First we need to determine whether the spring is indeed flat, and if not, how to make it so. Figure 5 shows the simple test rig to figure this out. My test setup used a drill press to force the spring down, but a hydraulic press, arbor press or milling machine would do the same job. Essentially the same effect could be derived by inserting a long bolt through the thrust-bearing plate and pulling the spring down by screwing it in. Note that you'll have to make the bolt on a lathe since the thread on both ends of the crank is a bastard 5/8" x 16 TPI UNS thread.

Since the diaphragm is very difficult to photograph, a ghost cross section is superimposed for clarity. With the flywheel suspended at the rim, the backing plate can float freely as the spring is compressed. The dial gauge is placed on one of the backing plate bolts and zeroed. The spring can then be compressed until the diaphragm is parallel to the flywheel and the dial gauge read. For the tests reported here, 0.060" of compression was required to flatten the spring. This amount must then be machined off the backing plate posts. Having to take material off the posts usually arises when either of the friction faces have been. To complete the setup, new spacers should be fabricated so that the drive straps are also parallel to the flywheel, especially if there has been material taken off the back of the flywheel for lightening.