What causes a HV transformer to internally arc? Lack of proper isolation?

[Red Squirrel]

Just messing around with some ferrite transformer parts I got off Digikey a while back, wound a HV transformer that I'm switching at about 40khz with a 555 timer and mosfet.  I might make a jaccob's ladder or something with it.   One thing I noticed though is that when I input 12v into it, I get arcing within the transformer itself, the arc seems to be between one end of the secondary, and another part of the secondary.  It's hard to tell though as a section just lights up blue, and I don't leave it on very long because it smells funny, and funny smells in electronics are usually not a good thing. 

Just wondering if this is just the thing of needing to isolate it better, ex: put tape between each layer?  Or if there is more to it that I need to take into account.  The final turn of the secondary also ends at the opposite side of the pins, so the wire crosses to go to the pin, so I wonder if that is an issue too, I imagine you want to avoid that?

Pic for your viewing pleasure:

[MosherIV]

Generally speaking, it will be the insultation on the wire breaking down, especially with the kind of voltage you are playing with.
I am not sure but I believe HV transformers use different/special coatings on the wire to cope with the HV.

The smell might be ozone, HV will create ozone.
It is not greate for you, in high quantities it is a carconogene (I am pretty sure I have spelt that wrong, must install a spell checker on this stupid tablet  )

[David Hess]

I am not sure but I believe HV transformers use different/special coatings on the wire to cope with the HV.

They use winding patterns which prevent adjacent areas from having too much potential difference.  Insulating tape is used between layers.

[Brumby]

Arcing and insulation breakdown go hand in hand.

The two suggestions I can offer are:
1. Better insulation between turns, which includes the coating of the wire and insulation between layers.
2. Minimise the voltage differential between turns of your secondary.  That is, if you have 1,000 turns creating your HV, don't put turn 1 near turn 1,000 - or you will have the maximum voltage stress at this point and an arc far more likely.

I believe the classic approach is to have several layers, with an insulating wrap between them.

The basic idea looks something like this - where the maximum stress between adjacent turns is 2KV...


There are other more complex arrangements where you might have several adjacent sections - or even separate coil sections ... but there will be others who will have a better knowledge on this.

My diagram at least gives you an idea...

[Buriedcode]

I'm not sure if you're asking what affects the breakdown voltage of transformer windings, or why its arcing at all?

In terms of insulation you have the insulation of the wires - usually enamel, sometimes teflon for when a very high voltage is generated, and layers of insulation tape between winding layers.  If you have say, 500 turns on the secondary, with 50 turns on each layer, and you're developing 1kV with it (depends on how you're driving it, as a coupled inductor, or transformer).  That's 2V per turn.  That means the insulation between windings only has to cope with 2V.  But a layer is 50 turns, at 2V each thats 100 volts for that layer. As you wind to one end, then back over it, you now have two layers with 2x100V = 200V between them at that end. Perhaps the enamel insulation can cope with that so you would insulate every two layers.

Another method is the split bobbin: http://modelengineeringwebsite.com/Blokker_I_C_ignition_files/origineel_blokker_circuit.jpg

It breaks up the coil into sections, with that one 6kV, each section is 1kV from the inner most winding, to the outer.

To further prevent internal breakdown, you can immerse the whole coil in oil, or pot it using a vaccuum pump to remove all the air (because air is easier to punch though that the potting compound).


If you're asking why its generating such a high voltage..
With nothing connected to the secondary you essentially have an inductor, and driving it with a square wave via a switch (be it MOSFET, bipolar etc..) you're building up the magnetic field, then when the switch opens it collapses generating current in both coils - that has to go somewhere, so the voltage rises to whatever it needs to be to get current flowing. This could be an arc at the secondary terminals, or an arc across the switch (breakdown of the MOSFET).  You can get several kV with an input of 12V with a turns ratio of only 1:10 - if you have the secondary connected to a spark gap that is the correct distance apart so it breaks down at a few kV.  But at the same time, the primary's voltage rises too, say if you somehow manage to get 1kV from a 1:10 transformer, the primary voltage would shoot up to 1kv/10 = 100V - that can kill your switch.

Driving with an AC waveform, push pull, you're building up the magnetic field, and then 'draining it' back out, with the output secondary voltage being a product of the input voltage across the primary, and the turns ratio.

Driving a transformer as a boost converter with the secondary not connected to anything and no voltage clamp across the switch, it'll either destroy the switch, or.. it will arc internally between the layers, burning away insulation, and making further arcing easy, or rather, reducing the voltage it needs to jump.

