Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: 8b/10b and EMI (how small the holes)

At 2:35 PM 0/3/17, Rick Walker wrote:
>Dear Ed,
>> Ed Grivna <elg@xxxxxxxxxxx> writes:
>> I think you missed Joe's point.  He makes reference to a 1-mm hole.
>> that is the size hole you would have if you routed only the
>> bare fiber through the hole.  This implies usage of a buried
>> module, or a module with a fiber pigtail.  In either case, the
>> fiber is routed through a metal plate, significantly removed from
>> the LASER diode and driver.  The plate is generally constructed as
>> a buried wall in the chassis, with a second bulkhead used to mount
>> connectors.
>I don't think I missed the point.  I was attempting to show that this
>extremely cumbersome technique is not widely used in low-cost systems.
>Most low-cost modules are not pigtailed.  They have a front-panel
>connectors.  In this environment, it is much more difficult to properly
>shield things.
>Are you suggesting that we should require a very expensive, telecom
>style of construction for low-cost 10GbE ports?  Remember that many
>systems may be built in metallized plastic packages with front-panel
>mounted non-pigtailed modules.

Ed did understand my point correctly.  Thanks, Ed.

I would add one thing though -- the 1-mm or 2-mm hole in the metal wall
could contain a glass ball lens rather than a fiber, so there is a simple
approach possible:  Laser inside a small metal or metallized cavity, ball
lens filling a hole in the cavity wall, focusing the photons on the tip of
a fiber connector just outside the cavity wall.  Another variation uses a
1-mm diameter GRIN lens in the cavity wall to match laser to fiber.  GRIN
lenses are cylinders, and are somewhat easier to handle than balls.

At 10 Gbaud, it may be necessary to use more metal and less plastic to pass
EMI.  The definition of "low-cost" may change as the fundamental frequency

Nor is it such a bad idea to confine the 10-gig signals to a small shielded
box, rather than expecting a large but inexpensive (read: sloppy) sheet
metal enclosure to be tight enough to contain such short-wavelength

I would also point out the lowly microwave oven.  The key seal in a
microwave oven is the door gasket, which must work even if the door is bent
and/or closed on a towel.  Current-day ovens operate at 2.54 GHz, with
something like 500 watts of RF power confined in the cooking chamber, with
only something like 10 or 100 milliwatts escaping.  (I don't recall the
exact requirement.)  Although some ovens in the 1970s used waveguide choke
seals, most now use a simple capacitive seal, which consists of a wide
metal flange covered with some kind of thin plastic film.  When the door
closes, the door flange and the doorframe flange together make a capacitor,
actually a waveguide beyond cutoff.  Very robust.  The basic purpose of the
plastic film was to prevent arcing from frightening the public, and it made
the seal less sensitive to manufacturing variations.  Ensuring 100%
electrical continuity all around the door seal, especially after household
abuse, is much harder than ensuring capacitance and waveguide-beyond-cutoff

For 10GbE, such a seal could be as low-tech as some double-stick tape
keeping the bent aluminium-foil shield from falling off.  At low power
levels, the dielectric properties of the film are not too important.