Bob hit on some very good items (which I also questioned, but didn’t comment due to the fact that many times a photo misrepresents what is actually there).
His first point was that if you draw a centerline from the mounts and extend that out to the wheel, the distance of that centerline, to the mounting point on the lower control arm (as viewed from a vertical, above the chassis perspective, with suspension in the loaded position), you should have an effective arm length that promotes a twisting action on the bar, rather than a vertical bending motion on the bar. The shorter this arm length, the less twisting force is applied (desired), and the more bending force is applied (not desired). In an extreme example, the arm length is zero, and thus no twist takes place at all. BTW, has this bar been bent or changed any way from the OEM use? The reason I ask is that if Bob’s point here is valid, it would mean the OEM application was flawed to begin with, which I highly doubt unless the bar has been flatten somehow. If the bar is un-modified, than moving it further forward from the arm does no good because you'd lose any advantage by introducing angles into the bar's links.
His second point is completely independent of the first one, and that is the distance of the mounts (and this is assuming you have a viable arm length). Look at this from an exaggerated perspective of having both mounts within an inch of each other. In that case, regardless of the bushings used, you’d have a force that would try to rotate the bar at a point centered between the mounts (in essence, the system sees this as a single mount rather than two of them), with no twisting motion provided as desired. As the mounts are moved outward from that exaggerated position, there is a shift from the rotation force to a twisting force on the arms, until you reach the ideal position where both mounts are a distance from each other that equals the distance of the control arm attachment points from each other (again, this is ideal but not practical). Yours are somewhere in the middle of these two extremes.
His third point is that rubber mounts compound all the problems noted above.
Based on your original posting photos, could you not locate the mounts on the horizontal surface immediately under the current vertical plate?. This would allow them to be moved further out laterally (out to the welded vertical brace), where it appears the original OEM mounts may have been located?
What would also help folks here is to get some photos from directly above the assembly, with the assembly position as it would be on the road, as will as directly in front of it, and directly to the side of it (the three planes parallel to the ground, perpendicular, and laterally). The angles of the posted photos, of an unload suspension, make it hard to see if there really is a problem (or degree of problem) here.