[Red Squirrel]

Hmm makes sense and I kinda suspected the pattern had to do with it.  At the start of winding I was following that pattern but then near the end it started to get a bit messy as some wires would "push" between other layers and push wires aside.   I imagine if I redo it but tape between each layer I'll have better luck.  Masking tape should do?

I imagine a permanently set would also be potted as mentioned.   

[Brumby]

Yes, neatness counts - for the reasons you are probably realising.

Masking tape wouldn't be my first choice - but it is better than nothing and will improve your results.

[Alex Eisenhut]

Jacob's Ladder? Maybe for ants, with that thing. You'll need a ignition transformer or neon sign job. These are lethal, be careful.

[Red Squirrel]

Mini one, just want to see what I could pull off with what I have.  I think once I rewind this better with proper isolation and reduce the primary turns I'll be cooking with fire. I was able to get some decent arcs with this one as is.  I also touched it by accident... thankfully I was only driving it with 5v at the time. That was a zinger.  The wave form I'm getting on the scope is not very pretty though so RMS voltage is probably much lower than the peak, lots of ringing, probably the result of not really doing any kind of calculations, just throwing it together.  Just a quick and dirty experiment.   From my understanding, frequency and input voltage play a big role in how many primary turns you need.  Too little or too much and it's not running efficiently. 

I'll get a bunch more turns going when I redo it and see where it brings me.  With one end grounded the other wire was making nearby CFL lights turn on.   But yeah very dangerous stuff at this point. I'm not even current limiting the input, I probably should. 

[Bud]

  As you wind to one end, then back over it, you now have two layers with 2x100V = 200V between them at that end.

That is why there is a winding technique where you do not back over but swing the wire right back and continue in the same direction . This reduces voltage potential between layers by half.
Also you have to maintain sufficient creepage to prevent arcing over the air between layers on the edges of the bobbin.

[timb]

Hmm makes sense and I kinda suspected the pattern had to do with it.  At the start of winding I was following that pattern but then near the end it started to get a bit messy as some wires would "push" between other layers and push wires aside.   I imagine if I redo it but tape between each layer I'll have better luck.  Masking tape should do?

I imagine a permanently set would also be potted as mentioned.

Get some Kapton (aka polymide) tape. You buy it in various widths, from 1/16" up to several inches or more. I find 1/4" to be the best general purpose width to have on hand.

It is heat resistant and has a fairly high breakdown voltage. You can buy it on Amazon or Digi-Key; it's not super expensive.

Failing that, you could pickup a roll of Teflon thread sealant from your local hardware store (or a plumbing supply place). It's not sticky like tape; it clings to itself sort of like Saran Wrap. It's very thin and stretchy so it will form to whatever you apply it to. Normally it's wrapped around the threads of metal water pipes to create a seal.

Make a few wraps around each layer; since it's teflon it should act as a nice dielectric. It will certainly be better than masking tape!

Also, keep in mind that masking tape is paper based and *will* catch fire if the transformer arcs. At least Kapton (and Teflon) shouldn't ignite.

[T3sl4co1l]

Get some Kapton (aka polymide) tape. You buy it in various widths, from 1/16" up to several inches or more. I find 1/4" to be the best general purpose width to have on hand.

Yellow polyester (e.g. Scotch #53 I think??) is the classic, which is a bit easier to use (a little bit more stretchy, stickier, and easier to bend and cut).

That said, polyimide is top tech material for high temperature, high strength and high voltage applications.

Remember that high voltage creeps around the edges of an insulator, so you need to use extra tape around the ends of each layer.  The best way to handle this is to wind less than the full width of the bobbin, put down a full-width layer of tape, then continue.  (You may find it's hard to keep the wire from falling off to the side, near the ends of each layer.  That happens.  Just deal with it.  Stick the wire in place with more tape, or something. )

[Red Squirrel]

Ohh never thought of teflon, I will try that since I have it on hand.  Maybe do one wrap around, then have the wire cross back to the beginning, then do another wrap around, then start new layer, think that should work?

Never heard of Kapton before that sounds interesting as well.    Did a search for "transformer tape" on ebay to see what I get and think I found what I was talking about:

http://www.ebay.ca/itm/MYLAR-HEAT-RESISTANT-ADHESIVE-INSULATION-TAPE-12MM-X-66M-YELLOW-High-Temp-/271644487876?var=&hash=item3f3f45f0c4:m:m9nKe-VOq8fl4YsyVBB8Vqg

Is that the stuff that is normally used?

