Stepping 12 to 800 volts: Arc-over on my transformer, is this even a good approach? / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Stepping 12 to 800 volts: Arc-over on my transformer, is this even a good approach?

1 2 next

Move Thread

LAN_403

DerpBrap

Fri Aug 28 2009, 06:49AM

Registered Member #2320 Joined: Fri Aug 28 2009, 06:22AM
Location:
Posts: 7

Hey everyone, I'm new to this forum and a novice electronics hobbyist. I've been building coilguns for a while but basically through trial and error and a whole lot of camera flashes.

I'm using a flyback driver similar to the one found here:

only instead of a 2N3055 I'm using the horizontal deflection MOSFET (SSH7N90) from the monitor a flyback came from. I tried both and they both work great, but the mosfet makes much better output of big clean arcs.
I thought I'd give an E-core ferrite power transformer a try because ferrite operates at similar frequencies to a flyback. I hooked a custom wound transformer of 10 turns primary (20AWG wire) and something like 730 turns secondary (28AWG wire) in place of the flyback. The core for the transformer came out of an old tube TV and happened to come apart without self destruction. It's a fantastic core. However, the first thing I noticed was ozone and fuzzy noises coming from the windings. The secondary coil was arcing down into the core :(
However, upon putting load on the terminals the arcing stopped and it functioned just fine. I also tried drawing an arc from the secondary and to my surprise, a hot and fuzzy 1cm long arc would form. I guess it was in resonance or something because this transformer should not be producing that much voltage, right?

I put some 1000V highspeed diodes on the secondary and attempted to charge an 800v, 420 uF capacitor bank. It worked great and charged it to full in what seemed like decent time.

Questions:

1) Why did the 1000v transformer arc like that? Was it in resonance like a tesla coil or something?
2) Is a ferrite E-core transformer a suitable solution for stepping 12 to 800 volts?
3) How can I prevent arc-overs on the secondary? I tried insulating every 3-5 layers on the secondary which did not help.
3b) would winding a 400v secondary and using a cockroft-walton multiplier to double it to 800 volts be a suitable solution?

4) I also read about Uzzor's boost converter which is supposed to be very efficient, but it's rated for 500 volts... can it be modified to produce 800V?

Thanks guys.

Dr. Dark Current

Fri Aug 28 2009, 08:22AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

With 10 turns on the primary and 12 volt supply voltage, you probably have something like 100+V flyback spikes on the primary, which translates to something like 7kV on the output unloaded.

To design a flyback transformer properly you should consider the primary flyback spikes, the transistor voltage rating and from here calculate the turns ratio. I'd choose something like flyback spikes of 4x supply voltage, which means you need to use a 100V MOSFET and with 10 turns on the primary, use 277 turns on the secondary.
Then you must make sure the phasing is correct and use a single rectifier diode to rectify the flyback pulse, not the forward one.
The no-load voltage will still be higher so you must make sure you don't let the output go open-circuit if the transformer is not designed to take the voltage.

Proud Mary

Fri Aug 28 2009, 11:28AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Hello, lad, and Welcome to our online community!

The 2N3055 circuit shown in your link is a rotten, one but its simplicity had lead to its mushrooming all over the web. It's no good because whilst 555 can source only up to 200mA on its third pin, the 2N3055 requires 5 AMPS base current (from memory) to work at its best. So 555 is simply too small, too under-powered, to make anything but the most feeble job of driving the 2N3055.

As you have found, 555 is more than enough to switch your MOSFET, because the gate of a MOSFET is (ideally) totally insulated from the current it controls, which, simply, it does by static electricty. You know the way you can rub a comb with a dry cloth and then use it to pick up little pieces of paper? Very basically, much the same happens inside a MOSFET, where charge on the gate can either repel or attract a flow of your paper pieces, so allowing current to either flow through the devices, or pinch it off.

