Sometimes no matter how well your vibration fixture is designed, you absolutely MUST use two or more control accelerometers to get all the way through your test spectrum. Now I’m NOT talking about using multiple control accelerometers to make a bad fixture work, there’s no excuse for that, I’m talking about a large, complex fixture that is designed optimally, but you can’t get around the laws of physics.
If you decide to control on the average of the accelerometers, fine, but understand what is really happening. When you look at that beautiful post test plot that shows you ran the test right in the middle of the spec, it’s an average of what really happening. Now look at the individual plots of each accelerometer. That’s the truth, and what it will show is that at some frequencies in some locations you have over-tested, and at some frequencies in some locations you have under-tested.
If you are concerned about your test specimen breaking due to the over test, then go ahead and don’t average, but instead control off of the peak values. You won’t over-test, but certainly when you look at the individual control plots you will find that there are places where you under-tested. Conversely, if you are concerned about under-test and not qualifying to the specified environment, then go ahead and control off of minimum values. Now your individual plots will show that you tested to the specified amplitudes everywhere, but will also show where you over-tested.
In summary, the “averaged” post test plot or the plot showing the combined peak or minimum, will look nice, but you REALLY need to look at the individual plots to see what ACTUALLY happened.