At 08:21 AM 7/29/2003 -0700, Michael Tope wrote:
>Actually, it depends on whether or not you connect the
>shields together at both ends. If you don't connect the
>shields together at both ends, then current can be
>capacitively coupled from the center conductors thru
>the dielectric to the "outside" of the shields. In this case
>the cable will look like unshielded twinlead with "fat"
>conductors of diameter equal to the O.D. of the coax
>shields and impedance determined by a combination
>of the spacing of the two coaxes and the dielectric
>properties of the coax jacket material. If the coax
>cables are physically close (as if taped together)
>then this configuration will have a very low characteristic
>impedance and hence be very lossy.
>
>If you short the shields together on both ends, then the
>electric field between the shields of the coaxes goes
>to zero. In this case, the impedance of the shielded
>twinlead becomes 2*Zo, where Zo is the characteristic
>impedance of the individual coax cable. I believe Jim,
>W6RMK is incorrect that the insertion loss will be
>approximately 1/2 that of the insertion loss of the
>individual coax that makes up the twinlead. Although
>he is correct that the current is reduced by a factor
>of 0.7 in each center conductor (as compared to
>a single coax) there are still two center conductors,
>so the net I^2*R loss is the same. Even if the loss
>is due to I^2*R loss in the shield, the same logic
>applies - the loss per foot is the same as that of
>the individual coax cables.
Mike is exactly right... Looking at loss as a dB/foot for the coax, you
divide the power into two pieces of coax, so the absolute loss (in watts)
will be half in each piece of coax, but you've got two coaxes, so the total
loss is exactly the same. (That's what I get for calculating in my head at
night, doh!)
I think it would be useful do a rigorous analysis of this transmission line
loss thing, particularly for the paralleled, seriesed, open wire line, etc.
cases. I've been fooling with various schemes where there's a lot of
reactive power circulating in lines to phased array elements.
>Because the impedance of the shielded twinlead
>is 3 to 4 times lower than twinlead or ladder line, this
>type of line will generally be more lossy than twinlead
>or ladder line having the same center conductor
>diameter. If you use matching transformers to keep
>the VSWR low, the loss should be tolerable. If not, you
>can still get decent loss performance by using very low
>loss coax. CATV hardline would be a good choice as
>it would yield a net characteristic impedance of 150
>ohms in a twinlead configuration.
Here's an interesting question.. Open wire line is reputed to have very low
loss compared to coax. Why is this? Is it because of the high Z, so low
currents? The dielectric loss in coax <30 MHz is pretty low. What's the
typical wire size on open wire line (RG-213 is AWG13)? Is the skin depth
different in a coaxial configuration compared to the twinlead/quadroline
configuration? Any EM fields folks want to take a crack at a rigorous
analytical description of the fields?
|