High Frequency Transformer Design and Troubleshooting

[smk037]

I am trying to do some high frequency dielectric strength testing. Right now, we are using a 500kHz generator capable of outputting ~100Vp-p and running it through a hand wound toroidal transformer. It is working okay but we are having problems with the waveform getting distorted and voltage drop when it heats up (it needs to run for 30 seconds continuously). I'm also having a problem with the insulation on the transformer windings breaking down due to the high voltage using 32AWG magnet wire. We tried with regular and double insulated wire but this is causing the toroid core to heat up enough to melt the insulation. We must also maintain a crest factor of 6 with a tolerance of +-10%.

I can not find a suitable ready made transformer or generator with this kind of output.

Any equipment suggestions? Also, if anyone has any ideas on how to make a more stable transformer, it would be much appreciated. I'm not well versed in high frequency magnetics.

 

[sophiecentaur]

It looks as if the core material may not be suitable for 500kHz operation and is too lossy (perhaps at a second or third harmonic (?).
I could suggest an air core and a single layer for the secondary ( a 1:100 ratio Tesla transformer). You may also need some tuning of the primary, to maximise your output power.
You are bordering on dangerous voltages, so be careful!

 

[sophiecentaur]

Perhaps the core material is not suitable for operation at such frequencies. It may be too lossy - particularly at harmonics. Why not use an air core and single layer secondary (Tesla Transformer) and some tuning of the primary circuit?

Take care at these voltage levels - you could hurt yourself.

 

[smk037]

Thank you, yes it is dangerous but we always test in a very controlled/safe environment.


How can I wrap an air core? Do I just wrap the secondary around a bobbin, the primary on top of the secondary and remove the bobbin?

Again, thank you for the help.

 

[sophiecentaur]

Thank you, yes it is dangerous but we always test in a very controlled/safe environment.How can I wrap an air core? Do I just wrap the secondary around a bobbin, the primary on top of the secondary and remove the bobbin?

Again, thank you for the help.

Different diameters of plastic waste pipe for primary (outside) and secondary (inside). Make the Primary shortish and near the bottom of the secondary to reduce the max E field between them - sprayed with some goo to hold the secondary windings in place. I used a lathe to get hundreds of turns (v. neatly) on a plastic tube when I did it (35yrs ago afair).
Tuning can make all the difference to this thing.

 

[berkeman]

Thank you, yes it is dangerous but we always test in a very controlled/safe environment.


How can I wrap an air core? Do I just wrap the secondary around a bobbin, the primary on top of the secondary and remove the bobbin?

Again, thank you for the help.

Perhaps you can just actively cool the transformer?

 

[smk037]

I've tried sticking a computer fan on it. I was thinking about maybe getting two little piezos and sandwiching the transformer in there, or maybe one piezo and a computer fan.

 

[sophiecentaur]

When in Rome . . . . .
The device you want is more like a Radio Transmitter than anything so why not use transmitter methods? That means an air core - very practical for the inductance values you are dealing with.
It surprises me that you are not rushing off to build at the mention of the words "Tesla Transformer" as it is flavour of the month with many PF contributors.
Afaic, it's horses for courses and that would seem to be a very suitable horse for this particular course. Only if you can get the right core material for 500kHz would there be any advantage in using a core (except if size were a big issue). Air is 'free', too. haha.

 

[jim hardy]

Sophie has a point - if the magnetic core is causing problems, can you make do without it?

A couple thoughts came up from your remarks:
You haven't said a word about the toroid core you're using.
If physical size dictates that you need a core perhaps you should focus some effort on it rather than the insulation on the wire.

Core loss is roughly in proportion to flux^2 X frequency^1.4 .
http://www.micrometals.com/appnotes/appnotedownloads/corelossupdate.pdf
So reducing flux is something to look at.

Here's a datasheet for a Magnetics Inc "Kool Mu" core that looks like it might be useable out to your frequency. The curves tell you a lot.
http://www.mag-inc.com/products/powder-cores/kool-mu/kool-mu-material-curves

If you place one turn around yours and observe it with a 'scope you'll be able to easily calculate how much flux you have, and can see clearly whether it's linear. If you are driving your core into saturation you'll see two effects:
1. It will get hot, see "core loss density" curves. You'd probably need to stay below 0.1 T for that particular KoolMu core at that frequency.
2. Secondary winding's induced voltage will have extremely high 'spikes' twice per cycle as core "flips", and you might pierce its insulation.. see Permeability vs AC Flux curve. Looks like you'd want to stay below ~ 0.5 Tesla for saturation in that KoolMu .. and it'd heat quickly at that level.

Your core must be useable at your frequency and flux levels.
Try a frequency sweep at pretty low voltage to see if your core performs out to your 500khz;
Try doubling or quadrupling the number of turns on both primary and secondary to reduce flux level, and observe whether secondary waveform changes.

Sophie's air core sounds attractive to me. More area but no ill-behaved iron.

old jim

EDIT: It's tedious to double the turns - you'll get same information by driving it at half or quarter voltage. In the linear region with sinewave excitation, flux is volts per turn after you straighten out the units .