[TowerTalk] Common Mode Choke

[Jim Brown]

On Sun, 12 Nov 2006 23:11:42 -1000, Kimo Chun wrote:

>I subsequently ordered 40 suppression cores of the Fair-Rite
>31 mix, part number 2631803802. Though they are of the 2.4 inch
>"toroid" variety they are categorized as "emi suppression cores",
>unlike the 43 mix which is used and listed in both categories. 

>I built four filters following W1HIS design using 41" of RG-213

W1HISs advice on the use of chokes to reduce noise and 
interaction is right on target, but his advice on how many cores 
and how many turns are needed is wild overkill. The reason is 
simple -- both you and he are measuring with inferior test 
equipment that gives a rather pessimistic number for impedance. 
Also, W1HIS seems not to appreciate the value of the resistive 
component of the impedance in noise suppression. 

>28.5 MHZ    Z = 220     X = 83 degrees
>21.2 MHZ    Z = 312     X = 83 degrees
>14.150 MHZ  Z = 498     X = 82 degrees
>7.100 MHZ   Z = 1233    X = 74 degrees
>3.65 MHZ    Z = >1500   n/a, beyond range of meter on both.
>1.83 MHZ    Z = >1500   n/a, beyond range of meter on both

You will find FAR better measured data for chokes wound on both #
31 and #43 cores on my website. They were done in an excellent lab 
by another member of this list who chooses to remain anonymous to 
avoid problems at work. 

See http://audiosystemsgroup.com/AESPaperFerritesASGWeb.pdf

and  http://audiosystemsgroup.com/SAC0305Ferrites.pdf 

which includes more tutorial information and more data. 

I concur with W1HIS that a total Z on the order of 4-5K is a good 
target for noise suppression. The overkill in his advice is that 
he recommends far more chokes than are necessary, because his 
measured data is poor. I got the same inaccurate results when I 
was measuring with an inexpensive antenna bridge like he (and you) 
used. 

For example, to achieve Z> 4K from 160 through 10M, all that is 
needed is 14 turns around one #31 toroid, in series with about 6 
turns around either #31 or #43. 

>As specified this mix works well at lower frequencies. I am
>not sure about the stray capacitance that comes to play in
>the design plus measurement technique. 

That is the major source of your measurement error. 

I am also not sure about the losses for this versus using 43.
>Still learning.

W1HIS is right on target in his analysis that a current balun 
(which in most implementations is a bunch of ferrite chokes in 
series) will fail destructively due to thermal runaway in the 
ferrite and overheating of the coax it surrounds IF the IMPEDANCE 
is not sufficient to reduce the common mode current to a very 
small value. But it is entirely OK for this IMPEDANCE to be 
resistive -- IF it is great enough. Why? Because losses increase 
proportionally with R, but as the SQUARE of the current. 

>Interaction between the antennas (at least interference-wise)
>was minimal in our full-power tests, today. With the addition
>of this common-mode choke in each line it was reduced to an
>almost unnoticeable level. 

Again, I think this is very good advice.

>I will likely proceed with building many more but with the
>43 mix for the higher bands. I will still try the 31 mix
>choke as well to compare the two. 

The #31 mix was developed several years ago. We tested the first 
pre-production samples of 2.4 #31 cores for the papers on my 
website, and subsequently ordered 1,800 of them as a group 
purchase for a half dozen ham clubs spread between CA, IN, IL, WI, 
and MN. I believe our order was their first of that new part.

As you can see from the data, the #31 mix has the major virtue of 
significantly higher R below about 5 MHz than the #43 mix, but is 
nearly equal to the #43 mix up to about 50 MHz. So above 7 MHz, it 
doesnt matter whether you use #31 or #43 -- they are equally 
effective. But if you need suppression on 160 or 80, the #31 mix 
is greatly superior. 

Another VERY important point. The RESISTIVE component of a ferrite 
choke will ALWAYS reduce the current, but the REACTIVE component 
can RESONATE with the wire that it surrounds and INCREASE the 
current. My research has shown that a ferrite choke is basically a  
parallel resonant circuit. For suppression, an ideal choke would 
be VERY lossy. For use as a transformer, a ferrite core should 
have very LOW loss. A choke balun (also known as a current 
balun) is NOT working as a transformer, thus loss is a GOOD thing 
-- IF there is enough of it. 

The #61 material used by W1JR has very low loss below about 20 
MHz, and is still pretty low up to 30 MHz. This makes it far more 
suitable for winding transformers that must handle power. 

>In the last two weeks we also successfully (we hope) took out
>mild interference to two neighbors computer speakers and boom-box
>using 43 mix "snap-it" cores on AC cords and speaker lines.

I work in pro audio, am chair of the Technical Committee on EMC of 
the Audio Engineering Society, and Vice-Chair of the EMC Working 
Group of the AES Standards Committee. You will find many research 
papers and tutorials on my website about RFI and audio systems. 
The tutorials on the pin 1 problem are particularly important -- 
this is the MAJOR cause of RFI in most consumer gear, and of 
nearly all RF in the shack problems. 

http://audiosystemsgroup.com/publish

The tutorials on my website are primarily aimed at pro audio. Im 
currently working on a version specifically aimed at ham radio. 

Someone else asked:

>>I have tried commercially made baluns in the 
>>past only to have them burn out at legal limit power levels.  

Yes, this is exactly the problem of not enough impedance and 
thermal runaway. 

73,

Jim Brown K9YC