I know of Dave. Lots of folks starting their auto conversions have drunk vendor Kool-aid and been bitten by cold, hard facts in the end, including you.
Yes, and there is no reason to see others go down the same road.
A few hours of testing clearly isn't enough to uncover basic design issues which may lurk on a redrive.
A thousand hours of "testing" won't uncover torsional issues if run at RPM outside a resonant range.
Here's an example. Assume an intersection of F1 and firing frequency resulting in peak oscillating torque at 1000 RPM, about 33 hz, an entirely likely ballpark for this system. Builder starts, pushes up to 5000, holds it there three hours. Actual period of peak driveline load was a few seconds when it passed through 1000, not the three hours at 5000.
If the builder passed through 1000 RPM at partial throttle, oscillating vibratory load was nowhere near the potential peak load. Nothing broke, so builder installs engine, goes flying, does lots of touch and goes during test. Out on the runway he moves from idle to full throttle very quickly, a throttle shove. The increased manifold pressure means peak resonant driveline load is radically higher than the part throttle case. Maybe something breaks outright. Just as likely, something begins to fatigue. That first drive? I got to 38 hours before the belt teeth came sailing back past the cockpit.
It's the million monkeys approach, blindly collecting "real world experience". Maybe one of the monkeys writes War and Peace. Maybe not.
Contrast with an analytical model, a list of
measured or carefully calculated inertias and stiffness values created prior to ever reaching the test stand. Condense it to as little three elements, and run a frequency prediction with a test stiffness value for the coupler. Adjust the predicted natural frequencies by varying the coupler stiffness. If necessary, vary other stiffness and inertia values. Then design a drive incorporating the specific inertia and stiffness values which gave the best result. Or, in the case of a purchased drive, select a coupler based on its catalog values, and design the necessary connecting flanges.
Now cut metal and go to the test stand. On the stand, confirm the model predictions. Compare the loads to known component capacities.
Everybody starts somewhere and learns a lot during this journey. I certainly did.
Yes. What I learned is that the analytical approach results in a far higher probability of success. Again, this is driveline engineering, a mature science well over 100 years old. There are no fundamental discoveries to be made, no noble experiments to prove or disprove a new theory. The task is far more mundane...
learn and apply known principles...education and recreation.