. This way the insulation will lay around the
wire layers without crinkling up and occupying too much winding height which
will lead to loose windings, and windings that won't fit under the permissible
winding height or into the available window height of the laminations when
inserted. It is precision work.

With OPT, the insulation may be 0.2mm polyester, and stiffer and more
difficult to handle but the same approach is used.
The result is that I get  strips of insulation which fit neatly between the bobbin
cheeks, and where I want it to be and in no other place.

The lathe is set ready with a bobbin as described above. An estimate of the
turns per layer for the wire size has been established during the design
process allowing for the fact that wire imperfections and very slight gaps
between wires will mean that the calculated number of tight turns per
layer will never or rarely actually be achieved, especially if the wire is
less than 0.6mm diva o/a and there are slight variations in bobbin traverse
width. So the design should always allow for two turns less than
theoretically possible for wire under 0.6mm.

The in-going and out-going wire entry points should be carefully planned
so to allow easy terminations after completion. Usually this means that
one side of the bobbin is devoted to primaries, the other to secondaries.
This gives no awkward 1/2 turns added anywhere, although with large
power transformers a half turn may have to be allowed so the correct
heater voltage is obtained from the few turns of thick wire involved.
But with a choke winding, there is just one side for the in wire and out
wire. Always start with sleeve insulation on the in wire so it projects 25mm
into the bobbin wind area, with 25mm outside. Allow 200mm of loose
wire end outside the bobbin and wind around a screw in the holder plates,
and watch that any wire ends cannot get tangled or gripped in stationery
gaps in bearing trunnions etc so thus avoiding snapping a lead out wire
and completely ruining a winding.

Use a choke wire size that will later be useful for an OPT primary
say 0.4mm dia.
With neat layer by layer winding, commence winding slowly. Soon the wires
will try to cross over each other or you get get gaps. Stop, unwind crossed
turns, and/or adjust the few turns together closely with the plastic blade you
have. Its better than a thumbnail. After some time you will know how to guide
the wire to minimize gaps and crossed turns and to be fast about it all.
NEVER use a metal blade or screw driver to adjust layer turns.
Speed does not seem possible at first. I may take 15 minutes to wind 40
turns on properly, especially during the first few layers where the wire bends
sharply 4 times on each turn. When you reach the end of the layer, apply a
small piece of ordinary sticky tape to hold the final turn tight. The turn count
should give about the right number of designed turns. If you seem to have
no gaps, and still need to wind two more turns on, check the wire size.
Check that what looks to be gapless is gapless, and try closing up the turns
together more tightly with your plastic blade.
Often this will give the wanted space for the last two wanted turns.
Such adjustments can take the time needed just to get a layer on.
Wind a layer of 0.05mm insulation on and have an overlap of 10mm in an area
which is not covered by iron when it is inserted.
cut off the spare insulation with scissors, and tape the insulation overlap tightly.
Consecutive overlaps will increase the height of the wind up which could
make the required wind up too fat to fit into the lamination windows if the
overlap and holding tape is in the wrong place.

The first few layers will be most difficult to wind because the shape of wires
under the insulation will tend to cause gaps between wires, and the wire
won't lay properly, so don't use such excessive wire tension that adjustment
of wire gaps and crossed turns is impossible for each 10mm of traverse
width. But you should get the wanted turn count on.

After about 5 layers of fine wire the bend around the bobbin is less sharp
and the wire gets easier to manage and by the 15th layer the traveling
gets easier but there may be a tendency for layer bulges and bumps at
the end of layers to develop. Also wire may tend to pull down past the
end of insulation which doesn't quite reach over to the bobbin flange
cheek. After a few chokes you begin to know just how many turns per
layer you should put on and how important it is to make bobbins with parallel
flange cheeks, and cut the insulation correctly, and never to rush a winding
project.

