Converting an Autotransformer into an Isolation Transformer

[WattsThat]

Here is a good link on DIY transformers

https://ludens.cl/Electron/trafos/trafos.html

[dmmartindale]

The primary winding consists of one-third 3mm wire and two-thirds 1.75mm wire. I’ve calculated that I need 230 turns for the primary and 240 turns for the secondary.

Do you understand why the winding uses two different wire sizes? The 3 mm wire can carry almost 3 times the current of the 1.75 mm wire. The 3 mm wire will be for the section of the winding that has the taps attached to it, since when the autotransformer is being used to either increase or reduce voltage, the portion of the winding between the input and output taps carries the full load current. That's what the 3 mm wire is sized to carry. The lower-voltage portion of the winding (below all of the taps) carries much less current in an autotransformer. That's why the transformer can use smaller wire.

For example, when an autotransformer is providing a 10% voltage boost, 91% of the input current passes directly to the output tap, going through only the portion of the winding between the two taps. Only the remaining 9% of the current flows through the portion of the winding between input tap and the neutral end of the winding. (Plus the whole winding has to handle the magnetizing current, but that's probably only a few percent of full load).

So your autotransformer has 3 mm wire for the portion of the winding that may have to handle the full input or output current, but smaller wire in the portion of the winding that handles much less current. An isolation transformer of the same VA rating needs two full windings using 3 mm wire for the whole winding, since the whole primary and secondary carry the full input or output current. That's going to need about 3.6 times the size of core window compared to the current mixture of 1.75 and 3 mm wire.

As for number of turns: How many turns are on the transformer winding now? This is the number of turns that is needed to keep the core out of saturation when operating with nominal line voltage applied to the transformer. That's the number of turns your new primary and secondary *each* need.

[ElectronSurf]

The primary winding consists of one-third 3mm wire and two-thirds 1.75mm wire. I’ve calculated that I need 230 turns for the primary and 240 turns for the secondary.

Do you understand why the winding uses two different wire sizes? The 3 mm wire can carry almost 3 times the current of the 1.75 mm wire. The 3 mm wire will be for the section of the winding that has the taps attached to it, since when the autotransformer is being used to either increase or reduce voltage, the portion of the winding between the input and output taps carries the full load current. That's what the 3 mm wire is sized to carry. The lower-voltage portion of the winding (below all of the taps) carries much less current in an autotransformer. That's why the transformer can use smaller wire.

For example, when an autotransformer is providing a 10% voltage boost, 91% of the input current passes directly to the output tap, going through only the portion of the winding between the two taps. Only the remaining 9% of the current flows through the portion of the winding between input tap and the neutral end of the winding. (Plus the whole winding has to handle the magnetizing current, but that's probably only a few percent of full load).

So your autotransformer has 3 mm wire for the portion of the winding that may have to handle the full input or output current, but smaller wire in the portion of the winding that handles much less current. An isolation transformer of the same VA rating needs two full windings using 3 mm wire for the whole winding, since the whole primary and secondary carry the full input or output current. That's going to need about 3.6 times the size of core window compared to the current mixture of 1.75 and 3 mm wire.

As for number of turns: How many turns are on the transformer winding now? This is the number of turns that is needed to keep the core out of saturation when operating with nominal line voltage applied to the transformer. That's the number of turns your new primary and secondary *each* need.

Yes, I'm familiar with the "How It Works" section. I just never wound or designed a transformer before. I did unwind the 3mm wire part, so I only have the 1.75mm wire left on the core. Since I want to use aluminum wire (accepting the downsides), I thought that if the core only has space for 2 mm wire, I could take advantage of the already existing primary wire (on the new bobbin that I'm designing). That would at least help with thermal performance while still having more conductivity than 2 mm aluminum wire.

I understand the concerns about this project, and I appreciate people pointing them out. The goal here is to make this work. Am I going to have less KVA? So be it. A less-than-perfect transformer? Tough luck! This is something I started, and I’m going to finish it with the help of users on this forum, hopefully.

[Gyro]

Some UPSs transformers have twin 220V or twin 110V windings and can be used as-is for an isolating transformer.
However not all UPSs are designed that way, so you need to find one that does and has the two windings. 
Usually one of the windings has some taps on it, but that's fine.
 
That's how I isolate my variac when needed. I put one of those UPS transformers into a metal box to make an isolating transformer I can insert when needed.

Unfortunately using twin windings on the 'primary' side of a transformer don't meet the requirements of a safety isolation transformer. They will most likely only be insulated from each other by the thickness of the wire lacquer.

