Topband: RFI cure kit

[Jim Brown]

On Thu, 11 Nov 2004 09:51:41 -0500, Eric Scace  K3NA wrote:

>What would you include in such a kit? 

I'll address what I've spent a lot of time studying -- common mode radiation from various 
computer and Ethernet sources. You might ask this question over on the RFI list, where 
guys like Dale, WA9ENA, and EMC engineer for Rockwell-Collins, and others like him, 
hang out. 

First, I would try to do as much of the networking wireless as possible. For whatever 
wired Ethernet remains, I would try to have a selection of the biggest Fair-Rite cores 
you can find in a selection of materials and use them to form multi-turn common mode 
chokes on the wired Ethernet signal and power supply wiring. The materials of choice 
for the lower bands are #78 and #31, while #31 and #43 are the better choices for the 
20-10. 

Fair-Rite 5978003801 is the largest toroid using #78 material (2.4" OD). #5943003801 
is the same thing in a #43.  These all allow lots of turns if you need them. 

#2643626002  is a cylinder 2" OD, 1" ID, 2" long of #43 material. 2631101902 is a 
cylinder 1.122" OD, .543" ID, 1.125" long of #31.  The larger one of these is big enough 
to get several turns of a CAT5 cable through it with connectors attached. The smaller 
one is tighter -- a couple of turns is all you can hope for. 

#0431177081 is a very large clamp-on (1" cable, 2.2" OD, 1.7" long); #0431176451 is 
a slightly smaller one  (0.71" cable, 1.5" OD, 1.87" long. Because these open up, you 
can get more turns through them with the connector attached. The clamp-ons are slightly 
less effective than the solid cores of the same size, and they cost more. 

I would like to list #78 materials in the cylinder and clamp-on forms, but I couldn't find 
any in the Fair-Rite catalog. 

Tom is right on target when he points out that what you are doing with any ferrite choke 
is attempting to form an L-section filter with it, whereby the choke is the series 
impedance and the line has to have some low impedance to short it out. Taking the line 
to a common point is the usual way to do that. But be careful about that -- the higher the 
frequency, the more difficult that is!  

But it is also true that, in general, you want that series impedance to be as large a 
resistance as practical, with as little X as practical, within the range of trash you are 
trying to suppress.  That is, you can also "brute-force" the problem a bit by simply 
making the equivalent series Z of the line that is trying to radiate as high as possible, 
and you do that by adding the lossy ferrite choke to it. It is no trick to get that series R up 
to 1Kohms over an octave or more using a multi-turn choke.

Some general thoughts:   

1) Many of the EMC radiation standards don't kick in below 30 MHz, so that's the range 
where we need to concentrate. 

2) None of this helps you with un-shielded boxes that radiate trash directly. 

3) Some common Ethernet birdie frequencies that will help you tweak these chokes for 
maximum smoke are around 14,030, 21,052, and the bottom 50 kHz of 10 meters. I've 
also identified two in the 30 meter band, but I've suppressed them so well that I don't 
remember exactly where they are. And as I recall, there's also one around 3512 or so. 
There are probably some on other bands and on the phone bands, but I'm a CW guy. 

73,

Jim Brown  K9YC