Best Circularly Polarized Array[edited]

[g86]

Suppose we are designing a Circularly Polarized (CP) array wich is consists of four sub arrays. Now which array should give best CP performance in terms of CP bandwidth, Axial ratio at center frequency and CP HPBW?

1. Sequential rotation of 4 sequentially rotated sub array
2. Array of 4 sequentially rotated sub array without any sequential rotation
2. Sequential rotation of linearly polarized sub arrays

=========
For simplicity assume all are 4x4 arrays and consist of four 2x2 arrays. I think now there won't be any confusion.
=========


Please give reasons..

:!: :idea: :?:

 

[flatulent]

easier to do it locally

It would be easier to make the circular at the lowest level in your system (individual antenna element). Then the combining networks and lines to make a phased array are not as critical and only have one line.

 

[g86]

Re: easier to do it locally

Treating the sub arrays as elements, your vote goes for linear array of sequentially rotated arrays (CP elements in your case). right?? But why you are not choosing others?

:!: :idea: :?:

It would be easier to make the circular at the lowest level in your system (individual antenna element). Then the combining networks and lines to make a phased array are not as critical and only have one line.

 

[flatulent]

simple is best

My philosophy is simple is best. By having each array element made circular right at its location you can precisely control the circularity ratio and opposite circular polarization rejection quite well.

Then make sub arrays of just a few elements. These are easier to manufacture and test. Then combine a few of these into a larger sub array which is easy to test.

 

[dowjones]

Hi,
I partially disagree with flatulent. The question of g86 is which is the best in terms of XP, and not which is the simplest. My agreement/disagreement points are explined below

1) I agree that the individual elements must be also circularly polarized. If you dont make this, you will have a gain loss of about 3db with respect to the maximum gain you could attain. In order to get this, you can think of single-feed patch antenna elements with perturbations.

2) The subarrays must be sequentially rotated. This improves the XP at off-boresight angles and also improves the input match of the subarray. Although you have to include 90deg transmission lines, the complexity of the network is not dramatic (assumed single-feed patches as in point 1). Then, your subarray gives good XP in a broad coverage and also good input match. Recall also that seq.rot. is a robust technique in respect the feed phase errors.

3) finally, to remove the residual XP which the subarray could not cancel out, the array should also implement sequential rotation. Again, the feed network does only needs additional 90deg lines, which is not so complex in printed technology.

4) I implicitly assume that the subarray is 2x2, which gives broad XP cancellation along the main cuts. please indicate if this assumption is correct.

regards

 

[g86]

I have simulater 4x4 array where sequential rotation of sequentially rotated 2x2 arays have been used and got broad side null. Are you telling the same thing?

:!: :idea: :?:

Hi,

4) I implicitly assume that the subarray is 2x2, which gives broad XP cancellation along the main cuts. please indicate if this assumption is correct.

regards

 

[dowjones]

well, in fact no. i was assuming a linear array of 4 subarrays (2x2 each one)- in other way a 4x1 array of 2x2 subarrays-, following your first post.

what you say however (4x4 of 2x2 subarrays) will give much broader XP cancellation than the linear case.

 

[g86]

I am not understanding why there will be XP cancellation. I was actually telling about 4x4 array (16 elements) means 2x2 array of 2x2 array.

BTW did you ment XP=cross polar? I am confused now.

:!: :idea: :?:

well, in fact no. i was assuming a linear array of 4 subarrays (2x2 each one)- in other way a 4x1 array of 2x2 subarrays-, following your first post.

what you say however (4x4 of 2x2 subarrays) will give much broader XP cancellation than the linear case.

 

[flatulent]

perfect elements

If the basic elements are exactly circular, it should not make any difference how they are combined. What you two seem to be discussing is the case where the basic element is really elliptical and the design of the overall array is done to reduce this effect by rotating the elements. If this is the case, I would suggest that making the overall pattern very fine would work the best. As an example. AAABBB is no as good as ABABAB.

 

[g86]

Re: perfect elements

The combination is directly related to Axial Ratio and combination matters. Even four linearly polarized antennas when make an array with sequential rotation it makes CP. Similarly all combination of circularly polarized antenna may not give you a good axial ratio bandwith.

You must consider:

1. HPBW for CP
2. Axial Ratio at center frequency
3. Axial ratio Bandwidth
4. Impedance Bandwidth

In our case 4th point can be neglected. All other three parameters changes with different combinations.

:!: :idea: :?:

If the basic elements are exactly circular, it should not make any difference how they are combined. What you two seem to be discussing is the case where the basic element is really elliptical and the design of the overall array is done to reduce this effect by rotating the elements. If this is the case, I would suggest that making the overall pattern very fine would work the best. As an example. AAABBB is no as good as ABABAB.