When you build an isolating transformer (floating secondary) you are creating a Class II insulation situation where the secondary side must be isolated from the primary (and the grounded core and casing) by Double or Reinforced insulation. Primary and secondary wound together don't meet this requirement.

@ElectronSurf needs to be very aware of this too. No skimping allowed. In European countries (not sure about Antarctica ), EN60950-1 requires 3kV AC withstand voltage. The US will have similar requirements.

[Psi]

Some UPSs transformers have twin 220V or twin 110V windings and can be used as-is for an isolating transformer.
However not all UPSs are designed that way, so you need to find one that does and has the two windings. 
Usually one of the windings has some taps on it, but that's fine.
 
That's how I isolate my variac when needed. I put one of those UPS transformers into a metal box to make an isolating transformer I can insert when needed.

Unfortunately using twin windings on the 'primary' side of a transformer don't meet the requirements of a safety isolation transformer. They will most likely only be insulated from each other by the thickness of the wire lacquer.

Pretty sure they are 1 primary winding and 1 secondary winding with 2 taps.   
The primary winding is the mains input and the secondary is the output winding with relays that switch between 5% step-up, normal or 5% step-down to keep the output voltage in spec even if input is out of spec. (plus another primary winding at much lower voltage for AC battery input).
But yes, i have no idea what the insulation is like between them.
I only use it for DIY stuff.

[ElectronSurf]

I laser-cut the bobbin design, but I don’t know what I was thinking when I didn’t make the top and bottom as a single piece! I spoke to the guy who does the core winding and related work, and he told me the weak point isn’t the body, since it will be supported by the center leg of the core. The weak part is actually the top and bottom pieces, because the wound wire could slip and push against them, possibly making them pop out. I think he’s right, this design needs a revision.

[Gyro]

Yes it's the unsupported bobbin cheeks that are prone to failure. With a brittle material, the joints between the base and cheeks could crack, opening up a potential leakage path. The risk of cheek breakage is particularly high during winding if they are not clamped.

You also need to (slightly) chamfer the the corners of the base of the bobbin to prevent them from kinking the wire and damaging the lacquer.

[MrAl]

I recently acquired a big 1kVA autotransformer, one of those commonly used as a "voltage stabilizer." I would like to rewind it so I can use it as an isolation transformer in my lab. I have zero experience with transformer winding.

My goal is to achieve 220VAC, 50Hz output at around 5A. Five amps is enough, though more would be better.

The core is an EI192 type, measuring 192x160x80mm. The center leg is 64x80mm. The existing winding has four taps, each adding about 15VAC.

I also found an online calculator to help with winding calculations.

My questions are:

1. Help? in general?

2. Can I keep the primary winding and simply add a secondary over it, or should I completely rewind both primary and secondary?

3. I’d like to use aluminum wire instead of copper to reduce cost. Is that acceptable?

4. Should I add a non-continuous copper shield between the two windings (with proper insulation) to reduce capacitive coupling?

5. How can I determine the magnetic flux density of this transformer? Should I assume mild steel at around 0.8 T?

Hi there,

The saturation flux density for most power transformers is 20kG which is 2T.  The design target though is usually 1.5T or less.
This is tested with a primary winding energized with a variac so you can turn up the voltage and watch the no load current or power.  The power should be 25 watts or less for a 500 watt transformer.  You might have a preference for a little more or a little less.
The calculation is:
Bmax=E/(4.44*F*N*Ac)
where
Bmax max flux density in Teslas,
E is the applied AC RMS sine voltage,
F is the frequency in Hertz,
N is the number of turns,
Ac is the area of the core in square meters.
If the primary meets that target and it tests ok, then the primary winding is good enough.  If not, add more turns and test again.
Also calculate voltage drop with the intended load current and wire diameter.  You don't want too much of a voltage drop but some is ok.

Aluminum wire would be used in low cost designs or lighter weight designs if it is possible.
For a one-off design, don't waste your time with aluminum wire.

You don't need a bobbin if you use layers of tape or fish paper and tape.  Use corners on the sharp corners, which can be fish paper (2000v/mil insulation).

You need enough insulation between primary and secondary to meet the isolation voltage requirement.  That would come from the type of tape and how many layers, and the layers should have overlapped edges.

You would need a foil separator between primary and secondary connected to ground if you expect high frequency noise or a noise sensitive load.

[Xena E]

[...] The risk of cheek breakage is particularly high during winding if they are not clamped. [...]

Always a good idea to take the time to make a mandrel to support the end cheeks fully while winding. Even if they don't break, if they spread at the outer edges it would likely stop the transformer being reassembled correctly.