With 0.4mm wire on a bobbin meant for 25mm tongue wasteless core, the
traverse width will be about 33mm, allowing about 68 turns. You might get
17 layers on which makes 1,156 turns. As a layer winding, this may take
you at least 5 hrs without the practice that tradespeople develop who would
do this in an hour or two or much less. The problem is that 5 hrs of your time
in a western nation could be worth $150, but a choke could be bought from
Hammond Engineering for $25. Spare a thought for the nimble Chinese
Person working for $2 per day, 12 hours a day, seven days a week, and
driven crazy with turns of wire in lousy hot noisy conditions with poor lighting
and a hard nosed boss. Under such conditions, there is no way one can be
sure there are no crossed over turns or gaps between turns, or a correct
turn count, which is why I NEVER buy any output transformer with a
dumbed down design made in asia. Crossed turns mean local pressure
points which lead to short circuited turns and early transformer failure.

For a choke a crossed turn is not a big disaster, although if a shorted turn
or turns occur, the inductance of the complete choke is reduced dramatically,
leading to a choke that does not filter. A shorted turn or turns in an OPT
leads to severe bass response problems and probable destruction of the
output tubes/devices.
When an OPT fails in an expensive amplifier it is a real problem because it
may be difficult to re-create a spare made to a one off design in 10 years' time.
So hence the OPT must be wound with great care, and used in amplifiers
with active protection measures against excessive tube currents which may
overheat thin wire primary windings and power supply choke windings
leading to shorted turns due to heat softened insulation.

The step from winding layered choke windings to an OPT is not a
huge step but then you have to learn about having maybe 30 ends of
windings and taps to deal with. That becomes easy as you learn that 30
connections means that you have 30+ wire ends to cope with so you
mark each one with masking tape and label it with a pen so you know
where you are as you go.
See the Image 1 below for the way to draw up the winding details
to allow the easy winding of an ultralinear OPT No1 mentioned
elsewhere in OPT theory pages.

Image 1.

For the above OPT, when you start winding, set the turn counter at zero
and begin with the primary at anode 2, bottom of the sketch, and label the
wire "1" and proceed left to right and when the layer is finished tape the
wire to the side of the bobbin holder temporarily while a layer of accurately
precut 0.05 polyester sheet is wound around the layer and taped into place
with a small piece of adhesive tape.
Proceed with the next layer. Before going right across this time from right
to left, stop to remove the tape holding the 0.05mm insulation in place and
continue to then complete the second layer to connection labeled "2"
which is brought out through a hole in the bobbin cheek and wound around
a screw head in the bobbin holder.
It is essential that wire not be allowed to slacken off at any point in the
winding process and it must be kept tight at all times.
Removing the small tape to hold the ends of insulation sheets prevents a
bulge developing after many such tape layers during the wind up. Always
overlap the insulation ends 10mm and locate the overlap adjacent to where
wire enters and leaves the bobbin lest the bulges from overlapping builds up
the height of the winding too greatly to be able to insert the E laminations.
The process of adding layers of wire and insulation proceeds upwards as
shown above. At the end of each layer, write down the turn number reached
and make sure the required number of turns is achieved in each layer.
Having 0.05mm insulation between each P layer which is usually fine wire
between 0.3mm and 0.6mm dia makes it easier to adjust groups of turns
together with a plastic thumb tool so that gaps between turns are avoided,
and all turns are pushed up close together all around the bobbin. No gaps
means you should achieve the right number of turns in each layer. Be prepared
to find the wire seems to have a mind of its own and gaps and crossed turns
may still occur, especially in the first 1/3 of the wind up where the wire has
to bend sharply around the rectangular bobbin which causes it to easily
cross over other turns or develop a gap, and be prepared to stop, wind
backwards a few turns, adjust a gap out, adjust a crossed turn out,
and re-proceed without tangling or kinking the wire. One needs to be alert.

