Dear LF Group,
At my QTH I find there are basically 2 types of "mains" LF QRM - The first
type seems to be harmonics of the mains frequency, and originating from
phase-controlled devices like dimmers and motor speed controls, and also
discharge lamps. If you look at the waveforms in these devices, they have
sharp spikes and ringing at each switching transition, so the harmonic
amplitude only rolls off slowly with increasing frequency, giving a fairly
uniform spectrum over a wide bandwidth. The second type are things like
SMPSUs, which have a "carrier" and harmonics of the switching frequency,
with superimposed modulation at multiples of 50Hz giving wide sidebands.
Tuning around, you can often identify the "carrier" - I currently have one
quite close to 70kHz, with a harmonic at 140kHz. The mechanism of the mains
frequency modulation of the witching frequency seems to lie with the bridge
rectifiers on the mains input. The rectifier diodes only conduct on the
peaks of the mains waveform, and so act like diode switches, switching on
and off at 50Hz, modulating the higher frequency noise currents flowing
down the mains cable.
Most of the noise seems to get from appliance to antenna via mains or other
wiring, and certainly seems to be more of an "H field" mode produced by
inductive coupling between loops of conductor and the antenna - I find that
loops invariably give poor results for this reason if they are near the
building, but once moved several metres away the noise level is much lower.
The noise seems to be able to propagate a long way down the cable at LF, so
mains noise can be high even down the road far away from buildings, if
there is a mains cable nearby. Using a ferrite rod antenna as a probe, and
a battery operated RX there is plenty of noise on the mains cables of my
house, even when the power is switched off completely, presumably coming
from the rest of the neighborhood. In the UK at least, there do not seem to
be regulatory limits on conducted EMI for frequencies below 150kHz, so
there is no particular incentive for designers to reduce LF noise going
down mains cables. For the same reason, mains filters often do not give
much attenuation at LF.
At my QTH, the mains noise sometimes shows up as discrete spectral lines at
50 or 100Hz intervals, however, much of the time allthough you can hear a
strong mains buzz in the noise, it just appears as a fairly uniform, wide
band mush on a spectrogram. It is more instructive to look at this kind of
noise at the receiver IF output in the time domain, preferably with a
fairly wide bandwidth, using a scope triggered from the "line" frequency.
Several peaks can be seen, some of which are synchronised with the mains
waveform, while others move back and forth on the trace. The amplitude of
the peaks seems to vary rapidly in a random sort of way. When I was having
severe noise problems a couple of months ago, I considered the noise
blanker type schemes that have been suggested, but since there are several
(8 or more) noise impulses in each mains cycle, and some of them have
continuously changing phase, this did not seem to be feasible in my case.
It would seem to be hard to filter this type of noise out of the received
signal - even when there are discrete spectral lines, they are usually
drifting quite fast.
Cheers, Jim Moritz
73 de M0BMU
|