Now, because the charge on the gate is a static charge - static meaning just what it says, the electricity doesn't go anywhere, no current is needed to maintain it. Which is why the 200mA output of 555 is much more than enough to satisfy it needs.

So we say that a bipolar transistor like 2N3055 is a current controlled device (and as we have seen it requires a devil of a lot of current to control it!) and MOSFETs are are voltage controlled devices.

Now imagine a splendid old Victorian capacitor consisting of two brass plates facing eachother, and insulated on glass rods. Think of a little hopper or silo above them trickling down like polystyrene or paper balls. When there is no charge, the balls will fall straight down between the two plates.
This is your current flowing through the device. What will happen if you know place a high voltage charge across the plates? The balls, as they fall, will be deflected by electromotive force, the force that will pick up paper pices with a comb. Increase the voltage still further, and al the balls will stick to one plate or the other and none will get through (this is how electrstatic smoke cleaners work in factory chimneys) At a certain voltage on the plates, all flow of balls will stop, and in a MOSFET this is called the pinch-off voltage.

Now depending on just what your balls are made of (

) they'll be attracted to either to the negative plate, or the positive plate, and this attraction or repulsion, positive or negative, is the basis of the great divide in the MOSFET species between N-type and P-type.

("Don't let me catch you drifting off there, boy"!

And so at length back to your own experiment, son, and how you can improve upon it. Nothing happens in this world in no time at all, do it?
You'll see that in our little experiment above, time must be taken for the metal plates to reach their full charge once the electricity is switch on. At the moment you switch the power on, current will flow in the circuit until the plates are fully charged, then it will stop because the two plates are insulated from eachother by air, and once they are charged there is no longer any difference in pressure ("potential difference") that would allow the electrons to move on from one place to another. They are stuck on the surface of the two facing places, with equal and opposite charges.

Now think of your MOSFET and the time it must take to charge up its gate. The bigger are the plates, the longer it must take to charge them, right? Now if you think of the GATE of your MOSFET as being one of those plates, you'll see that to control the big currents and high voltages that MOSFETS can handle, (think of huge numbers of the famous paper balls falling down between the plaes at high speed) those plates would havee to be very large indeed.

This is one of the down-sides of MOSFETS - their high input capacitance. Think of your square wave motoring along the wire highway from the output terminal (3) of your little 555. In real world electronics, there is of course no such thing as a pure, true square wave, since that would mean the voltage rose and fell in no time at all (t=0) but we'll forget about that real world detail while iwe imagine your first square wave pulse arriving at the gate of your MOSFET, on which there is as yet no charge. The pulse tried to go straight up vertically, but of course it can not because it must first charge up the capacitance of the gate to reach that full voltage. And the larger the capacitance of the gate, the longer that do take.

So how does this affect the performance of your apparatus, son? Remember that a voltage is induced in an inductor by the collapse of its magnetic field. The faster this collapse, the higher the voltage induced. This is the crucial difference between regular power transformers, where an alternating current flows continuously and voltages are stepped up or lowered in proportion to the ratio of the turns on the primary and secondary, and so--called "disruptive discharge transformers" like the motor car ignition coil.

The formula (derived from Faraday's Law) for calculating the voltage of a disruptive discharge transformer(DDT) is usually written like this:

e=L*( di/dt)

What this mean is that the voltage e produced by a DDT is a product of

(a) the size of its inductance L (higher voltage needs more turns of wire, as you'd expect)

and -

(b) the difference in the current through the coil (di) for a given difference difference in time (dt)

So the faster you can switch the current through the coil, the higher the voltage will be.

Now think back to the big gate capacitance of your MOSFET having to slowly charge up with each new pulse from the 555. And slowly discharge again before the next cycle. Instead of a dramatic 0 - 100 speed dash in zero seconds, the voltage rises only gradually in time, when what you want to get the highest voltage out of coil is the fastest possible rise time that you can achieve.

