In order for that ground wire to be part of a filtering
system, that wire must be short. Take a portable shortwave
receiver. Connect directly to earth ground - inches. Signal
is substantially reduced if not totally cut off. Connect same
shortwave receiver to same ground with 50 feet of copper
wire. Now that ground connection become a perfectly good
receiving antenna.

Grounding the filter through 50+ feet of safety ground wire
does not earth the filter. Instead that 50 foot wire would
become a transmitting antenna for any RF noise that leaked
out.

Many filters require that ground wire because of leakage
currents from AC electric. A line filter, like other low pass
filters, does not require an earth ground to be functional.
That filter simply acts as a dam - to stop higher frequency
signals from passing out of power supply into any AC wires.
Any signal that gets through that filter is then connected to
a good transmitting antenna - those AC wires. Purpose of
filter and of that 'lump' ferrite bead on monitor cable are to
stop (block, impede) RF signals from escaping the computer.
Connection to safety ground is necessary in some designs to
bleed off leakage currents. But a 50 foot connection to earth
ground would only act as a transmitting antenna.

V W Wall wrote:
larrymoencurly wrote:
V W Wall wrote in message ...

Is a ground needed for the input line filter to be effective even
if the filter doesn't connect to ground, as is the case with many
Antecs, including my 300W SmartPower?

The input line filter in the PS connects to "chassis ground",
i.e. the metal PS enclosure.

How can my PSU's filter connect to ground when it has no any
low-value Y-rated disk capacitors from line-ground, just X2
capacitors across the lines? It looks a lot like the filter in
this 380W Antec TruePower (lower left):

http://terasan.okiraku-pc.net/dengen/no48/true/open.jpg

and has a .47 uF yellow boxy capacitor across the AC receptacle,
a choke in series with one of the AC lines, followed by a .22 uF
capacitor across the lines, then a choke in series with each AC
line (both wrapped around the same core), and finally another .22
uF capacitor across the lines.

It's hard to tell from the picture. See:

http://www.pavouk.comp.cz/hw/Alim_PC_ATX_200W_fr.pdf (page 1)

for the kind of circuit I was refering to. Note the two 47 mmfd
capacitors with the mid point connected to the ground (shield).
In an actual PS, some heat sinks are connected to the case ground,
and some are floating. Note the center point of the two main
filter capacators connected to ground thru a varister.

The actual enclosure is bolted to the computer case so it is
always at the line ground wire potential.

Another whose filter that doesn't connect to ground is on page 12
of this example 90W ATX PSU (C1, C2, LF01):

http://us.st.com/stonline/books/pdf/docs/6412.pdf

There doesn't seem to be any ground (shield) here. You're correct,
the filter capacitors are between the hot line and neutral. There
are three different ground symbols in the diagram. It's hard to
figure which is which.
I assume the "three pronged" ones donate case ground.

RFI prevention is an emperical kind of thing. Each situation is
unique. I have seen several examples of the building wiring
acting as a good antenna, especially when not properly grounded.

Virg Wall