With the recent release of APM 2.9x for Arducopter, reducing vibration has become a top top priority. This is because inertial navigation is being implemented, starting with an inertial PID loop in altitude hold. Vibration problems show up in altitude hold and any mode that requires the use of the altitude controller. I was having problems with my quad climbing uncontrollably in forward flight (while in altitude hold) and then descending again when forward motion stopped. I thought it had something to do with transnational lift or air blowing into my APM case, but Randy (developer) helped me determine that the root cause of the problem was in fact vibration.
By enabling RAW logging I was able to see the acceleration forces on the quad in flight. The current thought is that during normal flight (not just a hover - you need to include data from high power settings) the Z acceleration should generally stay between -15 and -5. As shown by the green line on the graph below I was well outside that. I've got a motor that really whistles while flying, so I would guess it has a bent shaft. All my props are meticulously balanced. The red line shows the current vibration level the APM feels after I made a simple mod to the mounting of the APM. By no means was this my idea, but it works great and I highly recommend anyone having trouble with altitude give something like this a shot. Here is the wiki on Vibration Control. By importing the log data to Excel I was able to determine that the standard deviation of the red line is 1.4. I learned how to analyze flight data here. The second video discusses using Excel.
I previously had my APM mounted on a high quality gyro mounting tape, the same kind that the xBee is mounted on in the second picture. The ONLY modification I made to achieve the lower vibration to the APM was to switch to supporting the APM, receiver, xBee and a 42 gram plate with four really squishy earplugs. The diameter of the holes I drilled for the earplugs are a .25". The distance between the plate and frame is 7/32". I admit, 42 grams is A LOT of weight to just add to the quad, but the results are excellent. The all up weight of the supported structure is 111 grams. I first saw this mounting method here. The earplugs hold very well, I can actually pick up the whole quad by the APM and not have it all come disconnected. I messed around with using a little CA glue to hold the ear plugs in more permanently and found that thick CA and some accelerator did not attack the foam at all.
I hope this can help someone who has had trouble switching to 2.9.1.