Using toroidal power transformer in high-frequency application

[onemilimeter]

I wish to purchase the toroidal power transformer as shown in figure below. In my application, the transformer will be used to step up voltage. Assume that the turn-ratio is a = 220V/55V = 4.

The transformer, as we know, is designed for 50Hz application. However, I read somewhere that ferrite toroidal core can be used at higher frequencies, e.g. from few tens of kilohertz to hundreds of megahertz. I'm not sure the one I wish to purchase will work at 10kHz or not. The input voltage (to the 0-55V winding) is 25Vrms (10kHz) and hopefully the output (0-220V winding) of the transformer is able to supply at least 0.35A (rms) to a 250 ohm load.

Kindly share if you know any "side effect" when a power transformer (designed for 50Hz application) is used at 10kHz. Please discuss based on either Model-A or Model-B below. Thanks.

[Zad]

Impossible to know without full information about the performance of the core. Chances are the "Iron Losses" will go through the roof at that frequency, and it will rapidly overheat. Try doing some tests with an audio signal generator, a dummy load and an oscilloscope.

[Zero999]

Your best option is to buy a suitable powdered ferrite torrid core and wind your own which will be more efficient and lightweight than using a bulky and heavy transformer which is not designed.for it.

[mikeselectricstuff]

Eddy current losses increase with frequency, which is why ferrite is used instead of laminated iron at higher frequencies. 

[onemilimeter]

I will follow Hero999's suggestion...

I've been Googleing for a simple note on steps to design a transformer using a toroidal core for frequency higher than mains. I still could not locate one which is easy to understand for novice like me. Kindly share if you know any. Thanks.

[alm]

I don't think good transformer design is a particularly easy to understand subject, so it may be hard to find a novice-level text. There have been some references to good books on this forum some time ago, they may be helpful.

[Time]

Eddy current losses increase with frequency, which is why ferrite is used instead of laminated iron at higher frequencies. 

I thought you used laminated cores to counter eddy currents?

[onemilimeter]

I thought you used laminated cores to counter eddy currents?

I think laminated iron cores will be quite lossy at high frequency. Like Hero999 suggested, we should considered "powdered" iron core....

[onemilimeter]

Impossible to know without full information about the performance of the core. Chances are the "Iron Losses" will go through the roof at that frequency, and it will rapidly overheat. Try doing some tests with an audio signal generator, a dummy load and an oscilloscope.

If we know the full information about the toroidal core, is it possible to calculate all the parameters in Model A and Model B? Thanks.

[mkissin]

Impossible to know without full information about the performance of the core. Chances are the "Iron Losses" will go through the roof at that frequency, and it will rapidly overheat. Try doing some tests with an audio signal generator, a dummy load and an oscilloscope.

If we know the full information about the toroidal core, is it possible to calculate all the parameters in Model A and Model B? Thanks.

No, to find all those parameters you need to measure them from the actual transformer. They are determined by far more than just the core. Also, model A (The 'T' model) and B (The 'Cantilever' model) are the same, just with different representations. The 'T' model is more common.

Also, there are many different types of ferrite, and each is designed to operate in a specific band of frequencies. Powdered iron is something completely different and is generally used for inductors (chokes) rather than transformers, when energy storage is required. 10kHz would be OK for laminated iron (at the top end though), but ferrite would be much lower loss, and probably more appropriate.

Since you're using the transformer well under the maximum power rating, it'll probably work, but it'll also probably be rubbish (i.e. lossy, and with poor regulation).

HOWEVER... bear in mind that you seem to be using the transformer backwards in your appplication. Something for which it was not designed at all. If there are not enough turns on the secondary winding (your primary) then you may saturate the core, potentially destroying it and damaging the circuits connected to it.

To figure out if you're going to saturate it, you need to use the calculation:

Bmax= Vrms/(4.44*f*N*A)

This is only valid for sinusoidal signals, but it's close for others. If it's an iron core, you need to keep Bmax under about 1T, and if it's ferrite then 200uT is a good rule of thumb. You need to know N (the number of turns) and A (core area) which might be tough if the datasheet isn't really good.

[williefleete]

im picking at high frequencies the laminated core would act as a parasitic capacitor, the iron would act as plates and the enamel as the dielectric, capacitors act as a resistor at high frequencies and it would start to heat up i would imagine

[Psi]

Something i've always wanted to build is a transformer core efficiency tester.

You'd take a core you want to test (iron, powered iron, ferrite etc) and wind X turns of wire for a primary and again for a secondary then connect the 4 wires to the device.
A mcu would drive some power electronics to generate varying AC voltages at different frequencies on the primary winding. It would also measure the input and output power to calculate the efficiency so it could combine all this info and find the ideal frequency for best efficiency using the core being tested.

It would be useful to be able to test old transformer cores ive pulled out of things over the years so i could use them again.

[onemilimeter]

No, to find all those parameters you need to measure them from the actual transformer. They are determined by far more than just the core. Also, model A (The 'T' model) and B (The 'Cantilever' model) are the same, just with different representations. The 'T' model is more common.

If we have full information about the toroidal core, the number of primary and secondary winding turns, and AWG of the winding, can we calculate all the parameters in Model-A?

Thanks.

