[Elecraft] Kio2 cablength

[Jim Brown]

Hi guys, and thanks for the good questions. 

Add me to the list of those who strongly believe in connecting the shields of 
BALANCED interconnecting cables at the sending end only at low frequencies, and 
through a capacitor at the receiving end if the cable is a significant fraction of 
a wavelength at the frequency of an interfering signal. But it's important to 
understand WHY we need to do that -- the pin 1 problem!  

The key word here is BALANCED. Serial connections are UN-balanced, so any voltage 
drop on the return is added to the signal. So one thing we've learned in the audio 
world is to use as much beef as we can in the return of unbalanced circuits. If 
you reduce the resistance of the shield by half, you reduce the hum/buzz by 6 dB. 
Reduce the resistance by a factor of 5 and you reduce hum/buzz by 14 dB. Bill 
Whitlock and I both recommend using video coax (8241, 8281) that has a beefy 
copper braid for unbalanced audio connections, and this is one reason why. 

Another point, that I think we talked about in Dayton, Don, and that I've talked 
about on the reflector, is that the pin 1 problem is a major reason that we NEED 
to prevent that shield current. If you can eliminate the pin 1 problem, you've 
eliminated much of the coupling of that shield current into the equipment. The IR 
(or IX) drop is still there, but reducing the shield resistance helps that. 

Reminder of what the pin 1 problem is: A cable shield SHOULD go to the shielding 
enclosure of the equipment at each end. In the old days, that was easy -- the 
connector shell was bonded to the chassis by virtue of its mounting hardware. 
Today, in the interest of making things easier and cheaper to build, connectors 
are terminated to the circuit board, and a circuit trace connects the shell to the 
chassis eventually. But that circuit trace has both R and L, and other circuitry 
is also connected to that trace. So any shield current (like RF current induced on 
that cable shield, and any power-related hum/buzz that results from voltage 
differences between the connected equipment) flows on that trace and there's a 
drop across the trace. That drop gets injected into the circuitry based on the 
whim of the circuit layout specialist, and we've got RF in the shack (and buzz in 
the audio). 

Now, if you were using SHIELDED CAT5 (which is hard to find and I do have some), a 
case could be made for connecting the shield at only one end, and using the pairs 
as I've described, and with their returns tied to pin 5. But that shield doesn't 
do much below about 20 MHz -- I've tested it with the K2/100 feeding that long 
wire at full power all the way up to 10 meters, and it only improves stability 
slightly on 15m and more on 10m. And that's with the wire running within a foot or 
so of the RS232 cable!  I haven't done any testing of the long wire interface at 
high power (KW amp) above 80 meters, because the wire doesn't work well enough as 
an antenna above 80m for me to want to use it. :)  

All of the RS232 cables I've built for my rigs have the cable shields tied ONLY to 
the DB9 shell on both ends. So far, I've built interfaces for a TS850 (also works 
on the 791A VHF/UHF radio), Omni V, Icom 746 (which I use on 6m and 2m), and my 
K2's. On the radio end of the 850, the shield goes only to the shell of the DIN.

BTW -- the second reason why you want a beefy shield on coax that encounters noise 
below 1 MHz or so is that the shield cutoff frequency is reduced in direct 
proportion to that resistance! The shield cutoff frequency is the low frequency 
limit of where the shield starts working to cancel magnetic coupling to the center 
conductor -- below that frequency, it provides only electric field shielding. For 
a really beefy double copper braid, the cutoff frequency is on the order 1 kHz. 
For a typical foil/drain shield, it's more like 50 kHz. Also, shielding 
effectiveness above the cutoff frequency is inversely proportional to that 
resistance, so dividing the shield resistance by 10 improves the magnetic field 
shielding at 200 kHz by 20 dB. There's a nice derivation of this in Henry Ott's 
book. 

Another point re: connecting or not connecting at both ends. For the coax shield 
to work to cancel magnetic coupling, it must be connected at each end at the 
frequency of the interference. If the interference is at 200 kHz, you would need a 
capacitor that looks like a very low impedance at 200 kHz. No problem, you just 
need to choose the right value of you have that kind of interference. 

73, 

Jim Brown K9YC

On Thu, 7 Jul 2005 14:49:55 -0400, W3FPR - Don Wilhelm wrote:

>Adding to that grounding 'magical stuff', what I learned in large system
>computer design and testing experience was that shields are only DC grounded
>at one end, and that end is the 'driving' device or in the RS-232 world,
>that would be the box defined as the DCE - in the DTE devices, the shield
>would be grounded through a  capacitor, or alternately left open.  This was
>accomplished in the device wiring itself, the cable shield was always
>connected to pin 1 at both ends.