On Wed, 2011-06-08 at 17:50 +0900, Songsong Gee wrote:
> I have a question about SMA-SMA coaxial cable offered by Ettus
>
> Does this cable have an antenna effect?
> In other words, if I use this cable, receive gain or some performance
> measure is better than what I expected due to this cable becoming a
> kind of antenna.
As was mentioned, the cable itself does not exhibit any significant sort of
"antenna effect" unless it's somehow damaged (or perhaps the grounding is poor
on the USRP's PCB). However, the cable absolutely can act as an impedance
matching device if your actual antenna (or whatever is at the far end of the
cable) isn't 50ohms. Now, if the USRP's impedance (looking back into the
output) is precisely 50ohms, while the input impedance of the cable now changes
with its length, the VSWR is -- ignoring loss (it'll be quite small in short
cable) -- the same and it doesn't usually matter much. In actuality, though,
the USRP's output impedance isn't exactly 50ohms across wide bandwidths, and
hence you are actually matching better or worse to the load based on the length
of the cable and can see a difference in signal strength.
While I haven't actually experienced this myself with a USRP, I have seen it on
other radios to the tune of, say, +/-3dB signal strength differences depending
on the cable length used.
> Currently I'm looking for a certain cable which does not have such
> effect for very wide frequency range including very low frequency
> range i.e. DC to 400-500 MHz
At very low frequencies, the approximation of coax cable's impedance of 50ohms
is actually often not that horribly accurate. On the other hand, since coax
lengths in terms of wavelengths are commensurately smaller as well, the cable
itself tends to become more and more "transparent" and typically non-50ohm
characteristic impedance doesn't matter as much.
The standard "trick" for obtaining wideband input and output impedances over
wide frequencies ranges is to use resistive pads (tee or pi) -- a 6dB pad will
get you at least a 12dBreturn loss, for instance; if you can afford the extra
power to do this, it's by far the easiest way to go. (You'll note that test
equipment like spectrum analyzers are almost always spec'd with an internal
10dB
attenuator engaged!)
"Wideband," "good low-loss matching," and "easy" are a canonical "pick any two"
meme of RF design. (Indeed, you actually hit some fundamental limits in this
game -- e.g., fundamental limits on the Q of electrically small antennas are
depressingly low, and the Bode-Fano matching bandwidth limitations are
significant although typically not as troublesome.)
---Joel
_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