Your MOSFET is but an electronic switch. When the switch is ON, current builds in your coil at a rate determined by what is called its time constant (Google "time constant of an inductor" ) plus of course the lag caused by time needed to fully charge the gate capacitance.

When the gate voltage pinches off the current flowing through the MOSFET, switching it off, then the magnetic field has no choice but to collapse. But think about that collapsing magnetic field. Energy - electric energy - was needed to create it, wasn't it, so when it collapses that energy must go somewhere. (Rule Number 1 of the Universe: "Energy can neither be created nor destroyed") And where it goes is back into the electricity that created it in the first place. The faster the collapse, the greater the voltage induced in the coil.

So when you have a MOSFET that only turns on and off rather slowly, you are losing that true ON/OFF action which will get you the highest induced electromotive force (EMF) for any particular coil.

So what is to be done to speed up the gate charging rate? The simplest improvement, using components costing only a few pennies, is to place a buffer amplifier using two transistors between the 555 output and the MOSFET gate.

By the way, whatever badly thought through circuits you may see both here (sadly) and on the web generally, as a rule of thumb an oscillator (as is the 555 timer in this role) should always be followed by a buffer amplifier to (a) maintain a constant load on the oscillator (b) isolate the oscillator from the next amplifier stage, (c) help to keep all of your electricity flowing the way you want - it comes as a great surprise to many that a signal can travel seemingly 'backwards' from the output of an amplifier to the input, but there are no roadsigns on the wires, and since every wire in a circuit is connected in one way or another to all other wires as a network, we shouldn't be too surprised about signals flowing in places where we didn't expect them. A buffer amplifier between 555 and MOSFET reduces the chances of generally instability in such a simple circuit, and is the most obvious improvement you can make as a retro-fit add on.

I've gone on at length, and over-simplified some MOSFET operations , perhaps, to see you started on the right road. And now to address your specific questions:

1) Why did the 1000v transformer arc like that? Was it in resonance like a tesla coil or something?

You have generated too high a voltage for the insulation to withstand

2) Is a ferrite E-core transformer a suitable solution for stepping 12 to 800 volts?

Perfectly good. But don't foget you may use less components and time by direct voltage multiplication from your mains supply, especially if

3) How can I prevent arc-overs on the secondary? I tried insulating every 3-5 layers on the secondary which did not help.

Very old amateur radio books often give blow by blow descrptions of how to wind very high quality transformers yourself, so keep an eye out for them. If you're still around when I have more time I'll snag a chapter from a book and .pdf it to you. But generally, your insulation has failed.

In a coil, the voltage in adjacent windings is not generally high, but if you have your winding all higgedly-piggeldy than later turns with high voltages induced in them will be too close to turns with a much lower potential difference, and the insulation may be over-stressed and break down with an arc. Once this has happened, there is likely to be carbon track of burnt insulation between them, and carbon being a conductor of electricity, the fault will recur and get worse.

The obvious solution to your insulation problem is to immerse your coil in dielectric oil.Ordinary kitchen sunflower oil or rape seed oil is a good choice, and both are used in an increasing number of industrial power transformers because o it is environmentally friendly. Only use clean new oil out of an unopened bottle, so it won't have absrobed water from the air. If you go for an oil insulated transformer, rewind your transformer using layers of kraft paper (the brown paper used for parcels) between each winding. Don't use any kind of plastic insulation in the winding, is this will stop the oil soaking through. (By the way, huge high voltage industrial transformers very often use kraft paper and oil insulation, so don't be afraid that it doesn't sound very hi-tech!

)

Ideally, having re-wound your coil with kraft paper insulation, you should now place it in a vacuum chamber so that the oil can be forced though it, ensuring that that there are no air bubbles where corona breakdown could occur.

