Towertalk
[Top] [All Lists]

Re: [TowerTalk] Baluns/tutorial/notes.

To: "Tower Talk List" <towertalk@contesting.com>
Subject: Re: [TowerTalk] Baluns/tutorial/notes.
From: "Jim Brown" <jim@audiosystemsgroup.com>
Date: Sat, 22 May 2010 10:02:31 -0700
List-post: <towertalk@contesting.com">mailto:towertalk@contesting.com>
On Sat, 22 May 2010 16:46:58 +0100, Steve Hunt wrote:

>Jim,

>Thanks for that. Your explanation encouraged me to go and re-read your 
>paper. As always it was a "good read". Did I see some significant 
>changes from previous versions:

>1) Earlier versions of the paper stressed that 1:1 current baluns wound 
>with bifilar windings rather than coax would suffer significant losses 
>and heating because of flux leakage - I believe 30%-40% was mentioned. 
>The paper encouraged us to experiment by building a 1:1 balun with a 
>bifilar winding and noting the temperature rise when operating into a 
>well matched load. It was a view which always drew a lot of "flak" if I 
>ever quoted it on a forum.

>All of that material seems to be missing from the latest version of the 
>paper. In fact it seems to say something quite different - that bifilar 
>wound chokes don't suffer high losses and that "When wound with parallel 
>wires, the core sees the sum of flux from currents of opposite polarity; 
>the differential components cancel, leaving only the common mode flux 
>(due to the imbalance in the system)."

>That seems to be a sea change - or did I misunderstand?

You did not misunderstand at all. My earlier statements about high losses 
with differential flux were WRONG, and based on some careless experiments I 
did more than ten years ago before I started working with ferrites in any 
serious way. Based on questions from others, I recently set up careful tests 
with bifilar wound chokes on several types of cores and found losses due to 
differential flux to be VERY small. What I learned that I was mistaken about 
is that there is virtually no LEAKAGE FLUX -- that is, VIRTUALLY ALL of the 
flux from one winding couples to the other, so flux in the core perfectly 
cancels for a differential signal. 

>2) I believe the section on current baluns with impedance 
>transformations other than 1:1 is new. Earlier in the paper you warn 
>quite strongly that the flux in the cores of voltage baluns is directly 
>dependent on the differential-mode signal; I was surprised not to see 
>the same warning relating to current baluns in that new section.

>Again, perhaps I misunderstood .... it has been known ;)

No misunderstanding at all, It is new. I simply did more work and learned 
more things about how these things work. That's one of the great benefits of 
publishing things -- when you miss something or get it wrong, people tell you 
about it and you learn something in the process. :)  I've always liked to say 
that "he who does nothing does nothing wrong."  

If we define a voltage balun as one that couples power from one winding to 
another by transformer action through the core (that's my definition), then 
ANY voltage balun will produce heating at frequencies where the core is 
lossy. #43 and #31 cores are lossy at all frequencies above the AM BC band, 
so they will cause heating if used in a voltage balun. #61 cores are pretty 
low loss below about 10 MHz, but start getting lossy at 15 MHz and above. You 
can see that in the graph of u' and u'' on the Fair-Rite data sheet for each 
product -- u' represents series inductance, u'' represents series resistance.

The beauty of impedance transformation by series/parallel wiring of common 
mode chokes (the Guanella balun) is that they do NOT put differential flux in 
the core like a voltage balun. Each core DOES see the flux, but it shows up 
across the high Z common mode parallel resonance of the choke, so there's 
very little current (and thus very little heating) if the Z is high enough. 

73,

Jim Brown K9YC


_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk

<Prev in Thread] Current Thread [Next in Thread>