Problems with inverse Transfer Function Filter

[phono]

Hello,

I'm working on a inverse-TF-filter for compensating attenuation of my system.
Description of the System is:

Labview(sweepsignal(mag 1, 200-20Hz, 30sec):Fileoutput) - Amplifier - Shaker(B&K) - Tube - PCB-acc-Sensor - Labview(input)

The idea is:
measuring the Transfer Function with white noise.
inverting the TF
bring it back to time-domain by iFFT (impulse response)
convolution with sweep-signal
output filtered sweep file

I now expect a constant magnitude while measuring the vibration from the filtered signal.

But I received a signal which contains some kind of modulation with a low frequency.
(I worked only with the magnitude-part of my signal)

I hoped eliminating this effect by adding the phase-shift by complex-calculation but then I receive a signal 10 times higher as I need....

Another problem is, when I turn up the amplification of my shaker in an area that my filter has to attenuate the output signal - I still receive a impuls response which amplificates the sweep instead of scaling down the magnitude...

Does anybody know something about this problem?

 

[VisiGoth]

Make sure you check and then soften the transfer function before inverting it. I assume you have the complex transfer function, and it is not based on the poser spectral of white noise being flat only. (That is you took the ratio of the input and the output functions). If your output signal is not rich (fully populated with enough energy at all the frequencies) you could have a problem. You can manually limit the nulls before inverting. Or just limit the dynamic range of the TF before inverting. This sounds easy but is tricky. Good luck.