This is just a quick and dirty project but if I do plan to wind any transformers that I'll use I'll get that tape.

[Brumby]

Search for 'Kapton' - and you'll find all manner of sizes.

[David Hess]

Kapton tape is commonly used to mask off areas of a printed circuit board before wave soldering.  We used it to hold assemblies together when baking them in the oven.  I like to refer to it as aerospace tape.

[Jeroen3]

https://en.wikipedia.org/wiki/Corona_discharge
You'll need to vacuum dip coat the thing.

Source: company builds transformers.

[Red Squirrel]

Actually yeah are transformers typically dipped in varnish or something?      I probably won't do that until I decide to use it permanently though.  I found Kapton tape on amazon though might order some then rewind the transformer and use it between each layer.  Going to push my luck and reduce primary turns too. 
 
Of course I had to pick a HV transformer as my first try at making one, but I like a challenge.

[David Hess]

Actually yeah are transformers typically dipped in varnish or something?

They do not have to be but often are.  As Jeroen3 points out, vacuum potting the transformer helps a lot where a higher breakdown voltage is needed and this is standard on anything where corona discharge can be a problem like high voltage multipliers.

CCFL transformers get away without vacuum potting but take other steps to prevent problems.

[Jeroen3]

Vacuum potting is another step.
You must (vacuum) impregnate transformers if you want it to survive anything above mains. This is typically performed by dipping the entire thing in a bath of varnish, and let it soak for a while.
Another step to enhance durability is to (vacuum) pot the transformer. But this only adds to protection from the environment. Not necessary for easy indoors room temperature operation.

[R005T3r]

Actually, a true vacuum is not possible to obtain.... However, there's the SF6 that's used in High voltage mains applications, and as far as I know It's also used to power up 220KV/20KV city ultracompact transformer stations...

[CM800]

A nice way to do HV transformers it do put them in a nice little metal box with feedthroughs, then suck the air out of it from one 'port'
and once it is down to a decent vacuum, open up a second hole that connects to the bottom of an oil tank. The vacuum will then draw in all the oil which will fill in every crack even between windings without leaving air in them.

[MagicSmoker]

Jeezy petes are you all forgetting this is the *beginners* section?

Red Squirrel: 1 mil thick mylar (polyester) or kapton (polyimide) tape are most commonly used for inter-layer insulation in switchmode transformers and you can buy both types from Amazon (I prefer kapton myself). Alternatively, kynar or tefzel insulated wire wrapping wire has a surprisingly high breakdown voltage in practice (easily 1kV) so is a good choice for homemade HV secondaries; #28AWG (0.32mm d.) can handle about 300mA continuous and is a commonly available size (#30 is also common, and good for about 200mA). Another tip that helps a lot is to apply a light coat of clear acrylic nail polish to each layer after it is wound (then let it dry!) to help prevent wires in the next layer you wind from dropping down in between turns on the layer below. Also, there is no rule that says you have to put the start and finish of each winding on adjacent pins on the same side of the bobbin - sometimes, especially for HV windings - it makes more sense to start at the first pin on side of the bobbin and end at the first pin on the opposite side. This is especially useful when you want to minimize voltage stress between layers because you can then start each layer or a multi-layer winding on the same side of the bobbin then do the inter-layer connection on the bobbin pins, rather than run the wire across the top of the just-wound layer to start the next.

You can also buy magnet wire with more than one insulating film on it. The jargon for this is single build = 1 layer; heavy build = 2 layers; triple and quad build should be self-explanatory. Amazon carries mostly single build wire, but you can find heavy if you are careful (usually need to look for a larger diameter, if "heavy" isn't specified).

Good luck, have fun, and don't kill yourself unless you deserve to get a Darwin Award anyway.

[Red Squirrel]

Ordered the Kapcon tape so I'll probably play with that when it arrives.   If I feel eager to start before it gets here I'll try the teflon, but I feel an actual sticky tape should be much easier to work with as it will "lock in" the layers so I can let go as needed between each layer. 

I won't bother with vacuuming or dipping or anything for this, but if ever I make something I want to keep as a more permanent project then I'll probably look into those options.  Though I imagine I can still do that after the fact either way.

If I can get my hands on an old fridge compressor or something that could probably work to make a vacuum pump, or even an air compressor using the intake. 