In the above OPT where any taps in the primary have been designed to
occur at the end of a layer, the layer end is brought out by taping a wire
down along a layer, cutting the wire off the spool to allow say 200mm of wire
through the bobbin cheek to a ready screw in the bobbin holder. A label of
tape with number is attached. The 'screen 1' and 'screen 2' connections
at "4" and "15" are labeled the same for each in and out wire to the primary.

Where there are taps or ends of windings are brought out from somewhere
within a layer of wire, ie, for secondary connections H, I, J and K,  a layer
of fibre sleeving is placed on such a lead out to stop the crossed wires
crushing together to form a short. There will be a return wire treated similarly
so the winding can continue. Thin adhesive tape is used to secure such wire
lead outs as the subsections of secondary layers are wound on.

When I am done winding, I will have an OPT bobbin with many wires
wound around holding screws in the lathe plates. These wires are all
carefully unwound off the screws, and gathered together to allow
removal of the plates and bobbin off the lathe without yanking any
wires so tight that a break could happen. Never force anything.

The Es and Is of laminations should be arranged in piles ready for
insertion as soon as the bobbin is removed and the mandrel carefully
tapped out of the centre of the wound bobbin which is fragile, and
may try to bulge and spring apart if not "enclosed" with iron.

There is nothing so boring as stacking in the E&I laminations.
It is all too easy to get the correct sequence slightly muddled with an
E leg under another E leg from the opposite direction or one I lam short
or one too many. Check the layers of lams as you proceed each 5mm
in height, and redo where mistakes occur. Have the clamping yokes,
taped up insulated bolts, insulated washers, and nuts all ready for
assembly. When assembled, tap up the E&I lams to make the joins
gapless and the stack look plumb and square as the bolts are tightened.
When assembled, the bobbin will feel a little loose in the core due to
clearances. Place phenolic scrap plastic pressed in tight to eliminate
easy movement and make sure the end of the windup is well clear of
the core. At this point the circuit boards for terminations should be
wired/bolted to the outside of the transformer coils so that they are well
held to the bobbin cheeks with wood blockings or brackets.
The terminals can be turrets but I sometimes use plywood with small
brass screws, say No 4 gauge x 1/2" or 12mm long. These are available
from most hardware stores as cupboard hinge screws.
Brass plated steel screws are OK.

Where the transformer is potted, a flat phenolic of fibreglass heat
resistant board can be placed over the end of the lams with rows of
turrets or screws arranged to face into the chassis area when the item
is mounted on the chassis. Exact details can be chosen by copying well
made OPTs buy more serious suppliers. I never use flying leads of
different colors; usually there are far too many terminations on my OPTs
to be able to do that and a terminal board is necessary with a removable
box screwed down over the transformer.

When the OPT has been tested for termination voltage and phase
correctness, it is ready for varnishing or waxing.
Varnish is much more difficult to apply and is best left to someone
with a vacuum chamber and temperature  controlled oven.

In my town nobody has either which is easily available, so i made my own
vacuum chamber with an old pressure cooker container and I pull a 95%
vacuum by taking a pipe from the chamber to the *intake* of a small
compressor I bought. The oven is a frypan with an extended lid and gives
good heat control, although it took awhile to work out what setting of the
temperature dial to use to get an iron transformer up to the temperature
wanted to make the varnish cure properly so that the temperature was
uniform within the tranny.
But the second hand vac chamber cost $5, s/h cooker cost $5, and
compressor on special since no-one wanted a low power model from
Poland cost $100. I scrounged some auto tubing for vac leads....
Vacuuming the tranny is easy. Place the tranny submerged in a vat of
varnish within the chamber and tighten down the cover. I have a 10mm
steel plate 300mm square with rubber seals and 4 x 10mm bolts to clamp
over the cook pot. The vacuum pipe from the cover plate goes into the bottom
of a transition glass bottle with sealed top and second pipe taken to the
compressor input. So any fuming or liquid extraction from the vac chamber
can be seen in the bottle which will act as a liquid arrestor to stop varnish
being sucked into the compressor which would ruin the compressor.
After a few minutes, the vacuum is as strong as it will ever get and the
majority of air is extracted from the transformer inner cavities, along with
nearly all the moisture. The air is allowed to return to the vacuum
chamber, and the air pressure forces the varnish to penetrate inside the
transformer to occupy where a near vacuum exists. I repeat the operation
3 times.

