A disadvantage of simple antenna is they do not always perform to specified requirements, such as the required width of the main lobe of radiation pattern (RP). In response, we often utilize straight array (SA) containing N emitters; each of which is employed as a simple antenna. The distance ‘d’ and the current phase difference ‘h’ between adjacent emitters are permanent. Requirements to insure SA are given depending on F (θ) – normalized RP multiplier of SA to the angle θ of spherical coordinate systems. The basic requirement for F (θ) is to provide a given direction of a single main lobe. The additional requirements are smaller width of the main lobe and smaller levels of side lobes. We can provide the basic requirements for F (θ) with differing combinations of values for (d, h, N) of SR parameters, thus additional requirements in each case would differ. This in turn raises the logical question - how to purposefully select the optimal combination of values of (d, h, N), where we can provide the best value of additional requirements? Selecting the optimal combinations by sorting possible options is futile and requires a more focused approach. In this paper we recommend combining RPs to solve this problem. For example, to select the optimum value of N, it suggested to use a combined RP type F (θ, N) with dconst, hconst. As a result, one can make a reasoned choice of the optimum value of the parameter N. Similarly we can propose to use other types of combined RP : F (θ, h) with dconst, Nconst; F (θ, d) with hconst, Nconst. To build combined RP we can use suggested method «2D-3D-2D'». For example, to build a combined RP type F (θ, N) RP on a plane is used (2D) F (θ) with Nconst, dconst, hconst. After that the following RP system is placed in a spatial (3D), with x = θ, y = N, z = F. Then in spatial system many similar RP (2D) are placed in the range of values of Na <N <Nb, incrementing with ΔN → 0. As a result we obtain a spatial surface and its projection (2D') on the plane XOY reflects the combined RP we need. Similarly we're formed the other two types of combined RP. Thus, the resulting combined RP on the plane (θ, N), which reflects the breadth of available radiation lobes (main and side lobes) on zero level. Evidently, for small values of N width of the main lobe will be large, which is undesirable. With a further gradual increase of N we’ll receive smaller value of width of the main lobe. But this dependence is far from linear. At first, with increasing of N ( approximately up to 4 ... 6 ) width of the main lobe decreases sharply and then more slowly. The advantage of using of the combined RP on the plane is precisely the fact that it clearly demonstrates the specified dependency and allows consciously to choose the optimal (compromise) value. Even more significant advantage is the usage of combined RP type F (θ, d) with hconst, Nconst. With this RP we can see that for small values of d width of the main lobe will be significant. With a further gradual increase of d we will also receive smaller value of the width of the main lobe. Also, this dependence is far from linear. First, by increasing d the width of the main lobe decreases sharply and then more slowly. But in this case also we can see that the further increase of d we’re starting having side petals, and their width is growing. It is the use of combined RP that can help us reasonably choose a compromise solution that is close to optimal. This solution provides a rather small width of the main lobe with the acceptable level of side lobes. Furthermore, we can see a clear angular difference between the direction of the main and side lobes. In the case of small angular difference we can allow the presence of minor side lobes, and increasing the angle difference it possible to allow the increase of side lobes. From these examples it is clear that the use of combined RP allows each case to choose a reasonable compromise configuration of SA which close to optimal.