[UnijunctionTransistor]

Please tell us the formula you have used to calculate the number of primary turns.

The link is in the very first post.

Thanks, I had missed that. I only saw the one which has the tables for the core data.

[ElectronSurf]

Hi there,

The saturation flux density for most power transformers is 20kG which is 2T.  The design target though is usually 1.5T or less.
This is tested with a primary winding energized with a variac so you can turn up the voltage and watch the no load current or power.  The power should be 25 watts or less.  You might have a preference for a little more or a little less.
The calculation is:
Bmax=E/(4.44*F*N*Ac)
where
Bmax max flux density in Teslas,
E is the applied AC RMS sine voltage,
F is the frequency in Hertz,
N is the number of turns,
Ac is the area of the core in square meters.
If the primary meets that target and it tests ok, then the primary winding is good enough.  If not, add more turns and test again.
Also calculate voltage drop with the intended load current and wire diameter.  You don't want too much of a voltage drop but some is ok.

Aluminum wire would be used in low cost designs or lighter weight designs if it is possible.
For a one-off design, don't waste your time with aluminum wire.

You don't need a bobbin if you use layers of tape or fish paper and tape.  Use corners on the sharp corners, which can be fish paper (2000v/mil insulation).

You need enough insulation between primary and secondary to meet the isolation voltage requirement.  That would come from the type of tape and how many layers, and the layers should have overlapped edges.

You would need a foil separator between primary and secondary connected to ground if you expect high frequency noise or a noise sensitive load.

Very informative, thank you.
One question though; since the sheet between the primary and secondary can’t form a complete turn, does that mean the transformer will still have some capacitive coupling, and the copper sheet simply reduces it?

[UnijunctionTransistor]

What you have to do with the shield, is you wind it a one full turn plus some overlap. A few mm of overlap is all you require.
Then, to prevent this turn from shorting, you thoroughly have to insulate this overlap. When I worked for Universal Transformer in Indiana, we would use kapton tape as the insulation. Kapton tape is a very tough tape with excellent voltage withstanding capabilities.

[johansen]

One question though; since the sheet between the primary and secondary can’t form a complete turn, does that mean the transformer will still have some capacitive coupling, and the copper sheet simply reduces it?

well both coils will have several perhaps dozens of picofarads of capacitance to the shield, but if the shield is grounded to the core then the transfer capacitance from one coil to the other can be theoretically zero. it won't effectively be zero because you can't perfectly ground the core and the shield, because your ground wire will have some resistance and inductance in real life. so you'll still have some voltage on the shield that is capacitively coupled into the secondary.

can you get me the relevant core dimensions of your transformer?

if you search this forum for "transformer design spreadsheet" you'll find mine. its a little kludgy but it works, and will calculate the rate of temperature rise due to the heat produced to give you a mental perspective of how hot it will get. It does not calculate leakage inductance, so your real life voltage drop will be a few percent higher than predicted. It uses lookup tables in conjunction with the percentage winding factor to tell you what size wire will fit in the core. it can handle bobbin thicknesses less than zero so you have to check your work...

as others have said your 1kva auto transformer is not actually a 1 kva transformer. it may be limited to 300watts.

while you can go as high as 1.7T you'll have a lot of waste heat at no load, but the power handling capacity of a transformer scales with the square of the flux density. so its critical to get it right and know your thermal limits.

[ElectronSurf]

What you have to do with the shield, is you wind it a one full turn plus some overlap. A few mm of overlap is all you require.
Then, to prevent this turn from shorting, you thoroughly have to insulate this overlap. When I worked for Universal Transformer in Indiana, we would use kapton tape as the insulation. Kapton tape is a very tough tape with excellent voltage withstanding capabilities.

That actually makes sense. I thought the shield was supposed to look like a C, only hugging the primary. Thanks.

well both coils will have several perhaps dozens of picofarads of capacitance to the shield, but if the shield is grounded to the core then the transfer capacitance from one coil to the other can be theoretically zero. it won't effectively be zero because you can't perfectly ground the core and the shield, because your ground wire will have some resistance and inductance in real life. so you'll still have some voltage on the shield that is capacitively coupled into the secondary.

can you get me the relevant core dimensions of your transformer?

if you search this forum for "transformer design spreadsheet" you'll find mine. its a little kludgy but it works, and will calculate the rate of temperature rise due to the heat produced to give you a mental perspective of how hot it will get. It does not calculate leakage inductance, so your real life voltage drop will be a few percent higher than predicted. It uses lookup tables in conjunction with the percentage winding factor to tell you what size wire will fit in the core. it can handle bobbin thicknesses less than zero so you have to check your work...

as others have said your 1kva auto transformer is not actually a 1 kva transformer. it may be limited to 300watts.

while you can go as high as 1.7T you'll have a lot of waste heat at no load, but the power handling capacity of a transformer scales with the square of the flux density. so its critical to get it right and know your thermal limits.