During baking, the varnish solvent is expelled by the heat and any tiny
unfilled area will be at least wetted with varnish. The varnish will cure
with 4 hours at 125C, which is about 1/2 the temperature rating of the
polyester insulation materials used.
PVC insulation will melt and cause shorts so it cannot be used.

Waxing is also OK for OPT and is done  by simply soaking the transformer
immersed in a vat of wax kept at 90C for a couple of hours with the
transformer placed so air can easily escape from the many holes drilled
in the cheeks of the bobbin. wax is drawn in by capillary action.
Vacuuming isn't needed and would boil the wax. I have used candle wax
but it does melt at 50C, and although a tube amp may have been
designed to not have a temperature in the OPTs of more than 10C
above ambient, a slight fault could heat a winding and leave a puddle
of wax. And on hot summer days the OPT temperature can climb above 50C.
The wax does manage to silence the windings from screaming with a sound
of their own due to small movements of the wires and iron. After all, a
transformer winding endures magnetically caused forces similar to the
forces in an electric motor where the windings are deliberately allowed
to spin. But movement must be restricted in a transformer. SE transformers
usually howl more than PP types due to the gapped core vibration with
unbalanced winding currents.
Potting helps stop the noise, but I don't like potting over a waxed OPT.
Many amps I have serviced have had waxed OPTs and have such
puddles of wax left by tubes becoming saturated until they eventually
became open circuits, sometimes due to the winding fusing open,
or short circuits when they caused the mains fuse to blow. Alas, many
old OPTs in old amps with thin wire in their primaries would suffer a
shorted turn or an open fused winding , regardless of whether they
were waxed or varnished. I have seen this occur in Leak amps, and
I have a Luxman amp here with 2 OPTs with shorted turns, and a
noisy mains transformer; its a major amount of work needed for a fix.

I also sometimes use roof pitch for a potting compound which is
indescribably messy and smelly when it becomes liquid enough to
pour in around a varnished transformer . The temperature needed
is about 200C, and melting point is about twice that of wax, so it does
not soften and flow out of the pot very easily, and it is a good potting
compound which adheres to the potting can which thus is prevented
from vibrating and being noisy.

I find potting with molten roof pitch to be quite good although another
smelly process because the pitch gives off fumes but does need
considerable heat to make it melt. I have a Primus camping stove to
heat a steel pot of pitch for potting, and the temperature of the liquid
pitch is usually well above 100C and damned dangerous!
But its how they mostly did many transformers many years ago
because it was cheap, and you can SLOWLY heat up a can with a
potted tranny and remove it for re-winding if need be.

Quad II amps have a type of pitch which seems to have a type of wax
content which makes the compound have a melt point between pure
wax and pitch, but which is sufficiently high if a fault occurs. This
compound remains super sticky, and damps anything loose, but
nevertheless I have seen this black goo leak out all over a Quad II
amp from an over heated power tranny after bias failure has occurred.
There MUST be active protection against bias failures in all new tube
amps to prevent damage to precious hard to find power and output
trannies because tube saturation from bias failure may not cause a
fuse to blow!

There is an alternative to vacuum varnishing and oven baking !!!

