[RFI] Filtered connectors for video noise
Wed, 24 Oct 2001 20:00:49 -0700
Don't have any direct experience with filter pin connectors on video cards, but
have used them on other systems and do have a couple observations:
First, I'd suspect that the noise source is inside the monitor, not the video
card. The fact that you're seeing it near electrical wiring is a pretty good
indication that the dominant emissions mode is conducted emissions on the power
line rather than radiated. Since you don't get a lot of other noise from the
computer I'd be very suspicious of the monitor itself. Keep in mind that the
video drive (& possibly sweep depending on what the monitor does when not
connected) waveforms are directly related to the video waveform, & CRT drivers
handle a LOT more power than the drivers in a video card.
The big problem with using a filter pin connector in a application like this is
degredation of the video waveform. For example, an SVGA monitor running 600 x
800 resolution at 72 hz non-interlaced has a dot rate of about 34 MHz, which is
actually higher than the frequencies you're concerned with. If you look at the
fourier spectrum of a 34 MBPS digital signal, you'll see that the spectrum is
essentially continuous until you resolve down to the lowest individual
frequency component, in this case probably the 72 Hz refresh rate. Even a
tight CW filter won't resolve spectral components 72 Hz apart, hence the
appearance of broadband noise.
Anyway, if you try to filter the video signal directly with a low-pass filter
(which is what filter pin connectors generally are), any cutoff frequency that
suppresses components in the HF range will obviously have a serious impact on
the video waveform. Rule of thumb for digital signals is that filter cutoff
should be 3 to 5 times the bit rate. Even though the video waveforms are sort
of analog, you'd still need filter cutoffs several times the dot rate to avoid
fuzzing the edges of individual pixel elements on the display
Again, based on your description of signals on the electrical wiring, it might
be worthwhile to attack the montior (and maybe computer) power line itself.
I've seen lots of instances in my career where signals not related to power
supply switching circuits coupled onto the power lines. A large filter or
common-mode choke (remember to use enough cores/inductance to have a
significant impedance at HF) on either the monitor (most likely) or the
computer (less likely) power cord(s) may have the desired effect.
Let me know what happens.
Pete Smith wrote:
> I have what's gotta be the noisiest video card in the world, in my wife's
> computer. When it's on, I get severe broadband noise (no birdies here!) in
> an HF receiver not only in the room with the computer, but adjacent to the
> furnace thermostat, any electrical conductor, and out in the back yard.
> I'm pretty well fixed on the source because when I unplug the video cable
> from the rear of the computer the noise (on a portable HF radio right next
> to the computer) goes away completely.
> The good news seems to be that virtually all the noise is coming from the
> video card, rather than the switching PS or direct radiation, judging by
> hoiw much good disconnecting does. The bad news is that there don't seem
> to be enough snap-on chokes or cable length to begin to touch the problem.
> I estimate I need at least 30 dB of attenuation.
> The ARRL RFI Book suggests filtered connectors as an option, and a quick
> look at my Mouser catalogue lists male-female adaptor type units for ~$33.
> Obviously, I'd like some idea of whether this is likely to work. They are
> listed as coming in 310 pF Pi, 1000 pf PI and 4000 pf Pi configurations --
> presumably, the pF rating is the total bypass capacitance in a Pi
> configfuration that also includes a ferrite bead.
> Has anyone used these successfully? Is higher capacitance better, or at
> some point does it interfere with video performance?
> Alternatively, would I be better off to first try changing the video card?
> It is a generic S3 PCI video card, and a generic AGP of some sort wouldn't
> be much more expensive than the filtering. On the other hand if the
> filtering would work, I'd get one for the shack too.
> Advice much appreciated.
> 73, Pete N4ZR
> RFI mailing list