FAQ: The high voltage transformer

If you have a question about this topic, the answer is probably in here!
Post Reply
User avatar
Finn Hammer
Posts: 298
Joined: Sat Mar 05, 2016 7:21 am
Real name: Finn Hammer
Contact:

FAQ: The high voltage transformer

Post by Finn Hammer »

I am trying to keep this FAQ very basic. Covering every aspect of transformer theory has created many shelf-meters of books. Creating a functioning transformer has been considered a black art, which it is to some extent, and there are many reasons for that point of view. For one, there are the different units of magnetism, a lot of the literature relies heavily on math, few sources want to reveal the simple fact that all you need is the basic transformer formula.
With that at hand, it all comes down to actually starting to wind some coils.
If winding these coils gives you a sense of satisfaction, you could get hooked, if not, just forget about it and go buy what you desire. This is not for everybody, but it has given me great satisfaction over a period of 25 years.

Acquiring a high voltage supply suitable for fusor duty is not the easiest of tasks, and that is why many fusioneers, new as well as seasoned, consider to build their own. The electronic part of the project can be solved in many different ways, but there is no substitute for the step-up transformer. Since the solution I will address here uses a switching power supply, the transformer will employ a ferrite core.
This chart describes the various sizes of ferrite cores available in the UY group, the UY22A is particularly well sized for a fusor supply, I have used it successfully myself to deliver up to 2.5kW into a fusor.

UY core dimentional chart
UY core dimentional chart

A most frequent question when it comes to transformers is this: "How many turns do I need" sometimes followed by: "to avoid saturating the core".
Let me address the issue of saturation first, with an example using an iron cored transformer operating at 50/60 Hz .
The transformer primary coil is designed to have an inductance which enables it to have its rated voltage applied to it, without any noticeable current passing through it.
This inductance is called the magnetizing inductance and it is very big, but not infinitely so, that is why a small current will pass through it, even when the transformer is unloaded, and this current is called the magnetizing current, because that is what it does: it magnetizes the core.
To produce this high primary inductance, there has to be enough turns on the primary coil, and this number is related to the core cross section so that a beefier core requires fewer turns to meet the desired magnetizing inductance. If these two parameters are balanced correctly, the magnetizing inductance will be right, and the core will be magnetized to +-1.4 Tesla or thereabouts.
Changing the magnetic polarity of the core back and forth 50/60 times per second produces heat in the core, due to the hysteresis of the core material, but in an iron core at 50/60Hz, this is not a problem.

It is worth to take note here, that at this point of the design process, the problem of saturation has been solved: enough primary coil turns and you are good to go. No matter how much power you draw from the transformer later on will not increase the flux level in the transformer core.

If there are too few turns in the primary coil, however, the magnetic flux reaches a level where the core goes into saturation, and this leads to catastrophic results, because this diminishes the inductance of the primary coil, and causes the current through it to increase, sometimes to destructive levels.
If you take a properly designed transformer rated for 110V and apply 110 Volts to it, you will see that it can remain connected to this 110V source indefinitely. If you apply 220V to it, you will see the smoke.
This example of saturating the transformer core was made with an iron cored transformer at 50/60hz because it is in this type of transformer topology we need to worry about saturation of the core.

In a modern so-called switching power supply, where the frequency is usually higher than 20kHz, the losses inside the core becomes the limiting factor. Since every reversal of the magnetic field will deposit a small amount of energy in the core, as a heat loss, it follows that the losses go up with frequency.
Experimental results have shown that a ferrite transformer core can be driven close to saturation up to 20kHz, without developing thermal issues, and this is described like this: “Below 20kHz the core is saturation limited”.
Above 20kHz, the core will overheat if it is driven close to saturation, due to the high internal losses, and this is described like this: Above 20kHz the core is core loss limited”.
Since ferrites generally saturate at around 0.6 Tesla, this means that at frequencies well above 20kHz, the peak magnetic flux must be kept lower than 0.6 Tesla. How much is ultimately up to a practical test, but going as low as 0.1 tesla above 50kHz is not an unlikely reality.
But does that at all matter? Not really, all it takes to lower the flux level is to put more turns in the coils.
More turns in the primary coil corresponds to fewer volts per turn, and this in turn requires more turns in the secondary coil to reach a desired output voltage.
Is that a problem? Not really. There is little reason to produce more than 10kV peak to peak out of the transformer, and why is that?, you may very well ask. For one, at frequencies above 20kHz, with a stage capacitance of around 20nF you can build a full wave CW voltage multiplier with 8-10 stages which doesn’t result in more than a couple of hundred volts of ripple, another reason is, that a +-5kV transformer can be constructed as a dry transformer without need for insulating oil.
Before I describe the design process, a few words of caution regarding frequency of the switcher:

Being amateurs, we may harbor desires to meet or even exceed the results that the professionals at Spellman, Glassman, Bertan and other well-known high voltage supply manufacturers.
My advice: Don’t.
These guys know every trick in the book, neither do you and I, so perhaps don’t waste your time trying to produce that 100kHz hot rod supply, because it will cost you an endless amount of diodes and worry about overheating ferrite cores. Ask me how I know.
From my own experience, I can attest to it as a fact, that keeping the switching frequency around up to 35kHz will produce a nice, rugged and benign switcher, able to run forever without thermal issues. And when the neutron count from the fusor gets on the increase, I can also promise you, the least you want to worry about is your power supply.

