HV transformer isolation

[Neukyhm]

Hi there, I have built a 5.5kV transformer using AWG30 wire. I used epoxy resin to fix the windings and isolate the transformer.

I applied the resin to the secondary using a brush. The problem I have is that the curing time is very much and the resin will slide and fall from the transformer. Heating it won't help because the heat will speed up the curing process but the resin will become less viscous and slide anyway.

The question: is there a better sealant/adhesive/resin to use?

[Zero999]

Try applying the resin in more than one layer, so it's not thick enough to drip off.

Note: this is not an easy thing to do. Making the insulation between the windings good enough to hold off 5k5V, is not easy. Take care to ensure the voltage between each layer is no more than a few hundred volts and keep the wire clean and dry. If you wind by hand, then use cotton gloves.

[T3sl4co1l]

Place in a vacuum chamber, submerge in a pot of resin or varnish, pull vacuum, wait to stop bubbling, release vacuum.  (You may need a long cure resin (hours, or thermally activated) to do this.)  If there's time, cycle vacuum once or twice more just to be sure.  A vacuum pump is desirable here, not just a venturi kind (which is hard pressed to do better than 1/10th atmospheric).

Painting it on the outside isn't going to do very much (assuming you need it at all).  Or if you've used enough tape to deal with it as-is, it'll be fine -- at least in the short term, until corona and partial discharge eats it up.

If that's your concern, then all the more reason a vacuum is required, to fill the voids that PD occurs in.

Tim

[David Hess]

I agree with T3sl4co1l; pot it under a vacuum.  For an air wired voltage multiplier, I might accept brushing on corona dope and then potting in wax under atmospheric pressure but not for a high voltage transformer.

T3sl4co1l, what do you think about applying a couple atmospheres of pressure after vacuum potting?

[GerryBags]

Also, it might be worth opting for one of the low-viscosity, infusion grade epoxies. They are used for making carbon fiber items, laying up the fiber in a plastic bag in a mold, then sucking the air out of the bag and letting this thin resin in. Because of the low viscosity it is able to penetrate the weave of the carbon fiber, under vacuum, and make a solid matrix with no trapped air. This would make it ideal for getting right into the windings of a transformer. I think the temp requirements for most CF layup are generally more stringent than potting resins, but you'd have to check that.

[Jeroen3]

applying a couple atmospheres of pressure after vacuum potting?

That’s is called vacuum impregnating and the process isn’t easy.

[David Hess]

applying a couple atmospheres of pressure after vacuum potting?

That’s is called vacuum impregnating and the process isn’t easy.

What is the difficulty?  Use a pressure vessel instead of a vacuum chamber, apply a vacuum, release the vacuum, apply pressure, release pressure, and then remove the item from the bath of potting material.

[T3sl4co1l]

Pressure would have to be maintained until the material cures.  Which would work for resin, just heat the pressure vessel to activate, and let it sit until solid.  I guess there might be some oddness about heating a pressure vessel, just make sure it's good for it.

Then you have trapped bubbles under pressure, which will eventually diffuse back to ~atmospheric depending on material properties, though that could take very long indeed.  Well, at least if they're under pressure, they'll have higher breakdown voltage, which is a good thing.  Although they'll be smaller too, but, that means more E-field, which still works out.  Preferable to remove voids of course, but una-void-able (ha) ones under pressure is better than not.

It wouldn't work so well for varnish, where the pressure will reduce evaporation rate, and that's if you have ventilation in the chamber at all (say, a cold trap off to the side?).

Tim

[Jeroen3]

You evacuate the chamber before submerging it in resin. Then you apply pressure. And remove left over resin to prevent curing it as a block.

[fcb]

As others have said, vacuum impregnation of a resin will help.  But transformer design is also very important, if you are just randomly adding wire to the bobbin, then you risk having a potential between adjacent wires that exceeds the breakdown voltage of the wire enamel (in reality polyurethane or polyester), no amount/grade of resin is going to stop that happening.

You're probably close to the voltage where I would vacuum pot the whole transformer in RTV silicone or similar - it doesn't sound very big, probably just use a small potting box.

[jbb]

Sorry, should have said earlier: if you've got a large solid mass of epoxy over ferrite, you can get cracking of the ferrite due to CTE mismatch (Coefficient of Thermal Expansion).  Epoxy on the windings and bobbin is probably fine for CTE.

[NiHaoMike]

Sorry, should have said earlier: if you've got a large solid mass of epoxy over ferrite, you can get cracking of the ferrite due to CTE mismatch (Coefficient of Thermal Expansion).  Epoxy on the windings and bobbin is probably fine for CTE.

That's one reason to use gel type potting compounds. Not sure how they compare to the solid type for dielectric strength.

[David Hess]

Then you have trapped bubbles under pressure, which will eventually diffuse back to ~atmospheric depending on material properties, though that could take very long indeed.

Trapped bubbles of what?  Trapped bubbles of the remaining air from when vacuum was initially applied?  If so, then the same bubbles would exist from only using a vacuum.

I thought there might be some advantage if atmospheric pressure was not enough by itself to push the resin into the areas under vacuum inside the assembly.

[Decapitator]

There's a wealth of practical information about building transformers (and many other electronic topics) on Manfred Mornhinweg's ludens.cl website.

https://ludens.cl/Electron/Electron.html
https://ludens.cl/Electron/trafos/trafos.html
https://ludens.cl/Electron/tek310/tek310.html

[T3sl4co1l]

Trapped bubbles of what?  Trapped bubbles of the remaining air from when vacuum was initially applied?  If so, then the same bubbles would exist from only using a vacuum.

