After spending most of yesterday researching machine resonance, I chatted with one of the mechanical engineers at work today about the resonance problem. After drawing a picture of the motor/transmission assembly, it became rather obvious to both of us that we have the equivalent of a vibrating violin string. The motor/transmission is attached to the car somewhat rigidly at both ends. The main excitation mass in the system is the flywheel, which sits very nicely right in the middle of the structure. At the natural resonant frequency of the system, the tiny vibrations from the flywheel shake the center of the assembly up and down, just like passing a violin bow over a string.
Since the natural frequency of the system is sqrt(stiffness/mass) and mass is rather hard to change, he suggested adding stiff support bars to the bolts on the adapter plate and bolting the other ends of the bars to strong parts of the car chassis. This would increase the stiffness factor and raise the resonant frequency. This is similar to placing your finger in the middle of a vibrating violin string and having the frequency jump up an octave. The string still vibrates, just at a much higher frequency.
He did lend me an accelerometer which I tied to my analog oscilloscope to measure the three degrees of motion on the motor housing. I didn't get very accurate measurements because I didn't want to hold the system in resonance for too long, but it was good to see that the system underwent 5-6 Gs (5-6 times the force of gravity) during its oscillation at resonance.
On a slightly different note, I also probed the encoder signals inside the position feedback box attached to the DMOC445. In case I can't get a speedometer signal out of the DMOC, I just might risk tapping into the position encoder. It seems like a much easier idea than adding an opto sensor on the flywheel. The pulses coming out of the position box are close to a 50% duty cycle and we get 64 pulses per revolution of the flywheel. This is not surprising since the EE1EncoderPulses variable in the DMOC445 is 64.
At this point, I'm thinking of giving up and just limiting the RPM to 5400 and looking at ways to add support bars to the motor adapter plate later. I just want to get this thing going again.