Partial Discharge Free Transformers

Hi Uwe,

Use the transformer with a volumetric coil. Take two identical ferrite rings, make on them identical windings, carry them on distance 2-4 inches along an axis and make one winding coupling both ferrite rings.

For reduction of a leakage inductance it is possible to make such 3-4 or more windings in parallel. These windings should have isolation more than 10 kV and can be grounded.

Krass

Ohhh Krass that sounds interesting...

I ripped out of an old piece of test equipment a component that I wasn't sure what it was for, I've been meaning to 'play' with it some time.

It is a large ferrite toroid with three fixed ferrite toroids placed at 90 * to each other on the large toroid and each small one has a hundred or so turns on them...

Your description sounds just what this component is..?

You've spoilled my enjoyment of finding out now!

John

Hi, John

Using ferrite rings K40x25x22 carried on 5 inches I have received capacity 400 W in a load (SMPS frequency is 25 kHz) two weeks ago. The voltage across a load is 6 kV, a load is placed at 30 kV.

Krass

Hi Uweka,

On January 29 I posted a question to find High Voltage High Frequency insulation information on Corona (Partial Discharge) Inception for a transformer running at 40 - 100 kHz.

I was given a lead to the book " Insulation Materials For Design And Engineering Practice" by Frank M Clark published by John Wiley and Sons Inc. Very good book but unfortunately gives no corona information other than at power line frequencies.

We have contacted insulation suppliers and manufacturers and they have not been able to help.

We manufacture transformers and would love to help you. But I do not know where to find insulation which I can guarantee fits the bill.

I doubt whether the low capacitance transformer described by KRASS will necessarily be low corona - as the two are not really related and I don't know if you can tolerate having an earthed winding between them.

Also I do not understand your suggestion of bifilar primary.

I hope someone can give us both some advice.

Hi BigBirdAustralia et al,

I agree that there is not too much info on high freq corona floating around. (Perhaps we are looking in the wrong place??) I also think that Krass's idea can work well. All you need to do is get the dimensional proportions right for high voltage design. If you break it down into a problem of circular symetry, then this can guide you through.

Contemplate this: Think about the proportions (simply the diameters) of high voltage cables. You will find that D/d is a special ratio for these cables. D is diam of earth screen, and d is diam of HV core. t is thickness of insulation between them = (D-d)/2.

Lets keep D a constant size for now. If d is made smaller than a special (optimal) ratio, causing D/d to increase, then thickness of insulation increases, and so you may think that this should be able to withstand a higher voltage, but Mother Earth doesn't work like that because the electric field around the SMALLER HV core at d INCREASES due to the smaller dimensions.

On the other hand if you allow d to increase higher than that special ratio D/d, then the thickness of insulation DECREASES causing the electric field to increase as well.

So we find that when the ratio of D/d = e (yes, as of logarithmic fame e=2.71828...) then for a given size of D, if you then make d = D/e, then that cable will have optimal proportions for HV. So in practice, the size of d is determined for a given current carrying capacity, and so D is then determined for current and voltage capability.

If you look at cable manufacturers data, you will also find that the insulation often used is some special form of Polyethylene (PE), which can withstand perhaps 100kV per mm before it breaks down (I know I'm being simple here but just to get the engineering principles across). Then look at the actual field strengths that they allow you to run the cables at, and you will find that it is in the order of 7 to 8kV per mm.

So if you want to run in air, we know that air breaks down at about two orders of magnitude less than PE, perhaps 1kV per mm (also depends on pressure, humidity etc).

Now you should be able to apply these principles to the internal diameter of ferrite cores, which would be akin to D above; and the coupling winding going through the axis of the ferrite rings should have diameter d. You will find that you should make d much BIGGER than you probably contemplated in the first place, and this will reduce the chances of CD or PD. Make it a nice big round equipotential surface at diameter 1/e times smaller than the internal diameter of ferrite cores. Perhaps a single turn "oval track" type winding.