Dimensions are 192x160x80mm. The center leg is 64x80mm with 95mm of height. I don’t want to go higher, but rather lower, to reduce the no-load current and reduce core losses. I’ve chosen a magnetic flux density of 0.8T.


Redesigned the core; the top and bottom are now a single piece. I glued it with chloroform.

[MrAl]

Hi there,

The saturation flux density for most power transformers is 20kG which is 2T.  The design target though is usually 1.5T or less.
This is tested with a primary winding energized with a variac so you can turn up the voltage and watch the no load current or power.  The power should be 25 watts or less.  You might have a preference for a little more or a little less.
The calculation is:
Bmax=E/(4.44*F*N*Ac)
where
Bmax max flux density in Teslas,
E is the applied AC RMS sine voltage,
F is the frequency in Hertz,
N is the number of turns,
Ac is the area of the core in square meters.
If the primary meets that target and it tests ok, then the primary winding is good enough.  If not, add more turns and test again.
Also calculate voltage drop with the intended load current and wire diameter.  You don't want too much of a voltage drop but some is ok.

Aluminum wire would be used in low cost designs or lighter weight designs if it is possible.
For a one-off design, don't waste your time with aluminum wire.

You don't need a bobbin if you use layers of tape or fish paper and tape.  Use corners on the sharp corners, which can be fish paper (2000v/mil insulation).

You need enough insulation between primary and secondary to meet the isolation voltage requirement.  That would come from the type of tape and how many layers, and the layers should have overlapped edges.

You would need a foil separator between primary and secondary connected to ground if you expect high frequency noise or a noise sensitive load.

Very informative, thank you.
One question though; since the sheet between the primary and secondary can’t form a complete turn, does that mean the transformer will still have some capacitive coupling, and the copper sheet simply reduces it?

Hello again,

I meant to say 25 watts max primary no load power for a 500 watt transformer.  I updated my previous post.

As to the copper foil between primary and secondary windings, the physics is fairly simple.
Without the shield, the primary winding forms one plate of a capacitor with a dielectric, and the secondary winding forms the other plate.  That means we have an equivalent capacitor from the primary winding to the secondary winding.

When we insert the shield, we end up with the primary being one plate of the first capacitor and the shield being the other plate, and the secondary being one plate of a second capacitor and the shield being the other plate.  Thus, we end up with two equivalent capacitors in series from the primary to the secondary.  So far not that much has changed, but then we ground the shield, which means the center tap of the two capacitors is now connected to the ground for the primary winding.  That means noise from the primary is shunted to ground, and it's harder for noise to get into the secondary.

The tradeoff is secondary AC leakage to ground due to the equivalent secondary capacitor that connects to ground on one side and the secondary on the other side.  If there is a limit (such as in medical equipment) you may have to investigate that.  For a power transformer it's usually acceptable.  Testing a new design is always mandatory.

Since the shield acts as a single winding turn, the ends cannot be shorted together just like any other winding turn(s).  They have to either be separated by a small amount or insulated.  The lead from the shield should be a single broad strap rather than a single wire, and that should run out such that it ends up as close to the ground connection as possible when the transformer is installed.

[johansen]

Dimensions are 192x160x80mm. The center leg is 64x80mm with 95mm of height.

so after taking out 3mm on all sides for the bobbin and 1mm of space between two coils for isolation, you have enough room in theory for 220 turns (240vac at 60hz and .8T) of 13 gauge wire at 58% fill factor.

or 110 turns of 9 gauge wire.

your transformer core weighs 15.4 kilograms, is that correct?

with some work you can fit 220 turns of 12 gauge in there, or go for a pair of 15 gauge wire, the fill factor will be 80%


for the secondary if you want a 120vac secondary you can so two parallel 12 gauge wires, or 4 parallel 15 gauge wires.

you are looking at 5.3 kilograms of copper.

[ElectronSurf]

Dimensions are 192x160x80mm. The center leg is 64x80mm with 95mm of height.

so after taking out 3mm on all sides for the bobbin and 1mm of space between two coils for isolation, you have enough room in theory for 220 turns (240vac at 60hz and .8T) of 13 gauge wire at 58% fill factor.

or 110 turns of 9 gauge wire.

your transformer core weighs 15.4 kilograms, is that correct?

with some work you can fit 220 turns of 12 gauge in there, or go for a pair of 15 gauge wire, the fill factor will be 80%


for the secondary if you want a 120vac secondary you can so two parallel 12 gauge wires, or 4 parallel 15 gauge wires.

you are looking at 5.3 kilograms of copper.

I haven’t weighed the core, but it’s definitely above 10kg. The transformer guy told me that with 1.75mm wire for the primary, I might be able to fit 2.5mm wire for the secondary and GPT tells me the same. If I use 2mm wire for the secondary, it should be an easy fit.

And yes, it’s about 6kg of copper, which will cost around 100$. I can buy a ready-made toroidal isolation transformer for about the same price, which defeats the purpose of a DIY solution. That’s why I want to reuse the primary wire and use aluminum wire for the secondary.

[Gyro]

Give some thought to how you are going to bring out the primary and screen terminations while maintaining safety clearances from the secondary. Bringing them up the edges of the secondary isn't an option without double or reinforced insulation (as well as being a pain for your secondary winding process)*.

Also consider how you are going to terminate the Aluminium wire, it won't solder and is soft, so tends to cold flow if crimped or put in screw connectors (screws loosen).

Better to think about this stuff now rather than when you're half way through.


Edit: *Drilling holes out through the cheeks of the bobbin is probably the safest option.

[ElectronSurf]

Give some thought to how you are going to bring out the primary and screen terminations while maintaining safety clearances from the secondary. Bringing them up the edges of the secondary isn't an option without double or reinforced insulation (as well as being a pain for your secondary winding process)*.

Also consider how you are going to terminate the Aluminium wire, it won't solder and is soft, so tends to cold flow if crimped or put in screw connectors (screws loosen).

Better to think about this stuff now rather than when you're half way through.


Edit: *Drilling holes out through the cheeks of the bobbin is probably the safest option.

I’m not going to bring the primary lacquered wire out directly, instead I’ll solder it to a stranded insulated wire inside a fiber glass sleeve, which should provide proper insulation.

Yeah, aluminum tends to form a non-conductive oxide layer. I’ll use brass screw terminal and then fill it with Vaseline to keep the moisture out?

[Xena E]

I've seen transformers with aluminia wire successfully use a form of terminal similar to the Wago connector: it maintains grip on the wire despite the tendancy to cold flow. Good idea to provide an oxygen barrier with some vas or sil grease as well.

[johansen]

Yeah, aluminum tends to form a non-conductive oxide layer. I’ll use brass screw terminal and then fill it with Vaseline to keep the moisture out?

aluminum is now soldered commertially on a handful of products such as the common mode choke in inverter driven microwaves, as well as transformers in certain cheap APC backups. many cheap microwave oven have spring terminals on the bobbin that the aluminum wire is pressed into.

with some effort, you can scrub using a shitty soldering iron tip, an aluminum wire and solder it. i am pretty sure i did this with 99% tin solder years ago, was able to solder a to-220 mosfet directly to an aluminum heatsink, but it looks like zinc tin alloy solder made for aluminum works better.

anyhow i wouldn't worry about it. just run your aluminum wire directly into a receptacle, circuit breaker or contactor terminal strip rated for aluminum.  there will be enough "spring" in the contacts to keep the connection good for a decade.

[ElectronSurf]

I brought the bobbin and wires to someone who had a winding machine to get it wound. He asked for about $80 just to use the machine! These mofos make me giggle.

So after showing him the middle finger, I decided to build my own winding machine using a drill and a counter with a Hall sensor and an ATMEGA MCU. Hopefully, this will last me a lifetime whenever I want to wind coils in the future.

The idea is to build something like the picture attached. I’ll update this thread once I make some progress.

[UnijunctionTransistor]

I have wound transformers using a drill myself. But you really require to have the wire tensioned for a tight winding. A rough and simple way is to clamp the wire between a couple of teflon strips. You adjust the clamp’s pressure to obtain the desired tension.

[johansen]

I have wound transformers using a drill myself. But you really require to have the wire tensioned for a tight winding. A rough and simple way is to clamp the wire between a couple of teflon strips. You adjust the clamp’s pressure to obtain the desired tension.

for only 220 turns or so, do it by hand @OP

[ElectronSurf]

I wrote a script (with help from AI) based on this calculator for PowerShell to help calculate EI transformer windings. Simply right-click the script and select 'Run with PowerShell'.

This is V1. I might add calculations for wire length and weight, no-load primary current, and more in the future.