I had a power transformer and an output transformer that were both from old tube amps and which both had shorted windings. I was in need of a transformer for a bench power supply and also a multi-impedance output transformer for testing so I decided to have a go at doing a transformer rewind.

I did a lot of reading about how to do rewinds. The biggest issue that I was concerned about was how to dismantle the core if it had been vacuum impregnated with varnish. Both of the transformers I had were very obviously coated in varnish. I made a mistake here in assuming that meant they were varnish impregnated. Going on that assumption I looked up the best way to dissolve the old varnish. As with all things Internet, it is wise to read multiple articles. I found several people recommending paint thinner as a good solvent.

I followed the instructions and put my power transformer into a one gallon paint can (a new one from Lowes) and filled the can with paint thinner. I left it covered over outside for a week.

The paint thinner made almost no change to the varnish, but as I discovered later, it softened the varnish coating of the wire which made for a lot of mess later. So the first lesson: Don’t use paint thinner, and don’t assume the transformer is impregnated.

Some time later in the process I realized that the two transformers I had were dipped in varnish probably to prevent corrosion, ingress of moisture, and maybe a little stabilization of the laminations to prevent them from moving.

The first step in the disassembly is to remove the laminations from the bobbin. These are in there very tight and held in there by a small amount of surface corrosion in some cases, bits of varnish that crept into the cracks, and the same force that allows lizards to climb my shiny plastic fence – magic.

The best way I have found to remove the laminations without damaging them is to make a drift from a piece of hacksaw blade and use it to gently tap out the first few laminations while the transformer is held in a vice. Before I discovered the hacksaw blade idea I used a hardened steel chisel. This works too but you have to be very careful about the placement of the tip, and the tip needs to be pretty sharp. Once the first couple of E and I laminations have been removed the rest can be separated with a sharp knife before being pushed out. It takes quite a bit of time to remove all the laminations.

The laminations need to be kept somewhere safe so they won’t get scratched or otherwise damaged before the rebuild. If one or two get mangled during the disassembly it’s not going to matter as I found it very hard to get all of them back into the bobbin later. I always had one or two left over.

The laminations removed from the bobbin

The next step is to remove the old windings from the bobbin. For each winding I noted the size of the wire and how many turns there were. This is VERY tedious to do by hand for the high voltage windings as they have hundreds of turns. I counted the turns off in batches of 50. Then cut the wire, balled it up and put it on the bench. I was a bit worried about losing count. Thicker wire like that used for the heater secondaries is easy to get off. Some windings are bifilar wound. Just make a note of how they were connected if you plan to reproduce the winding.

DO NOT be tempted to reuse the old wire. Recycle it away or make art from it. As I got to the bottom of the fine windings the wire kept braking because its insulation had melted off (the cause of the original failure).

High voltage winding wire balls

Once the bobbins were empty I cleaned them up as best I could and removed any old wire fragments from the terminals. The the terminals were re-tinned to be sure they would solder OK with the new wire. Cleaning up the gunk left form the paint thinner dunking was pretty unpleasant.

An empty bobbin

My power transformer had two bobbins: one for the primaries, and one for the secondaries. The output transformer I rewound later only had a single bobbin.

After the experience of counting the wire off the high voltage secondaries I had serious doubts I could accurately rewind and count the turns. At my age remembering where I put my flip flops is enough of a challenge.

To avoid the dreaded “How many turns was that?” question, I built a turns counting rewind machine. I used an Arduino with an LCD display from Sparkfun and a couple of hall-effect magnetic sensors placed near a wooden rotor that has two bits of magnetic plastic (from my Eufy) taped to it.

The tall box was designed so I could see the display behind the rotor. To stop it falling over on the bench when I was clumsy I put a big chunk of wood in the bottom.

For the rewind I bought the wire I’d need from Remington. You can buy direct from them or from their store on Uncle Jeff’s website for just a bit more.

After rewinding the primary bobbin I did a quick test to make sure I hadn’t screwed up. I wound a few turns of wire (an exact number that I counted) on the secondary bobbin and then reassembled the whole thing for a bench test. Once I was happy that the primary turns were correct and the two 110V windings matched, I wound the secondaries.

For the secondaries I wanted to have two HT voltages (with a center tap) and a center-tapped 6.3V heater. I spent some time making sure I had the correct number of turns for each secondary segment. I color coded the terminals with acrylic paint and made a winding chart to be sure I did it right. There were not enough terminals on the bobbin for all my connections so I had to use tails for the ends of the heater winding. I used a common center tap post for both the HT and heater windings. The main secondary also had a tap for a bias voltage so in all the main secondary had 6 terminal posts.

Here is my winding list:

Winding order:

300 V red – 175 V or – 275 turns
175 V or – GND blk – 385 turns
GND blk – 40 V yel – 88 turns
40 V yel – 175 V or – 297 turns
175 V or – 300 V red – 275 turns
6.3 – GND bifilar 7 turns
GND – 6.3 bifilar 7 turns

Tab colors:

grn – 6.3
blk GND
red 300
or 175
yel 40

Procedure:
300V
Start at large red post
Wind 275 turns
End at large orange post
Tape whole winding
175V
Start at large orange post
wind 385 turns
end at large black post
Tape whole winding
40V
start at large black post
wind 88 turns
end at small yellow post
Tape whole winding
175V
start at small yellow post
wind 297 turns
end at small orange post
Tape whole winding
300V
start at small orange post
wind 275 turns
end at small red post
Apply two complete layers of tape to completely cover winding

6.3V
Bifilar wind two strands but laying down one at a time
Winding A
Connect end to green tail wire and sleeve with heatshrink
Tape wire in place on bobbin
Temp tape green wire to shaft
Wind 7 turns with one wire space between turns
Estimate wire required to terminate at large black post and cut
Tape temp to side of bobbin
Winding B
Strip end and attach to large black post
wind 7 turns laying down with previous winding and tightening both
tape windings at least some to hold in place
cut the end and solder to green tail. Sleeve with heatshrink
pull tail out from bobbin and temp tape to shaft.
Complete winding A by attaching loose end to large black post.
Tape final windings in place

The transformer as used in my PSU (partial schematic)

In retrospect I had no reason to use two 120 V primaries. I should have just used slightly thicker wire and had one.

Here are a couple of shots of the finished transformer before it went into my bench power supply.