We manufacture sensors to measure corona discharges, and power frequency voltages without making physical contact. Take a look at Pulse Peak sensor: www.suretech.co.za

Hope this helps,

NeilJ

Hi BigBirdAustralia, "I do not know if you can tolerate having an earthed winding between them." I have written "can be grounded", but not "have to be grounded". If you wish to have small partial discharge you should leave an additional winding under a flying voltage. If you wish to protect your equipment (SMPS for example) from damage when load breaks down to the earth, you should earth an additional winding. Krass

Hi Krass

Fair comment, I did assume that it would be grounded - my mistake.

But in my experience floating or flying windings or shields tend to settle at their own floating voltage, and that makes it very difficult to ensure that the insulation is not stressed beyond the Corona Inception Voltage.

The main problem is that insulation manufacturers do not - to my knowledge - puiblish a lot of useful data except at powerline frequencies and sometimes at aircraft requency of 400 Hz.

Cheers

BigBirdAustralia

Hi Uweka,

I think that I may be slowly getting somewhere after NeilJ's reply. I have emailed him.

Could you advise the input and output voltages and currents? And perhaps contact me on our very simple website www.dyne.com.au which we are in the process of updating.

Cheers

BigBirdAustralia

Hi Uweka, If there are fine bubbles in isolation it will cause a corona. It is necessary to fill in the transformer by compound in vacuum. Krass

Corona discharge occurs as narrow pulses of energy that happen in quick succession after each other. The corona sensors we have built to observe CD and PD activity has enable us to see these individual pulses. And so if you are working at 50kHz, this is still low compared to the frequency associated with CD activity. If for example you have a transformer running at 50kHz, corona activity will have its major energy component at 50kHz, which occur as a stream of tiny (width) pulses, and occur normally at the voltage PEAKS of the wave. There could be one / a few or tens (or perhaps hundreds) of narrow CD pulses that occur on each peak of 50kHz wave. So therefore if you look at the energy spectrum of CD, you will find it starting at 50kHz in this example, and the CD noise will extend into the GHz due to the very narrow pulse widths. This is the reason that one can use RF antennas to "sniff" out CD activity. The major energy band that CD will occupy is in the units / tens (perhaps hundreds) of MHz in this case. For 50Hz power systems, the CD pulses are still the same sort of narrow width (perhaps 1uS or less), but because 50Hz peak is present for 1000 times longer than a 50kHz peak, the number of CD pulses will be many more. So now if you look at the energy spectrum of CD emanating from the 50Hz equipment, it will have a major component at 50 / 100Hz, most of the CD energy will be in the tens / hundreds / thousands of kHz region. But there will still be CD energy extending into the GHz region as well.

Corona discharge normally emanates from a surface that has small radius of curvature, and so the electric field AT THAT POINT exceeds the breakdown strength of air; and so small discharges at that point will occur into the air.

Partial discharge is normally associated with discharge that occurs within air-bubbles (gas-bubbles) that are trapped within insulation. The practical effect of the partial discharge works like a woodpecker inside the bubble, who chips away at the insulation from the inside as he tries to escape. Eventually the woodpecker will have done his work (perhaps assisted by water-trees), and as he finally breaks out of his bubble, he usually leaves behind him a trail of smoke of immense proportions; he normally has a flash photographer present.

Hi,

Thank you all for your thoughts.

Krass, I`ve understood your hint about earthing right. Years ago I built inductive voltage dividers with divider errors <0.1ppm. They had two cores per decade coupled with a magnetizing winding. I`m just experimenting with a small 2-core-solution for the above problem and think its good to have the possibility to use the grounding point or not - everyone involved in HV/HC-applications knows that sometimes grounding somewhere is ok and sometimes not, depending on the application. You have to regard the (dis)advantages - grounding nearly doubles the coupling capacity and lowers the isolation voltage if your primary is near ground but often leads to lower disturbances. Apart from that I know that vacuum compound is a must. In the meantime I found a company in my hometown who could and would do that for us.

Neil, thanks for your offer in transformer design, I had done the mathmatics before and now I`m testing whether it works (and seems to ...), perhaps I find a manufacturer when I can give them a drawing of a proven design.

BigBirdAustralia, the proposal for a bifilar winding might improve the leakage induction and coupling - I thought, but as I know now the effect is minimal though lightly measurable.
Best regards to all Uwe