For a low frequency 50/60Hz current transformer, the accepted wisdom is that if a current passes through the primary while there is lo load on the secondary, then the transformer will be damaged and the world will come to an end (worst case scenario).
I have never considered the case of an RF current transformer, as used in an ndB coupler, such as this home made 30dB coupler of mine:
So I did some tests using a signal generator (HP8640B), my 30dB coupler and my home made RF power meter which is based on an AD8307.
I set the sig gen to 10MHz 0dBm, and connected via a 1metre long coax cable to the power meter. The power meter read -0.3dBm
I then inserted the 30db coupler at the power meter end of the cable. The coupler was terminated with 50R on its sample port. The power meter again read -0.3dBm.
I then changed the load on the coupler sample port, and got the following results:
Load Meter
Open circuit -4.3 dBm
Short circuit -0.3 dBm
100R -0.3dBm
Clearly something odd is happening when the couple sample port is open circuit.
To investigate further, I used an AIM4170 Antenna Analyser (a simple form of a Vector Network Analyser), connected up like this:
Sweeping the frequency from 3 to 33 MHz gave the following results:
Looking directly at the input of the power meter, a nice flat VSWR / Return Loss from 3 to 33 MHz.
With the 30dB coupler at the input of the power meter, again a reasonably flat VSWR / Return Loss from 3 to 33 MHz.
With the 30dB coupler sample port un terminated (open circuit), something serious is happening in the 10MHz region.
With the 30dB coupler sample port terminated with a short circuit, we are back to the fairly flat response.
And similar results when terminated with 100R.
It is quite by chance that I started the experiments with the signal generator giving 10MHz, I thought that it was a low enough frequency to give good results when looking at the 30dB coupler sample port with a 200MHz scope.
I have not included my scope test results here, but I did see quite large voltages when the coupler was unterminated.
What I did notice was the voltage on the sample port was very dependant on what was connected to the port, even just putting a BNC "Tee piece" could make a visible difference. I guess the small capacitance of the T piece was enough to affect the high impedance of the unterminated sample port.
JimB