[Elecraft] OT Cat 7 Ethernet Cable

[Clay Autery]

On 3/7/2016 9:50 PM, Jim Brown wrote:
> On Mon,3/7/2016 6:47 PM, Clay Autery wrote:
>> Any device putting out RFI as common mode are NOT in spec
>
> Perhaps, but such devices are VERY common in the consumer world. I
> have no contact with the pro world of IT.

Yep... just like a LOT of consumer grade electronics, the manufacturers
intentionally ignore or falsify the FCC certifications.  Generally, you
get what you pay for.

>> A non-inclusive list of causes:
>>
>> Badly designed and/or shielded switch mode power supply (wall wart OR
>> internal).
>
> The Ethernet birdies/carriers I listed are from the data stream, NOT
> from the PSU. Of course the PSU often generates trash, but it's
> DIFFERENT trash. :)

That's a very definitive statement...  while an unlikely cause of
multiple HF region emissions, switch-mode PS frequently switch at
frequencies in/around HF frequencies.

>>   A wall wart can radiate via the DC cord to the device and/or
>> send the signal into the device via the DC cable and then radiate via
>> the chassis or signal cables..  USUALLY internal PSs are adequately
>> shielded... but poor input filters or none at all can let the bad stuff
>> through.
>
> Trash that radiates to our antennas is COMMON mode, not differential
> mode, and common mode is caused by improper return paths, and it
> cannot be "filtered." Filtering affects differential mode trash.

As I said, it is a NON-inclusive list and there are multiple ways to
generate "trash".
>
>> Wireless radios that are improperly designed or shielded can also be the
>> culprit.
>
> I can't see any way that a 900 MHz or 1.8 GHz radio can cause RFI at
> HF. It is the baseband digital signals that have HF components.

A wireless "router" radio is a radio just like any other...  they take
an input signal, massage it and send out a signal...  they are subject
to the SAME design issues as any other radio.  IF frequency(ies),
mixers, oscillators, et al.  Subject to the same mixing products and
improperly shielded can radiate any product, harmonic, etc.
>
>> Not having a separate logical and chassis ground can be a problem...
>
> WRONG -- that is a CAUSE of problems, not a solution. See Henry Ott's
> classic text on EMC.
Perhaps I should have been more precise...  logical ground and a chassis
shield.  Although, there ARE in fact many electronic devices where the
device itself has an isolated logical "ground" and the entire device is
located WITHIN a totally separate shield from end to end essentially
floating the devices on each end and the signals inside their own "space".

>> Shielded cables work best in my experience when you can create an
>> isolated chassis (or supplemental Faraday cage type superior to the
>> chassis (or internal to a plastic case) where the outer shield of the
>> cable and connector housing can be connected in *one continuous shield*
>> around the device(s) and the cable, etc, etc...
>
> EXACTLY WRONG. For shielding to be effective, it must be continuous,
> including a termination to the shielding enclosure at both ends.
> Again, see Ott.

Read what I wrote again.  Chassis/shield to cable connector shield to
cable shield to connector shield on the other end to the
chassis/shield.  One continuous shield...
What may have confused you is that I was considering both well-designed
devices with chassis designs that act as or contribute to the shielding
vs. the majority of consumer grade devices which do NOT have a shielding
chassis...  e.g. plastic case with venting, etc.  MOST of this consumer
grade stuff has 8P8C modular jacks that are designed to accept UTP RJ-45
connectors made of plastic thus they do not provide any means for a
metal shielded connector (a la STP CAT 5/5e/6+ or all CAT 7 and above. 
IF you didn't do anything else, the STP cable and connectors would do
little to assist because they are NOT connected to the
chassis/supplemental shield and have no path to ground.

On a device with a plastic case providing no overall shielding and/or if
the radio shield or PS shield, or any other potential "trash" maker
(common OR differential) circuit is not properly designed and/or
shielded, and/or filtered, or other mitigating method, you're peeing
into the wind by simply adding shielded cables.  You can connect a drain
to ground from the cable/connector housing and that will help...  BUT
what you really have to do to clean up a non-FCC compliant device
putting out harmful/unwanted IF is to either/both/hybrid:

1) Construct an internal shield that will block the radiation from
leaving the enclosure AND connects to the signal cable shields and
thence to ground...  OR, if you can't do it internally,...
2) create and EXTERNAL shield with the same electrical characteristics.
3) Some hybridization of the 1 and 2.

Bottom line...  we said the exact same thing...

>> CAT 7(+) would make an excellent 4 circuit control wire for remotes and
>> the like since it has shield per pair...
>
> Those working in the world of pro audio deal every day with microphone
> signals in the range of -140 dBu, with required dynamic range of at
> least 110 dB, and typically have 24-32 mics active in a given
> recording or production. Any coherent noise that is present in
> multiple inputs will degrade s/n by 3 dB per doubling of the number of
> inputs, and modulation of an RF signal present on multiple inputs will
> be coherent after detection. My point is that pro audio faces a s/n
> problem FAR worse that we do in the radio or computer networking
> world. In that world, we have learned that TWISTING is at least as
> important as shielding, and often MORE important.

On that we agree...  Back in the days of IDE hard disk drives, the
standard was 12-24 inches MAX cable length because of the flat ribbon
cable (and other considerations).  I successfully designed shielded
twisted-pair cables terminated to the 40-conductor internal connectors. 
We got multiple increases in the spec length with no performance
degradation out to many feet and external HDD enclosures (used for
thermal design and other in chassis considerations).  Ultimately, after
tweaking wire size, twist rate, and adding overall shielding (properly
terminated), we were able to remotely locate ALL of the HDD for a 12
board/ 1 master encryption cluster more than 100 feet away with perfect
performance and no RF, cross-talk, or timing issues.  Same thing with
Ethernet cables and others.  You can frequently multiply the spec length
or signal frequency/data capacity using many of the same methods.

Signals are signals... whether they are in audio interconnects, mic
cables, computers, et al.  You do what is required to keep signal
integrity and not cause external interference.

> That said, serious contesters with extensive IT background have
> reported that STP does kill Ethernet birdies, and I believe them. I've
> never heard a suggestion that CAT7 was needed, nor have I ever read a
> spec for any of these cables.

IF they have an "extensive IT background" they'd know a floating STP
cable won't work nearly as good as one where the shield is properly
terminated/utilized.


Have a good evening...

______________________
Clay Autery, KG5LKV
MONTAC Enterprises
(318) 518-1389