[Zero999]

Actually yeah are transformers typically dipped in varnish or something?

They do not have to be but often are.  As Jeroen3 points out, vacuum potting the transformer helps a lot where a higher breakdown voltage is needed and this is standard on anything where corona discharge can be a problem like high voltage multipliers.

CCFL transformers get away without vacuum potting but take other steps to prevent problems.

Well CCFLs work at relatively low voltages, around the 1kV range, so it's fairly easy to keep the voltage between layers below a couple of hundred volts.

[David Hess]

Actually yeah are transformers typically dipped in varnish or something?

They do not have to be but often are.  As Jeroen3 points out, vacuum potting the transformer helps a lot where a higher breakdown voltage is needed and this is standard on anything where corona discharge can be a problem like high voltage multipliers.

CCFL transformers get away without vacuum potting but take other steps to prevent problems.

Well CCFLs work at relatively low voltages, around the 1kV range, so it's fairly easy to keep the voltage between layers below a couple of hundred volts.

The CCFL ignition voltage is significantly higher.  Many of these transformers wind the secondary in separate isolated sections.

[Zero999]

Actually yeah are transformers typically dipped in varnish or something?

They do not have to be but often are.  As Jeroen3 points out, vacuum potting the transformer helps a lot where a higher breakdown voltage is needed and this is standard on anything where corona discharge can be a problem like high voltage multipliers.

CCFL transformers get away without vacuum potting but take other steps to prevent problems.

Well CCFLs work at relatively low voltages, around the 1kV range, so it's fairly easy to keep the voltage between layers below a couple of hundred volts.

The CCFL ignition voltage is significantly higher.  Many of these transformers wind the secondary in separate isolated sections.

1kV is the typical starting voltage for a CCFL. Once started the voltage drops to a few hundred volts.
http://www.ti.com/lit/ds/symlink/ucc2973.pdf

[timb]

Ordered the Kapcon tape so I'll probably play with that when it arrives.   If I feel eager to start before it gets here I'll try the teflon, but I feel an actual sticky tape should be much easier to work with as it will "lock in" the layers so I can let go as needed between each layer. 

I won't bother with vacuuming or dipping or anything for this, but if ever I make something I want to keep as a more permanent project then I'll probably look into those options.  Though I imagine I can still do that after the fact either way.

If I can get my hands on an old fridge compressor or something that could probably work to make a vacuum pump, or even an air compressor using the intake.

You can get small vacuum pumps relatively cheap these days. I picked up a little unit a few years back for >$200, including hoses, gauges and a 5lb tank of refrigerant! It's designed for evacuating automotive A/C systems, which is what I got it for.

Most auto shops were quoting me over $100 to do a purge, vacuum and charge on my system, so I figured for a bit more, I'd have all the tools to do it myself. Glad I did! I've done like 10 cars for friends and family with it, plus the pump itself has come in *very* handy in a couple of projects!

Just keep an eye out on Craigslist, I see them on there every now and then. You can also buy just a pump on eBay for >$100.

[R005T3r]

A nice way to do HV transformers it do put them in a nice little metal box with feedthroughs, then suck the air out of it from one 'port'
and once it is down to a decent vacuum, open up a second hole that connects to the bottom of an oil tank. The vacuum will then draw in all the oil which will fill in every crack even between windings without leaving air in them.

Actually instead of oil I would have used epoxy resin: epoxies have very good insulation proprieties. I'd be careful with oil because the bloody box can explode if something goes on fire.
I know for sure that in mains power distribution transformer oil is used only for cooling not for insulation: they still stick to the windings  enamel  for that.

[SeanB]

If potting you just fill the case with enough polyester resin to cover the part, then pump the case down to a vacuum slowly, so the air has enough time to come out of solution, and not foam up too much that it goes into the pump ( a small clear filter housing is good there, so as to enable you to see the foam and not allow it into the pump), and after it has been pumped down for 10 minutes you simply turn off the pump, allow the vacuum to equalise back to ambient and remove the parts. Then bake at 80C in an oven for 4 hours to cure the single part varnish.

Change the vacuum pump oil at this time as well, it will not be too good with all the dissolved solvents in it.  You can buy a special ( at a price) low boiling point vacuum pump oil, but a straight mineral oil ( I use a Napthetic compressor oil, Viscosity 68, meant for R22 or R600 use, not a POE or PAG oil) will work there as well. Recycle the oil properly, or use it as lubricant for gates, doors and other non critical operations.