When I was in school (when tiny dinosaurs ruled the Earth) we did an experiment to map nerve endings on the skin. We had someone remove their shirt and lie down on a table, then we took 2 toothpicks and pressed against the skin of the back and asked whether it's 1 toothpick or 2.

When the toothpicks are close together, they activate a single neuron, which the subject reports as a single point. Moving the points further apart, eventually the 2nd toothpick activates a 2nd neuron, and the subject reports feeling 2 points. That's the boundary between the (sensation area of the) neuron of the 1st toothpick and another neuron.

Using this method and a pen, you can draw a map of the areas sensed by single neurons on the back, and various parts of the body. IIRC, the map of the back had rather large (1.5" wide) blotches, so that my back of 14" by 12" would be sensed by about 75 neurons, more or less. A matrix of roughly 8 x 9 neurons reports touch sensation on the back.

So I took a look at the literature today, and sure enough someone mapped of the entire body using this method. Below are relevant results (from that study).

Using the successive stimuli graph (due to motor motion), the theoretical resolution of touch sensors on the skin (the width of the neuron sensing area) is about 1.5cm for the forearm, 2.5cm for the calf, and 1.5cm for the belly (using the numbers for the lower back).

The motors themselves are 10mm discs. It seems that having motors close together will probably not give the wearer a greater sense of resolution, when used on the forearm, calf, or belly.

Source: Mancini F, Bauleo A, Cole J, et al. Whole-body mapping of spatial acuity for pain and touch. Ann Neurol. 2014;75(6):917-924. doi:10.1002/ana.24179