Antenna array gain explanation

Hello,

I am designing a 64 element patch array. with circular polarization and a bandwidth of 12%. my single patch gives a gain of 7dB. I have designed a circular series feed for a 4 element array. when I simulated it i have about -6.7dB in all the outputs and near 90 degree of phase shift for the polarization. However when testing the feed network together with the patches i get only 10.8 dB gain. I was expecting to get something around 12 dB. where could the other 2 dB's be lost?

in the same way the 16 element array using a parallel feed, that shows -6.3 dB in each branch only gives 14.4 dB gain, when i was expecting something about 16 dB. why is it happening?

Thank you!

Hello,

Try to increase the distance between the array elements. The new radiation pattern is the product of the element pattern (your 7 dB single patch) and the array pattern (radiation pattern if all elements are isotropical sources). If the array elements are to close, the array pattern will be too wide.

For the array pattern, the free space wavelength is of importance, not the wavelength in the dielectric. When you make an array on material with effective eps = 4, the elements may be separated by 0.5lambda (based on the PCB material), but based on the free space wavelength, it is just 0.25 lambda.

You can also compare it with two dipoles fed via a splitter, when you place them too close together, they behave as a single dipole (and the mutual coupling will be significant). When you place them very far apart, you get the 3 dB additional gain with many lobes.

complete other thing, did you check the current distribution (phase and amplitude) to make sure you have equal phase and amplitude (to get maximum gain)?

Hi Wim, thank you for the reply!


Actually I was testing with different separations, but it is a bit difficult to get a series feed with good balance for each case, takes too much time to fit electric and physics distances within my limits, my design is being done for a Rogers 5880 substrate, it is in the Ka Band, so my eps=2.2

In my actual best design I have a separation of 5.1mm ( my free space wavelength is 8.75mm) and i get just 11.2 dB I wouldnt like to separate them more cause it creates grating lobes and also i want to fit the 64 array in a maximum 70mmx70mm area.

in the 4array feed i have checked the distribution. the power is split almost equally, there is a small difference of about 0.5dB between the extreme cases, and the phase is shifted between them close to 90 degree, this is done for get a circular polarization. (I will add a picture later on). in the 16 array cases it is lot better, the power and phase are very well balanced.


Ok, then if the reason it is the separation i consider they are already quite well separated more than half of the free space wavelength.
should I really try in separate them more?
what is the disadvantage of increase the separation? less directivity?
in what depends the gain of the array in addition to the element separation?

Thanks again

'Hello,

When you increase the separation too much, gain doesn't increase much, but grating lobes will, so at some point, further increasing doesn't help you. As you have space limitation, the practically obtainable gain is limited by the size and it will reduce if you use feed taper to get better side lobe suppression (in case of the 64 elem. array).

There may be some contribution to the overall radiation patterns because of the current in the feed lines. Did you simulate the circuit with all arrays having its own source also to avoid the feed problems and play with the separation, feed taper, etc? Can the difference between directivity and gain be explained by mismatch loss and copper/dielectric loss?

Regarding the current distribution, did you check the current distribution in the patches itself (phase and ampitude)?

untitled2.bmp

as you can see in the image there are some feed lines that are considerable wide and they could contribute to the radiation pattern. I only did the simulation of the single patches to play with the separation, as the parallel feed is more simple and doesnt have the challenge of get a phase shift it is quite straight forward to design, but maybe it worth to do it.

the dielectric should of course add some losses and having the trade off between bandwidth and gain i am using a regular 10 mils substrate that is not very lossy.
I havent check the current distribution in the patches itself, i just perform the simulation of the feed giving the desired output to the patches, and I think there is some unbalanced distribution produced by the serial feed.

I appreciate your help. this is the first time I am designing such kind of array, (antenna have never been my strength) and i would like to know according to your experience if I have a 7dB gain single patch which gain could I expect from a 4 elements array? are my losses higher than expected, or is it normal to have around 2dB losses?

Once again, Thank you!

I think i have figured out that the "problem" is that I am using the ADS Momentum simulator for this design, and the Gain calculation in the antenna parameters is done with an horizontal polarization, as my antenna is radiating in circular there is a polarization mismatch causing losses in the measurement but not in the antenna.

Can someone corroborated it? how to setup the antenna parameters measurement in the ADS?

Hello,

This can be the problem, so you have to find how to enable circular polarization for the radiation pattern. I don't have experience with ADS Momentum so I can't help you with that. If your antenna produces true circular polarisation, you will gain 3 dB.