Your solution is to submerge your transformer in oil in a small saucepan, and then slowly heat it up to say, 100 centrigrade, so that the oil becomes as sunny as possible, and then let it cool with a tight lid on. (The reason for keeping moist air out as much as possibe is not that, clearly, absorbed water will reduce the dielectric strength of the oil, but that the dissolved water will form a constant boiling point mixture with some factions of the oil, so no amount of re-heating can ever make it chemically dry again. Anyway do a few cycles of heating and cooling like this, as air bubbles will expand as the oil is the heated and so break free. And give it a good agitation as long as your patience lasts. Finally, keep the transformer in a warm place (so the oil stays thin) for a week or so. The good thing is that oil transformers are inherently self-healing - oil will flow into any gap created by arcing, whilst with solid dielectric insulators, once they have been punctured by an arc the hole stays there forever, a dead duck.

If you wind your coil so that a cross section is very much a triangle, rather than a rectangle, you'll see that the distance across which a destructive discharge must occur is very much increased. If you wind with this triangular cross section, but keep your insulation out to the full width, you'll see that the amount of insulation a spark must get round increases with voltage gradient.

Lastly, a quick note on a better style of winding for a mullti-layer coil high voltage coil.

3b) would winding a 400v secondary and using a cockroft-walton multiplier to double it to 800 volts be a suitable solution?

A good idea. C&W are most suited to low power applications as yours sounds to be. How many milliamps are you hoping to get out of it?

Got to go! It's lunch time.

Harry.

DerpBrap

Sat Aug 29 2009, 05:08AM

Registered Member #2320 Joined: Fri Aug 28 2009, 06:22AM
Location:
Posts: 7

:O

Thank you so much for the detailed post! Really really informative. I'll experiment with the suggestions and see what happens.

The goal here is a capacitor charger so the CW multiplier would be pumping through as much amperage as possible... I'm going to play with that over the weekend and just see what happens but if it's better suited to low current operations then it might not be best for this.

Oh and I have thought about rectified mains for simple and high speed charging but this is going to be a portable thing so it's got to be DC-DC step up.

UPDATE:
I found a bobbin of radio shack magnet wire so I set out to arbitrarily make a new transformer. Using the triangular shape the secondary ended up being about 1000 turns (yikes) so I wound a primary of 30 turns to try and get about 500 volts out. It works! no internal arcing this time which is nice. However, the 12 volt line is only running about 3 amps. I suppose that's to do with the inductance of the extra turns on the primary. It does work, though.
Next time (tomorrow most likely) I'm planning on using a fresh bobbin and trying out a transformer with fewer turns overall, like what Dr. Kilovolt said. Hopefully that'll get the overall power up. The arcing transformer yesterday was drawing more than my ammeter can read which is more than 10 amps, so it must be possible to go extremely high power with this system. Once the transformer is working correctly I'll be tuning the rest of the circuit properly as well.

I'm not paying much attention to screwing up because I've got a local source of well priced magnet wire and too much spare time... The more it doesn't work the more I have to think. that's good right? :P

Proud Mary

Sat Aug 29 2009, 08:54AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Ha, ha, DerBrap, you are putting the cart before the horse, I see!

Let me show you a different way of going about things:

(i) Is an engineering solution to a problem possible?
(ii) How can it be done within the real-world constraints of budget, time frame, availability of parts, labour skills, social costs?

You may feel free to ignore these first, but I put them here to show you a type of engineering method which is well worth the understanding.

Specify problem exactly:

In your case: To charge capacitors.

Far too vague you see. You must specify, at least:

Energy to be stored in the capacitor for your particular application.

But why not take a breather here, while you memorise these two fundamental concepts:

One Coulomb is the quantity of electric charge transported in one Second by a steady current of one Ampere.

