That  'primer' is one of those publications I mentioned  that confuses measurement equipment noise floor with the real noise floor and has a misleading spectral plot.

None the less it is quite revealing.


First, on HF in about 10kHz bandwidth, you might expect ~ -100dBm in a quiet spot between signals, in 30 times that bandwidth as they have here , perhaps -85. At the low end this is plot dominated by the 1/f frequency response of their current pick up, and then the white noise of the stage that follows it, However, the good news is that they will be making the HF detection part of their AFDDs quite deaf.

The periodic nature of the purple trace is  misleading, it is not really because the spectrum really is periodic like that with peaks and troughs every 1MHz of so. Rather it is just because this plot was taken with a receiver sweep rate such that approximately ten mains cycles passed during the time of  10MHz of sweep - the drop outs are actually due to the arc being absent for some of the mains cycle.

if the spectrum analyser had dwelt on a frequency showing a low level or notch, until the next mains cycle came along, energy would be present and it would have filled in.




Another image is also telling.


As is shows us pictorially how the microcontroller is reaching it's trip/no trip decision.

It also suggests how one may make both an all electronic test box and a countermeasure (anti-tripping) box, should either or both be required for the Siemens units.

regards

M.

That  'primer' is one of those publications I mentioned  that confuses measurement equipment noise floor with the real noise floor and has a misleading spectral plot.

None the less it is quite revealing.


First, on HF in about 10kHz bandwidth, you might expect ~ -100dBm in a quiet spot between signals, in 30 times that bandwidth as they have here , perhaps -85. At the low end this is plot dominated by the 1/f frequency response of their current pick up, and then the white noise of the stage that follows it, However, the good news is that they will be making the HF detection part of their AFDDs quite deaf.

The periodic nature of the purple trace is  misleading, it is not really because the spectrum really is periodic like that with peaks and troughs every 1MHz of so. Rather it is just because this plot was taken with a receiver sweep rate such that approximately ten mains cycles passed during the time of  10MHz of sweep - the drop outs are actually due to the arc being absent for some of the mains cycle.

if the spectrum analyser had dwelt on a frequency showing a low level or notch, until the next mains cycle came along, energy would be present and it would have filled in.




Another image is also telling.


As is shows us pictorially how the microcontroller is reaching it's trip/no trip decision.

It also suggests how one may make both an all electronic test box and a countermeasure (anti-tripping) box, should either or both be required for the Siemens units.

regards

M.