problem seteering a phased array

Hello,

I've been working on a 1x8 linear phase array. The antenna system is working perfect, but I came across with a problem when making the algorithms for steering.

Imagine your phase shifters give a maximum of 400º shift for a particular frequency. Now suppose the required shifts for the eight elements are for example: 0 90 180 270 360 450 540 630º.

I programmed an algorithm that used the periodicity of the signal so those shifts are the same as: 0 90 180 270 360 90 180 270. I moduled all phase shifts to 360.

When testing the radiation pattern in a anechoic chamber, the result was disturbing. I got a maximum in the desired direction but the whole pattern was very smooth, with less lobes that it should have, much less directivity.

Then we used another algorithm that worked. Our phase shifter are switched lines shifters, so in this new algorithm we worked with temporal retards instead of thinking of degrees. The thing is... it works.

Then my question is, why doesn't it work moduling the shifts to 360? shouldn't the pattern be the same??

I'll be thinking about it,
thanks a lot in advance

Hello,

What algorithm you are speaking about ? to steer the linear array there is very simple and exact formula ?


> When testing the radiation pattern in a anechoic chamber, the result was
> disturbing. I got a maximum in the desired direction but the whole pattern was > very smooth, with less lobes that it should have, much less directivity.

What is the inter-element spacing ? how large is the bandwidth ?

> Then we used another algorithm that worked...

Again , what algortuhm ? for what ?

> Then my question is, why doesn't it work moduling the shifts to 360? shouldn't
> the pattern be the same??

It works for sure..

The system needs enough isolation between elements. The coupling between antenna elements shall be greater than 15dB, ideally greater than 20dB.

d is less than landa/2.
the bandwith is 2 to 6 GHz.
We used switched lines so the same configuration worked for the whole bandwith.
One algorithm worked. We put shifts suchs as: 100, 200, 300, 400, 500,...
In other algorithm, we modulated to 360º and the shifts we put are: 100, 200, 300, 40, 140,...
The algorithm where we modulate doesn't work.
A friend thinks that modulating to 360 is equivalent to reducing the inter-element distance, but I think something we made is wrong. what do you think?

The results were that the pattern was pretty smooth, with the maximum in the right direction, but with no secundary lobes, as if the sum of many grating lobes had happened. In the right pattern, we got 7 secondary lobes, pretty well shaped and with good beamwith and NLS.

hope I have explained my self.
what do you think about modulating the phase shift to 360? It should work fine, right?????

thanks guys

> d is less than landa/2.
> the bandwith is 2 to 6 GHz.

WOW, a very large bandwidth !!
what is the central frequency -> if 4 Ghz and its the inter-element spacing is 0.5Lambda, you have at band edges 0.25Lambda and 1Lambda !!! this can explain the large difference in number and topology of side lobe reginon !!
Moreover at 2GHz the elements esemems to me too close each othe
r (high copuling) and at 6GHz too far (grating lobes in scanning)

What is your range of scanning angle ? Do you need to move the beam on several pointing angle or just at a fixed direction ?

Concenring the excitation phase this comments applies:

if you use "fixed phase" (i.e. not frequency dependent) the beam squints with frequency change, and the beam can be largely different wrt the predicted one !
Using delay lines (for a fixed pointing angle) you overcame the problem !
the former is the case where you use "Rounded phase shifter" the latter the case where you use "delay lines".

Thus is not a problem of algorithm (still not well clear what you means with this word), but of bandwidth application. So for your final HW implementation what you need is "real true delay lines" and not phase shifters

I hope this can help you.

those problems are solved. I used switched lines, so the steering works for the whole range of frequencies.

In fact, the antenna has been made and tested.
BUT, we some steering algorthims. One of them, used modulation to 360, and that one didn't work.

Check out the above messages. I want to know if anyone know if there could be anyproblem when modulating to 360. The phased shifter is of switched lines and can get up to 500º or so for 2GHz, but we decided for an algorithm to modulate to 360. I know this could only work in a smaller bandwith (think about it if you want). Modulating to 360 doesn't let you steer in one direction for the whole range 2 to 6. But, I measured the pattern in one frequency, i think it was 3GHz, and I calculated the control bits for 3GHz. I want to know if you think that modulating to 360 can generate grating lobes.

Thanks guys

It has to work, I didi it several time this approach during test activities.