An example and explanation of unexpected common mode choke flashover

An online discussion is developing the design of an ultimate common mode choke, at it reached a stage considered final when a transmit test revealed it could not withstand the unstated transmitter power.

The designer did report measurement at the choke looking into the feed line giving Z=493-j740Ω @ 3.8MHz. There are questions about the validity / uncertainty of the measurement, but let's take is as correct for the purpose of this discussion.

We can calculate the expected differential peak voltage at a given power level at the point where Z=493-j740Ω.

Let's use Python as a complex number calculator to calculate based on Z=493-j740Ω.

>>> import math
>>> import cmath
>>> p=1500
>>> z=493-740j
>>> g=(1/z).real
>>> '%0.2e'%g
'6.24e-04'
>>> vpk=(2*p/g)**0.5
>>> '%0.2e'%vpk
'2.19e+03'

Above, the solution of \(V_{pk}=\sqrt{\frac{2 p}{real(\frac1{z})}}\) gives the peak voltage as 2190V.

If your analyser / VNA gives the impedance as an equivalent parallel Rp || Xp, then the following might be more convenient using Rp.

>>> import math
>>> import cmath
>>> p=1500
>>> rp=1603.75
>>> vpk=(2*p*rp)**0.5
>>> '%0.2e'%vpk
'2.19e+03'

Above, the solution of \(V_{pk}=\sqrt{2 p r_p}\) gives the peak voltage as 2190V.

Another simpler alternative is Calculate Vmax, Vmin, Imax, Imin for lossless line from Z (or Y or VSWR) and Zo.

Again, the answer is 2190Vpk.

Which ever way you get it, is 2190Vpk ok?

Given that the prototype balun was wound with enamelled copper wire, you might think it should easily withstand 5-10kV wire to wire… but it would appear to have failed in this case at less than a third of that. The voltage withstand of the enamel insulation depends on many things, and can be degraded by sharp bends (eg around sharp core edges if present). In this case the core material was #31 which is not an extremely good insulator, so that might degrade things.

My experience having tested the withstand of some samples of enamelled copper wire is that it not a good choice for over perhaps 1000Vpk for several reasons set out at On use of enamelled wire in transmitting baluns.

It is very easy to damage enamel insulation with a single flashover, and thereafter the carbon track remaining in the enamel results in greatly reduced voltage withstand. The the arc might exist for milliseconds, that is sufficient to permanently damage the insulation… and therefore the whole choke.

The calculations given here are for differential voltage at the point where Zin is known. The differential voltage at other parts of the choke may be higher or lower… higher would be a bigger problem but it is not higher in this scenario.