[TowerTalk] Trifilar Winding Question

[Richard (Rick) Karlquist]

Andy wrote:
>> I did a study of this about 10 years ago.  Garden variety ribbon
>> cable has a characteristic impedance of 150 ohms.
> 
> The characteristic impedance is dependent on how the cable is used ... which
> wires are signal, and which ground.  For example, G-S-S-G is different than
> a G-S-G-S-G configuration.

Split off 2 wires from the ribbon to form "twinlead".
Now measure its characteristic impedance the usual way.
That will usually be 150 ohms.  Once you know that number,
you can calculate the impedance of any other configuration,
such as the ones you mentioned.  BTW, G-S-S-G has the same
impedance as G-S-G.  I have wound transformers both ways
and there is no difference.

> 
> Not having done a thorough search ... I recall seeing ~110-120 ohms as a
> typical impedance for flat ribbon some years back, for certain uses.  (I
> think this was old SCSI, not sure anymore.)

The vast majority of ribbon cable is 150 ohms.  BTW, if you want
100 ohms, just scavenge the twisted pairs from a CAT-5 cable.

> OTOH, and again not being very well versed on such things, I'm not sure how
> the characteristic impedance of the wires matters in this case when winding

The characteristic impedance of the wires is extremely
important to the performance of the transformer.
A 50 ohm to 450 ohm transformer wound with ribbon cable
will have a bandwidth of something like 1 to 60 MHz,
with 1.25:1 SWR.  The same transformer wound with 3
wires randomly twisted together to make a "trifilar"
winding will have much less bandwidth.  I did a bunch
of simulations that show that having the correct
characteristic impedance is the "magic" ingredient in
broadband transformers.  This is one of the most poorly
understood aspects of transformer building.

> transformers.  Plus the impedance would surely be reduced by the proximity
> with lots of other metal around it (other coils, the core, etc.).

As I said before, I did extremely careful measurements that show
that the characteristic impedance is not affected by proximity
to the ferrite core.  If these measurements are not done correctly,
it is easy to erroneously conclude that it is affected.  I
have access to an Agilent E5071C four-port network analyzer that is
capable of doing this measurement accurately.  (I design
network analyzers for Agilent).  It is true that metal can affect
the characteristic impedance, however, it has to be practically
touching the ribbon cable.

Rick N6RK


> 
> Andy