[onemilimeter]

To figure out if you're going to saturate it, you need to use the calculation:

Bmax= Vrms/(4.44*f*N*A)

This is only valid for sinusoidal signals, but it's close for others. If it's an iron core, you need to keep Bmax under about 1T, and if it's ferrite then 200uT is a good rule of thumb. You need to know N (the number of turns) and A (core area) which might be tough if the datasheet isn't really good.

In my application, the voltage and current waveforms will be in pure sinusoidal.

By the way, what're the units for f and A? Are they (Hz or kHz) and (cm^2 or m^2) respectively?

Thanks.

[Zero999]

Something i've always wanted to build is a transformer core efficiency tester.

You'd take a core you want to test (iron, powered iron, ferrite etc) and wind X turns of wire for a primary and again for a secondary then connect the 4 wires to the device.
A mcu would drive some power electronics to generate varying AC voltages at different frequencies on the primary winding. It would also measure the input and output power to calculate the efficiency so it could combine all this info and find the ideal frequency for best efficiency using the core being tested.

No need.

All you need to do is measure the resistance of the primary and secondary, measure the efficiency of the transformer, then you can work out the losses due to the windings (Ohm's law) and subtract them from the actual losses.

[mkissin]

If we have full information about the toroidal core, the number of primary and secondary winding turns, and AWG of the winding, can we calculate all the parameters in Model-A?

No. There are two ways to find those parameters....measure them, or simulate them using a really expensive piece of FEM software (ANSYS/ANSOFT or COMSOL are the usual ones, but there are others). Under no circumstances are you going to be able to calculate them, although an experienced magnetics designer may be able to ballpark them for you.

As for transformer efficiency, you really need to measure that at power because core losses are directly related to the voltage across the windings, and don't scale nicely if you move out of the linear region of the core materials BH curve. They also change with waveform (i.e are you driving this transformer sinusoidally, which is almost never, or with a full-bridge inverter, or a flyback, or whatever). The copper loss is, as stated, reasonably easy to get close as long as you measure at the operational frequency, but skin effect and proximity effects will still be triggered by harmonic content in your current waveforms so you'd ideally want to use the same drive waveforms.

The challenge posed by magnetics cannot be overstated

[Psi]

Something i've always wanted to build is a transformer core efficiency tester.

You'd take a core you want to test (iron, powered iron, ferrite etc) and wind X turns of wire for a primary and again for a secondary then connect the 4 wires to the device.
A mcu would drive some power electronics to generate varying AC voltages at different frequencies on the primary winding. It would also measure the input and output power to calculate the efficiency so it could combine all this info and find the ideal frequency for best efficiency using the core being tested.

No need.

All you need to do is measure the resistance of the primary and secondary, measure the efficiency of the transformer, then you can work out the losses due to the windings (Ohm's law) and subtract them from the actual losses.

The issue is that i have a lot of nice empty transformer cores around here in various sizes and shapes that could be quite useful if i knew what material they were made from and as such what frequecny they would work at.
90% of them have no markings at all.

[onemilimeter]

After reading some design notes, I obtain the following conclusions:

[1] The magnetising inductance (Xm=2*pi*Lm) should be large enough to limit the input current when the maximum input voltage is applied.

[2] The leakage inductance (Xp=2*pi*Lp, Xs=2*pi*Ls) should be made as small as possible to avoid loading effect due to leakage inductance especially at rated output current and 10kHz.

[3] To reduce leakage inductance, we should arrange the primary and secondary winding in such a way that the magnetic coupling (coupling coefficient) between them is near unity.

Can I assume that to design a transformer using a toroidal core the coupling coefficient will be very near to unity, and hence, the leakage inductance can be neglected?

Thanks.

[mkissin]

All your equations are missing the 'f' term that makes them frequency-dependant. Other than that, they are correct.

X = 2 * pi * f * L

That just gives the imedance (X) of any given inductance (L) at a frequency (f).

Just using a torroidal core doesn't make your leakage inductance negligible. You need to ensure that the turns you wind into the core are evenly spaced so that your primary coil uses the entire length of the core. Ditto for the secondary. You also need to ensure that the two coils are wound on top of each other (as they must be if you follow the first rule). Don't forget to add a layer of insulating material (electrical tape is generally OK, Kapton is better) to get your isolation voltage up.

[onemilimeter]

All your equations are missing the 'f' term that makes them frequency-dependant. Other than that, they are correct.

X = 2 * pi * f * L

That just gives the imedance (X) of any given inductance (L) at a frequency (f).

Just using a torroidal core doesn't make your leakage inductance negligible. You need to ensure that the turns you wind into the core are evenly spaced so that your primary coil uses the entire length of the core. Ditto for the secondary. You also need to ensure that the two coils are wound on top of each other (as they must be if you follow the first rule). Don't forget to add a layer of insulating material (electrical tape is generally OK, Kapton is better) to get your isolation voltage up.

Thanks for your advices. If I follow the approach you suggested to wind the primary and secondary winding on a toroidal core, do you think, based on your experience, will I obtain a magnetic coupling coefficient of more than 0.95?

Thanks.

[mkissin]

You should be able to get a coupling of around 0.95 without too much trouble, as long as you wind it carefully.