At this point in the FAQ and onwards, I will describe the design procedure for a ferrite cored high voltage transformer to be used with a switching supply, one that will be suitable for use with a fusor.
You cannot proceed without selecting the core, since this part will define the rest of the process.
The cores that you find and scrounge from color tv’s flyback transformer are not really suitable because of the narrow window, which places the high voltage end of the secondary too close to the core. This means that the resulting transformer will be risky to operate, and even under oil will be prone to failure.
One core that has shown it’s worth is designated UY22A. The legs of the core are 22mm in diameter, so the core area is 3.8cm^2
With this parameter on the table, it is possible to proceed to the calculation of the needed amount of windings on the primary coil.
To do this, use the standard transformer equation, which goes like this:
N(turns) = Vin*10/delta B*Ae*F*2
Where
Vin is the buss voltage for H-Bridge, 0.5*buss voltage for half bridge
Delta B stands for peak flux in Tesla
Ae stands for core cross-section on square centimeters
And F stands for switching frequency in kilohertz
Assuming a half bridge, a buss voltage of 340V, a delta flux of 0.32 Tesla, a core cross section of 3.8 cm^2 and a switching frequency of 35kHz, filling into the formula:
N(turns) = 170*10/0.32*3.8*35*2 = 19.97 turns.
In the practical world, this will amount to 20 turns, which corresponds to 8.5 volts per turn.
The secondary coil will develop the same 8.5 volts per turn, so if you want a 5kV transformer (10kVp-p) you have to wind 588 turns on that secondary. If you want to drive a full wave doubler (and you do!) you need 2 coils center tapped to drive each leg of the doubler with opposite polarity simultaneously.
To connect the transformer secondaries correctly to the voltage multiplier, follow this schematic:

Connection between high voltage transformer and voltage multiplier
Connection between high voltage transformer and voltage multiplier
Cheers, Finn Hammer
User avatar
Finn Hammer
Posts: 298
Joined: Sat Mar 05, 2016 7:21 am
Real name: Finn Hammer
Contact:

FAQ: Phasing the secondary windings

Post by Finn Hammer »

Phasing he secondary coils in the high voltage transformer.

This post is dealing with the proper phasing of the secondary coils.
To drive a full wave multiplier, the transformer has to have a center tapped winding, as shown in the previous installment of this FAQ. In the case of a coil with just one continuous layer, the center tap is obvious: It is a connection to the middle of the coil, which produces a voltage output relative to the center tap, which is negative in one end and positive in the other.
A single layer centertapped coil
A single layer centertapped coil

What is important to note here is that the coil as a full is wound in the same direction.
In a high voltage transformer, there are special requirements to where the potentials of the coil can be placed. Normally, you don’t want the high potential to be closest to the core, and that is why high voltage transformers often have pie wound coils. These coils can have the grounded connection close to the core and the high voltage terminal furthest away from the core.


A pie wound coil
A pie wound coil

It can be puzzling to try to imagine how to wind such coils, should they be 2 identical coils, or should the one of them be wound in opposite direction?

If you look at this first drawing of 2 identical coils with 3 turns each, positioned alongside each other, and centertapped, you will see that this arrangement satisfies 2 of the requirements. The low voltage end is closest to the core and the high voltage is furthest away. But you will also see that the winding direction is effectly reversed, producing the same polarity on the ends of the coils.

non-flipped but identical secondaries
non-flipped but identical secondaries

In this next drawing, one of the coils have been flipped 180 degrees, and now the winding direction is the same so the center tap is sitting at the middle of the full output, producing a +- voltage relative to this center tap.

Secondaries, one flipped 180deg.
Secondaries, one flipped 180deg.

Winding pie coils is not for everybody, instead a compartmentalized layout is suitable if you have a lathe or 3d printer. These coils look like this one

20220605_120449.jpg

You will need 2 of these coils for a center tapped high voltage transformer, and just like the pie wound counterpart, the coils must be identical and one of them must be flipped 180deg, to form the center tap in between the coils, so that the highest potential windings are furthest apart.


You have now heard, that in a center tapped high voltage transformer, the two coils are identical, but the one of them is flipped 180 deg. The center tap is created by joining the bottom of the two coils.
If you actually wind the coils, place them on a core, and excite it, you will know it!.

Please post comments in this thread in the high voltageforum viewtopic.php?t=14508

Cheers, Finn Hammer
Post Reply

Return to “FAQs: High Voltage”