[RFI] DSC Alarm Panel 75 Meters/False Alarms

[Jim Brown]

On Wed, 17 Nov 2004 09:06:27 -0500, Pete Smith wrote:

>This raises a question I have wondered about.  Years ago, I had 
>interference to a stereo that originated with the speaker leads.  Not 
>knowing any better, I went to Radio Shack and got a couple of RF chokes and 
>a couple of disk ceramic caps, and put them in series and shunt.  Worked great.
>
>If the sensors on this alarm system are DC, wouldn't a similar fix be the 
>best way to go about cleaning up the sensor lines?

It depends on how the RF is getting in. If it is a pin 1 problem, it probably would not fix it. 
If it was differential mode, it probably would. 

To understand why, I'll review what "the pin 1 problem" is.  A cable shield, if there is one, 
should go directly to the shielding enclosure, not to the circuit board. If it goes to the 
circuit board first, the path from that shield to "ground" along the PCB and other wiring 
will have inductance, and the received antenna current flowing through that inductance 
will put an IZ drop across various points on the PCB depending on the board layout. 
Thus any RF current on the cable shield is injected into the circuitry, where it is likely to 
be detected. 

There are two practical fixes for this problem. One is to move the shield to where it 
belongs -- the shielding enclosure, or, in the case of an unshielded enclosure, the star 
ground point. The other is to choke off the current (or reduce its strength to the point 
where detection either no longer occurs or is no longer audible/causing data errors. 

Now, consider an unshielded cable going to an unbalanced input, with the signal return 
going to "circuit common" on the circuit board. Now you have the same problem, except 
that it is common mode current on the signal pair, not the shield. The solution here is to 
first terminate the circuit return to the shielding enclosure. By virtue of the capacitance 
between the conductors, this will reduce the common mode RF at the input. The second 
part of this solution (without changing the equipment) is to use twisted pair. This 
reduces the loop area, and thus the differential mode RF, and it also makes the RF 
more common mode. And the common mode has been taken to the shielding 
enclosure.  Again, choking the common mode current is a good thing.

There is another mechanism at play with shielded twisted pair cables. It is called shield-
current-induced noise (SCIN), and it converts RF shield current into a differential-mode 
voltage on the signal pair. It happens when the shield current couples asymetrically to 
the signal pair -- that is, more to the red than the black. That most typically happens with 
foil/drain shields, and it occurs because the drain wire is usually twisted at the same 
rate (lay) as the signal pair, and much closer to one signal conductor than the other. 
SCIN is a MAJOR contributor to RFI between about 100 kHz and 10 MHz. Above that 
frequency, shield current symmetry tends not to be disturbed by the drain wire, flowing 
more uniformly in the foil instead.  I use the word "contributor" because the victim 
equipment must contribute its inadequate filtering of the input or output to which that 
cable is connected. A simple RC or LC filter (like what you did) at the equipment would 
fix this, assuming that it didn't also degrade the signal.  But choking the shield current 
also fixes SCIN, and it can't degrade the signal. 

I've published some research on both of these mechanisms in four AES papers, and 
they can be downloaded from the AES website.  


Jim Brown  K9YC