Getting access to the ballast has been problematic, but I constructed a
noise source using a single MOSFET switching at 75 kHz. This generated
plenty of noise in the 40 meter band where I was having trouble. I
placed my load about 15 feet from the source which is the same distance
that the lamp is from the ballast. I was then able to try different
choking arrangements while listening to the noise on 40 meters and
monitoring the common mode current on the line with a clamp on RF
probe. I found that a common choke on the lamp side or the power side
of the generator would work about the same with about 12 to 15 dB of
reduction. Another choke on the other side of the ballast gave about 3
dB more reduction. I viewed this as a short dipole on 40 meters with
the noise generator being similar to a transmitter. I then modeled this
in EZNEC and got the same results. My chokes consisted of 18 turns of
trifilar wire (I have 3 wires to choke) on an FT240-31 core. I found
that if I wound 9 turns close spaced on one side of the core and 9 turns
close spaced on the other side, I could move the self resonant point
into the 40 meter band, although the Q was low and the resonance was
broad. I got close to 8000 ohms of impedance in the 40 meter band as
measured on my AIM 4170 network analyzer. I have not had a chance to
place these new chokes on the neighbors installation yet, but I am
hopeful it will attenuate enough.
I also built a 50/50 LISN and made some measurements at a hydroponic
store on the Lumatek LK-1000 and found it to be about 30 dB above FCC
part 18 limits in the 40 meter band. These things are really bad. I do
have an open case with the FCC concerning this particular ballast.
If anyone is attempting to locate noise like this using the loop on my
web site, it has been my experience that this kind of noise is almost
always vertically polarized, so the null is broadside to the loop and
the peak is edgewise to the loop when it is held perpendicular to the
ground. I believe this occurs because the house wiring acts as an
antenna, but since it is so close to the ground, most of the
horizontally polarized signal is radiated straight up whereas the
vertically polarized signal is radiated toward you close to the ground.
I have verified this by exciting my house wiring with a signal generator
and going outside with the loop to see if I can identify the direction
of the signal.
On 1/15/2011 10:42 AM, Jim Brown wrote:
> On 1/15/2011 7:50 AM, Tom Thompson wrote:
>> have had some success mitigating the noise with common mode chokes right
>> at the ballasts.
> As I understand it, noise is radiated by wiring on both sides of the
> ballasts, including the fixtures themselves (if, for example, they are
> typical linear fluorescent fixtures). You might consider choking the
> secondary (fixture side) if that is practical. I'm also told that
> shielding the fixtures can help. I would expect the noise radiated on
> the secondary side to emphasize the higher frequencies.
> When winding the common mode chokes, pay attention to their resonance
> with respect to the frequencies where you are experiencing the greatest
> problems. There is measured data for several practical chokes in
> http://audiosystemsgroup.com/RFI-Ham.pdf Pay special attention to the
> data for a bifilar winding on a single #31 core, and try to do the
> equivalent with flexible cord (all three conductors) feeding the
> fixture. Also, consider a larger choke (more turns) in series with a
> smaller one.
> 73, Jim Brown K9YC
> RFI mailing list
RFI mailing list