In August 2006 I wound a pair of OPTs for a pair of SET amps using
845 tubes. I trialled the use of polyurethane two pack floor varnish
instead of using electrical varnish. The product here is known as
Wattyl Estapol Polyurethane 7008, and is available in 1/2 litre cans
of part A and part B.
This product is a very hard wearing timber floor and bench top coating
for general use. It has good insulation properties and its dielectric constant
does not increase the self capacitance of the windings to any great extent
when applied. When equal volumes of part A and B are mixed together
in a clean small container and stirred well the clear and fluid liquid
remains a liquid for about 8 hours at room temperature. It slowly cures
to become a hard plastic which will adhere reasonably well to most
insulation materials but not dissolve them. I tested polyester and
polythene and found no dissolving tendency, and wire insulation is not
affected. After setting up to commence winding up a transformer,
a batch of mixed polyurethane is prepared. Only about 30ml is needed
at first. I apply the liquid with a swab made from cloth wound around a
6mm dowel and wired into place; its cheaper than mucking around with
a brush which requires cleaning in a mixture of Estapol Reducer and
methylated spirits. I throw away the application swabs after each
fresh small batch of polyurethane is mixed. The polyurethane is applied
generously to consecutive wire and insulation layers both before and after
and between all multiple layers of insulation where say three 0.2mm
polyester sheets are piled up to make up the 0.6mm required. Make
sure the workshop is well ventilated and has an extraction fan going
to draw away the toxic fumes.

Don't try to thin down the mixed polyurethane; it would ruin the product.
This whole process is indescribably messy, and the polyurethane will
try to get everywhere; it will drip all over the bench, smear onto you
hands and gear where it will drive you nuts. Have a bottle of methylated
spirits and lots of clean scrap cloth to clean you hands and gear as
required until things are not sticky.
Be sharp, be alert, be careful and quick, but don't rush and stuff things up.
If you don't complete a wind-up within say 6 hours, STOP, and leave it for
2 days at least before continuing. BUT, after you stop, and record your
turns and where you are, apply the timber blocks and clamp to the wind-up
so that when the curing of the polyurethane occurs over the next two days
the wind-up will have any bulge removed before continuing. If you don't
clamp up the wind-up then the polyurethane will harden and you will never
be able to compress the winding later. When finishing to move again in two
days, always finish after an insulation layer and don't paint over it with
polyurethane before the clamping blocks are used. Don't disturb the clamped
bobbin during the two days; the polyurethane fractures easily while 1/2 way
along the curing time.
 

The very smelly fumes given off by liquid Estapol 7008 could be toxic
to some people or cause an allergic reaction.
Skin damage could also occur if not cleaned off immediately.
A well ventilated work place is essential and an anti chemical fume 
face mask recommended.
Gloves will make you very clumsy, so be careful how you apply the
material.
Practice makes perfect.
I can say categorically that the process is a complete PAIN IN THE ARSE
but the rewards for taking the extra care to get it right avoids the possibility
that the vacuum impregnation of electrical varnish and baking doesn't
quite reach all voids within the wound up bobbin. Once having applied
the polyurethane at a low cost and having taken only 10% more time to
wind a bobbin, the vacuum and bake process does not have to be done
at all, and the trauma of heating the completed tranny does not affect
insulation.

When the transformer falls from the pot when stirring it with a poker in
the fire, the core with winding can be pushed into the fire further to heat
it till just red, then allow it to cool over night. Next day the bolts and wire
can be sawn off with an angle grinder and the iron is all there for re-use,
and unaffected by the slight additional annealing. The pot may be worse
for wear, and may need panel beating or copying. I roasted a wheel
barrow full of old fused or shorted chokes and transformers in 2003,
and now have a stock of hard to get laminations in many sizes for re-use
in filter chokes. There was some GOSS lams in among the lot and the
µ of the iron was no different after the firing than before it.

Winding mains transformers is subject to National Safety Regulations
laid out in the National Standard Codes of what ever country you find
yourself within and these should all be carefully adhered to before
winding anything that is connected to the mains.

Regardless of where you get your information from, if you wind a
mains transformer and it causes a  shock to somebody then
don't blame anyone but yourself, because you wound it,
not me. The legal systems of most countries will blame you,
and nobody else.