 

[dowjones]

ok g86,
my thoughts were in the direction which indicates flatulent, in the sense that the sequential rotation improves the axial ratio of a single element (elliptically polarized). I was not considering lineal elements because there is a 3db gain loss with respect to an array made with circularly polarized elements.

other thing is the array configuration. any seq.rot. configuration gives ideal 0 crosspol (XP) at boresight. This is got because the crosspol adds out of phase at theta=0. however the cancellation depends on the spatial delay between elements, and worsens with the angle. that is what i meant when talkin about broader XP cancellation. by the way, g86 in your first post you talken about a linear array (to me this means Nx1), while now it seems that your interest is a planar (MxM). that confused me and maybe my comments have confused you.

on the other hand, and as flatulent notices, if the array is composed of circ. pol. elements with ideal XP, it has no sense to use seq. rot. since it does not provides any improvement.

 

[g86]

ok g86,
my thoughts were in the direction which indicates flatulent, in the sense that the sequential rotation improves the axial ratio of a single element (elliptically polarized). I was not considering lineal elements because there is a 3db gain loss with respect to an array made with circularly polarized elements.

I am still not understandig the loss you are talking about.

other thing is the array configuration. any seq.rot. configuration gives ideal 0 crosspol (XP) at boresight. This is got because the crosspol adds out of phase at theta=0. however the cancellation depends on the spatial delay between elements, and worsens with the angle. that is what i meant when talkin about broader XP cancellation. by the way, g86 in your first post you talken about a linear array (to me this means Nx1), while now it seems that your interest is a planar (MxM). that confused me and maybe my comments have confused you.

I see. It should be linearly polarized array. May be that is the source of everybody's confusion.

on the other hand, and as flatulent notices, if the array is composed of circ. pol. elements with ideal XP, it has no sense to use seq. rot. since it does not provides any improvement.

I am editing the 1st post. It will be linearly polarized array and not a linear array. Sorry friends ...

:!: :idea: :?:

 

[JoeSchmo]

Rotation of CP Elements is Desirable...

if your element has an unbalanced feed.

 

[Lupin]

If your requirement si just low Array Axial Ratio, I'ld use a sequential rotation of sequentially rotated elements. The single element must have at least 2 input ports... To have a low AR at single element level is a good starting point...

Regards
Lupin

 

[dowjones]

ok g86, ill try to explain the 3db loss issue. I recognize that I think usually on patch antenna elements (meybe this is not your case, but it is useful for my explanation).

suppose that your patch elements have a single feed and radiate linear polarization. if the linear polarization gain is G(dB), your circular polarization gain is G-3 (dB).

Extending this, if you have 2 elements rotated 90deg, the gain of your array will be 3 + (G-3), i.e. G dB referred to a circular polarization standard.

On the other hand, suppose that the patch element has 2 inputs so that it generates a CP (circ.pol) wave. Then the gain of your single element is G (dB) in CP, and if you make an array of 2 elements (rotated or not), the gain of your array is 3 + G.

Therefore, the gain of your array with LP elements is 3 dB below the gain of the array with CP elements.

Again my starting assumption is that the single element is capable to radiate a CP wave.

I hope the loss issue is more clear with this explanation

regards

 

[g86]

ok g86, ill try to explain the 3db loss issue. I recognize that I think usually on patch antenna elements (meybe this is not your case, but it is useful for my explanation).

suppose that your patch elements have a single feed and radiate linear polarization. if the linear polarization gain is G(dB), your circular polarization gain is G-3 (dB).

Extending this, if you have 2 elements rotated 90deg, the gain of your array will be 3 + (G-3), i.e. G dB referred to a circular polarization standard.

Very correct

On the other hand, suppose that the patch element has 2 inputs so that it generates a CP (circ.pol) wave. Then the gain of your single element is G (dB) in CP, and if you make an array of 2 elements (rotated or not), the gain of your array is 3 + G.

But this G is not same as previous G. Earlier one was LP gan and this G is CP gain and it is in practical almos three dB less than LP G. These two G s are not same

:!: :idea: :?:

 

[dowjones]

maybe it is not exactly the same, but it is practically the same. recall that if a linear polarization antenna has a gain G in linear polarization (wrt a LP standard), the same antenna with two feeds and 90deg phase has a gain G in circular polarization (wrt a CP standard)

(wrt: with respect to)

 

[dowjones]

(sorry i did not finish my previous post)

... and therefore, if we call G to the gain of the CP antenna measured wrt a CP standrad, then the gain of the same antenna with a single feed(LP) measured wrt a CP standard is G-3 (dB)