Many researchers introduced an array of generators for broadband energy harvesting. The array has been studied in comparison to a single element from this array, but never compared to a single reference harvester with same volume as the whole array. This paper presents a theoretical study of evaluating the performance of the array harvester in comparison to the reference harvester. Power from the reference harvester as well as from the array is analytically calculated. The array is compared to the reference harvester when loaded by their optimal resistances which provide maximum power capability. The comparison is divided into two sections: firstly when the elements of the array are tuned to resonate at matching frequencies and secondly when they are tuned to non-matching resonance frequencies. The comparisons lead to two significant limits of the working bandwidth of the array: the lower and the upper limit. Between the two limits, the power produced from the array is less than the reference harvester, but with a small additional bandwidth. Below the lower limit, the array has no advantage over the reference harvester. Above the upper limit, output power of the array is inconsistent. Hence, design guidelines for the array are provided.