K7ITM wrote:
On Aug 4, 11:49 am, Chris wrote:
I've been looking at a few satcom antennas for a project I'm working
on. We were originally looking at parabolic dishes but have recently
taken a look at slot arrays. A low profile solution is desirable
which would lead to a small dish or a low (but very wide) slot array.
Some of the people we've talked to have warned us that slot arrays are
narrow bandwidth. I thought I understood this issue, but I'm
beginning to doubt myself.

Suppose a slot array could transmit at Ka frequencies from 30 to 32
GHz at a bandwidth of 3%. If the center frequency of transmission is
31GHz a 3% bandwidth would be equal to 930MHz or 465MHz on each side
of the center freq. This may be narrow to an antenna designer but
this is more spectrum than we'd ever hope to get on the satellite
anyway. I'm a little confused over why this is an "issue", which
makes me think I have a misunderstanding of the terminology.

Can anyone clear this up for me?

One of the things that occurs to me is that it can make a big
difference just how you feed the array. This is something I've
pondered with respect to linear arrays of dipoles at lower
frequencies. If I feed the array from one end and depend on the
length of transmission medium between the feedpoints, e.g. one
wavelength between each radiator so they're fed in-phase, the phasing
will change linearly with frequency. This is hardly noticeable if the
array is short and therefore has low gain; but if it's long so the
beamwidth is, say, 10 degrees, then a 3% change in frequency (.03*360
degrees) will tilt the pattern by about one beamwidth. In a waveguide-
fed system where the propagation velocity depends significantly on
frequency I suppose things could be even worse.

This is actually used deliberately to steer a beam by changing the
frequency. Consider a radar with a magnetron that can be tuned...



But an alternate way to feed such an array is with a tree of splitters
so that the propagation from the input port to each radiator is the
same time delay. In that case, the radiators will stay in phase and
yield a high gain over a considerable bandwidth. In such a case, you
do still have to account for the fact that the mutual impedance among
the antennas will cause the elements at different points along the
array to represent different loads to the transmission tree "leaves"
and therefore cause different phase shifts, but that's something you
can account for.

And for which phased array designers spend significant amount of time.
There are "sweet spots" that are easier than others, and if you have
lots and lots of elements, then the fact that edge elements have
different neighbors than all the ones in the middle might be ignored.
All depends on the sidelobe requirements, mostly.



I recall once reading about a radar antenna that used a "geodesic
lens" to give equal time delay from the feedpoint to each radiating
element; it was basically a fan-shaped waveguide with a wrinkle in it,
deepest at the middle and flat on the outer edges, which caused the
central transmission path to be the same electrical length as the
outer paths (and all the others).

There's about as many schemes for feeding a phased array as grains of
sand at the beach it seems.