In my opinion the first one-way wheels hold the key to success, assuming that the internal mechanisms in the later ones were based on the earlier ones. Although we know that the Kassel wheel produced about eight bangs on its falling side, we have no knowledge of how many noises accompanied the spinning of the earlier ones - just that a loud noise was produced. I mention this because it might be wise to leave aside any assumption that there would need to be eight bangs to somehow include in the earlier more basic wheel. Bessler implied that he was able to barely induce a wheel to turn with just one cross-bar inside it, which could mean one pair of weights operating within a single but complete mechanism.
So I, at least, continue to work on producing a one-way wheel, but with five mechanisms which I believe Bessler indicated, is the most that can be fitted into the wheel. That indicates to me that the more mechanisms the better - and five seems to me to be the answer, or part of it. So four would not produce as much torque as five and three even less.
Many people work on the theory that because there were about eight bangs on the side towards which the Kassel wheel turned, that fact can be assumed as relevant to the other wheels, but I believe that the earlier ones were simpler with less mechanisms inside and therefore fewer sources of noise. Being of a simpler design they should be easier to replicate - why try to build a two-way wheel when a one-way wheel would prove the point.