The main requirement of a mains transformer from the Authorities
point of view is ISOLATION, and SAFETY and to achieve good isolation
a vertically divided bobbin is the surest way. The mains primary is
designed to fill half the bobbin on one side, and the secondaries will fill the
other side. Many mains transformers are wound this way but they often use
random windings everywhere which is poor quality when one considers
that a mains transformer is permitted to have a T rise of 40C above ambient.
Such a T-rise is deadly over time to random windings because of the many
crossed over turns and localized pressure points on wire insulation which
tends to crush to cause shorted turn/s especially during a fault event
when the tranny may have a T rise of much more than 40C.
I like to wind all my mains transformers with GOSS and with B = 0.9Tesla
max with very well current rated and nicely layered windings so T-rise is
less than 10C above ambient, and the transformer is never highly
stressed, and unlikely to ever fail during the next 50 years.
I won't be around to mend one.

Image 2.

A batch of power and output transformers for 300 watt amps on the bench
The clamping yokes are made from aluminium angles. The sizes can be
estimated by the 300mm long ruler to the right side.
The two OPTs near the ruler have boards to terminate the ends of the 12
separate secondary windings to get waste-free load matching.

Image 3.

Another view of the primary and secondary boards for the 300watt OPT.

Image 4.

In this OPT I have 3mm thick aluminium angle yokes with hardwood
blocks with brass screws for the terminations to P and S windings,
P and one end and S at the other. The sizes can be estimated by the
centimetre ruler.  It weighs over 15Kgs.

Image 5.

Here we see the 300 watt amp OPT bobbin being wound on the home
built winding lathe.
Behind the G-clamp is a box with the electric drill used for the drive power
with lamp sockets on top for varying the motor speed. The bobbin was
hand made and you can see the plywood bobbin clamping plates with a large
plastic handled nut that tightens the assembly on the drive shaft.
Just above the roll of masking tape is a hand rest for resting hands while
feeding wire to the bobbin.

Image 6.

Its messy business winding transformers and here we have a picture of
a tap being brought out from a 48 turn Secondary winding on a 300w OPT
bobbin.
There is fibreglass sleeving and covering tape to keep the last wound turns
taught and the tap in position while the rest of the layer is completed over
the black coloured insulation material.

Image 7.

Here is a close up of  the 300w amp power transformer with the two
hand made bobbins for the OPTs showing the timber mandrel inserted
into one bobbin. The empty bobbin has grey electrical cardboard former
and you can see the white fibreglass cheek plates and all the holes to
allow the wires to enter or leave at whatever height is needed. 

METRIC WINDING WIRE SIZE CHART
The metric winding wire sizes were kindly given to me by a local Sydney
wire and transformer parts supplier, Blackburn Electric Wires Pty. Ltd.
55 Garema Circuit, Kingsgrove, NSW 2208.
Ph. (02) 9750.3133 Fax. (02) 9759.0245
They do not appear to have a website but are VERY good to deal with
by mail order.

The original chart contained the same copper sizes as shown for grade 1
with less enamel thickness and grade 3 with more enamel thickness.
I only use grade 2 which is the only grade shown in the chart below.
Grade 2 is the main grade stocked by my supplier because it is the
industry norm for 99% of high temperature rated winding wire for
electric motors and stressful industrial applications. Trying to get
triple insulated winding wire is almost impossible so making copies
of McIntosh OPT is difficult.

The range of sizes shown are not all obtainable off the shelf, and to
get some sizes a wait for an order is involved, so I sometimes have to
design around the wire size available, which adds to the challenge.
Anyone not used to measuring in millimetres better start getting used
to metric because here the diameter measurement matters more than
the wire guage, and there are is AWG, SWG, BS, all very confusing,
and I don't have conversion charts so if you work in gauges and inches
and feet, provide your own solutions.
Before winding anything, make sure you have an accurate micrometer
to confirm that the size is correct.