I thought there might be some advantage if atmospheric pressure was not enough by itself to push the resin into the areas under vacuum inside the assembly.

Yes -- there will always be bubbles, because there's no perfect vacuum, nor perfectly nonvolatile resin (though a separate degassing step might help, also, I suppose the components could be degassed with heating to remove volatiles efficiently?), or perfectly permeable assembly (how much gas will exit in the time given, how much resin will flow in?).

The question is, is it good enough?  For sure, using vacuum at all is a big start, and the bubbles should be very small indeed after pressurization (reduced by whatever the pressure ratio is).  I wonder if they might even disappear, because of surface tension and solubility.

It's probably down in the realm of empirical results at this point: how good is good?  Do a burn-in test for different conditions, and there you go.  Of course, that only helps if you need a known lifetime, for a production run.  One-offs?

Tim

[David Hess]

Yes -- there will always be bubbles, because there's no perfect vacuum, nor perfectly nonvolatile resin (though a separate degassing step might help, also, I suppose the components could be degassed with heating to remove volatiles efficiently?), or perfectly permeable assembly (how much gas will exit in the time given, how much resin will flow in?).

We did this when potting the cavities in load cells where the strain gauges are mounted although I an not sure what the justification was.  Degassing the epoxy gel after mixing produced crystal clear instead of a fogged translucent result but I think we did the same thing for non-transparent epoxies implying that it was for more than aesthetics.

The question is, is it good enough?  For sure, using vacuum at all is a big start, and the bubbles should be very small indeed after pressurization (reduced by whatever the pressure ratio is).  I wonder if they might even disappear, because of surface tension and solubility.

I suspect if you did it right that the gas remaining after a vacuum is applied would completely dissolve in the degassed epoxy.  Most of the advantage however comes from applying a vacuum since the ratio of pressures (gas volumes) will be much greater.  I was more concerned with getting the epoxy to completely displace as much of the free space as possible.

It's probably down in the realm of empirical results at this point: how good is good?  Do a burn-in test for different conditions, and there you go.  Of course, that only helps if you need a known lifetime, for a production run.  One-offs?

I have been anticipating very small production runs, accelerated testing at higher than operating voltage under worst case conditions, and destructive testing on samples in lieu of a long term testing program.  The last time I did destructive testing like this I could not get the assembly to fail but I am sure that would not be a problem when high voltages are involved.

[jmelson]

The transformer died almost instantly with a visible arc inside the epoxy 

I have to find a better insulator. I got that epoxy from RS Components, so it is supposed to be high quality, but it was very difficult to apply it to a ETD59 coil former, it had lots of bubbles.

Notice the core I'm using: ETD59.

Did you have layers of insulating material between winding layers?  This is standard practice.  You wind a layer on the coil form, then lay down a layer of insulating paper.  Then, wind the next layer of windings, and repeat.  This creates an additional amount of insulation between each winding layer.  The insulating paper is longer that the circumference around the winding, so the ends of the paper overlap.

Jon

[T3sl4co1l]

Hmm, "paper" sounds very old fashioned, but then I remember, there are plastic fiber products (usually polypropylene felt, I think?) intended for transformer winding -- the fiber allows resin in.

Just don't forget that this exists, and fall back on ordinary (cellulose) paper -- those days are long over!

Tim

[fcb]

Not all enamelled wire is the same (!)

These guys make the dogs-danglies of transformer wire: http://www.tex-e.com/

[jbb]

If higher leakage inductance is OK, a special bobbin with a sectioned winding (I.e. split the winding into multiple smaller lumps) might help.

[Gyro]

Is that ptfe plumbers' thread sealing tape that I see?

[T3sl4co1l]

PTFE tape has nil thickness to begin with, and is porous and cold-flows under pressure.  In short, it's useless here!

Actually using masking tape isn't a showstopper here -- as long as you only need class A insulation, and take adequate precautions to deal with the possibility of it catching fire (it's not exactly 94V-0 grade!).

"Electrical tape" is definitely not the thing here, as it is also very soft, especially when hot.  Or duct tape, or anything else rather on the gooey side.

It's well worth getting some legit yellow transformer tape!  It's only $20, and a small roll will supply many transformers.  I forget which Scotch part number is usual, but any moderate temperature (105C+), polyester tape will do.  The adhesive is usually acrylic based.

It goes on kind of loose, not the stickiest thing, but the adhesive softens on heating -- usually when the assembly is impregnated with varnish, then baked to cure it.  But even if you've just got solid layers of polyester film separating the layers, without resin, you'll stand a much better chance of success!

Tim

[fcb]

We use AT4004.

[james_s]

A trick that is used in some HV transformers is to use a bobbin that splits the winding into multiple side by side windings that are in series. When dealing with high frequency it becomes even harder to prevent corona and arcing, you may want to pot the entire winding using a casing similar to those used for CRT flyback transformers. For prototyping and experimentation immersing in mineral oil is a practical approach. It can be messy but it is a very effective insulator. I've wound HV transformers that I placed in a glass jar of mineral oil with the leads feeding through insulators in the lid. Just don't use stranded wire leads, the oil will wick up the wire.

[T3sl4co1l]

Bank wound construction (or "pi wound" like the old chokes) is fantastic for decreasing capacitance.  Downside: more leakage inductance.  Well, in short, it's a method to raise the characteristic impedance of the winding.  At high turns ratios, it's impossible to get a Zo anywhere near what you're using it for, so every little bit helps.  (Such a construction might achieve Zo ~ low kohms, whereas you might be using it for ~mA at ~kV = Mohms!)

CCFL transformers are cheaply available, constructed in this way.  I don't know how much OCV they generate, but it could be a few kV.  I've seen small Tesla coils made with them.

Tim