AND

One Coulomb is also the quantity of charge stored by a capacitance of one Farad charged to a Potential Difference of one Volt:

So without wandering off further into theory basics, only once you more closely characterize your problem

(for example only) To charge a capacitor of 100uF to a potential of 800V

can you decide what sort of circuit would be most suitable for achieving that (always, the in real world, within the practical constraints of cost, availability of materiel, labour skills, time to completion, acceptable Mean Time Between Failure etc etc)

So, laddie, what's the size of your capacitor, what maximum voltage does it say on the can, and what voltage d'ye want tae charge it tae?

teravolt

Sat Aug 29 2009, 04:11PM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

DerpBrap welcome,if you want portibility there are a lot of good circuits like

if you want it even easyer you could use a 120 to 480 transformer connected to a full wave doubler. the current would only be limited buy the size of your transformer and doubler capacitors. If you want 3 120 to 120vac transformers with the outputs in series for 360 vac would work

DerpBrap

Sun Aug 30 2009, 12:11AM

Registered Member #2320 Joined: Fri Aug 28 2009, 06:22AM
Location:
Posts: 7

Harry wrote ...
So, laddie, what's the size of your capacitor, what maximum voltage does it say on the can, and what voltage d'ye want tae charge it tae?

I have two 400 volt, 820 uF capacitors to charge. I'd like to use them in series for 800 volts total and charge them to as close to 800V as possible.

:)

Proud Mary

Sun Aug 30 2009, 11:50AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

DerpBrap wrote ...

I have two 400 volt, 820 uF capacitors to charge. I'd like to use them in series for 800 volts total and charge them to as close to 800V as possible.

Two 820uF in series is one capacitor of 410uF.

If you want your capacitors to last, keep your working voltage a good bit lower than the maximum state on the can. Not everyone will agree, but I would suggest a maximum of about 650V across the two in series, especially in the type of rapid discharge experiments I'd guess you have in mind.

To charge up a capacitor of 410uF to 650V, only a milliamp or two will be sufficient, not great amperes as you had imagined!

Hope this helps!

All the best,

Harry.

DerpBrap

Wed Sept 02 2009, 06:43AM

Registered Member #2320 Joined: Fri Aug 28 2009, 06:22AM
Location:
Posts: 7

I've been re-reading your post about disruptive discharge operations, Harry. Do you know any good reading material on this subject as I'm only beginning to understand what's actually going on in the flyback operation.

The issue of flybacks creating way more voltage than the expected turns:volts ratio (like with AC wall warts) and predicting what voltage will actually come out of a given system is stumping me entirely. Like why (numerically and forumula-wise) did a transformer with 750 turns secondary produce about 400-500 volts with a 30 turn primary, but when the primary was shrunk to 15 turns it put out over 1000 volts? Tried running it on a couple different MOSFETs, too. a 900 volt SSH7N90A, a 200 volt 2SK2520, and a 60 volt IRFZ44V. All of them produced similar voltages out (if i recall, i didn't write anything down when i really should have).

It looks like a simple explanation might be that since the turns were halved, the inductance was basically halved (increasing the transformer's ability to fill up with and collapse a charge on each pulse), so the output voltage doubled?
No idea if that's close to what's really going on. At least I'm trying to understand, right?

Also, what's the difference between this flyback circuit and a camera flash charger circuit? The camera flash uses the same kind of E-core ferrite transformer with a small switching transistor oscillating/pulsing in audible frequencies. How come I've found they follow the standard turns:voltage ratios?

Hehe I should be studying electrical science instead of computers; this stuff is very interesting :P

Steve Conner

Wed Sept 02 2009, 09:34AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

DerpBrap wrote ...

Also, what's the difference between this flyback circuit and a camera flash charger circuit?

Probably none.

wrote ...

The camera flash uses the same kind of E-core ferrite transformer with a small switching transistor oscillating/pulsing in audible frequencies. How come I've found they follow the standard turns:voltage ratios?

Probably just coincidence. A flyback converter can output any voltage whatsoever if it's unloaded. The voltage is usually limited by breakdown of the transistor driving it.

For extra credit: Explain why the whining noise from the flash charger increases in pitch as the capacitor charges up.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob