[Gb-ccb] RFI screening requirements

Roger Norrod rnorrod at nrao.edu
Thu Aug 21 07:37:33 EDT 2003


At 04:38 PM 8/20/2003 -0700, Martin Shepherd wrote:
>
>For example, at the preliminary design review I asked whether a
>honeycomb RFI ventilation filter would be ok for the cooling fan,
>given that the attenuation of this filter dropped off rapidly above
>10GHz. I was told that this fan-screen would meet GB's RFI
>requirements. Now however, there is concern at GB that the 2.5mm hole
>in an all-metal optic-fibre feedthrough connector would not meet Green
>Bank's RFI requirements. These two data-points appear to be
>contradictory.

Not really.  A single hole would indeed not be a problem, but several
located on a housing face can function as an array and can radiate at quite
low frequencies.   An extreme example: The reason honeycomb filters are
used rather than a flat perforated plate is that the honeycomb has depth,
so the many little tubes act as waveguides below cutoff.  Rather than try
to do detailed EM analysis of a situation with a specific array of holes,
we tend to the conservative approach and try to plug everything.  Bitter
experience has led us there.

Also, the situation with a metal ferrule in a metal fiber connector housing
is similar to that of a loose fitting screw in a metal wall.  Most EMC
texts will tell you those radiate badly, acting like a low-impedance
coaxial line.  Probably the fiber connector would not be too bad because
with the non-metallic fiber coming in and out, there is not much radiating
length, but we're being conservative again.

>
>I am guessing that the difficulty in specifying this revolves around
>the unknown of how much radiation a single-board computer and FPGA
>will generate at microwave frequencies. I don't have the experience
>necessary to make an educated guess at this, whereas I assume that
>people at GB do.
>
>So, could somebody at GB please provide me with a specification for
>the cutoff frequency needed of the RFI screening?
>
>

I'm forwarding this to Jeff Acree, the head of our Interference Protection
Group to see if he can provide some input.

Roger




More information about the gb-ccb mailing list