Measurement transducer impulse response using an exponential Sine - - PowerPoint PPT Presentation
Measurement transducer impulse response using an exponential Sine Sweep method Jorge Otero and Ivan Felis Con Content I. I. In Intr trod oducti tion on II. II. Me Method III. III. Ele Electromechan anic ical t al tran ansducer
Jorge Otero and Ivan Felis
º
The impulse method, which ch is based on recording measuring signal, can be used to obtaining transfer funct ction of common elect ctroacoustic c systems and their impedance
ccurring in the ultrasonic c transduce cers can be observed and diagnosed in detail when the impulse method in associ ciation with digital proce cessing of the signal. One of these methods with sweep sinus requires an anech choic c ch chamber in order to delete the influence ce of the ro room and get a good signal-to to-no noise r
. On t n the he o
her ha r hand, us nd, using ng a a qua quasi-anech choic c ch chamber give the po possibi bility o
the he m measur urement i in a n a s small r room. . In In this paper er, the e method proposed ed by A. Farina to get the e impulse e res esponse e in rooms is implem emen ented ed in elect ctromech chanical transduce cers and an ultrasonic c transduce cer through exponential sweeps. For several reasons, sweeps are a far better ch choice ce for transfer-funct ction measurements than noise sequence ces MLS.
Jorge Otero and Ivan Felis.
In order to test the method in twice ce sorts of transduce cers, an anech choic c ch chamber was used to reduce ce the number
ctions in the walls, floor, and ce cell in the case of elect ctromech chanical loudspeaker transduce
case with piezoelect ctric c ce ceramic c PZT a sweep sinus was emitted in a water medium. In both cases, to avoid reflect ctions from surface ces the signal has been cu cut before the first reflect ction acc ccording to the size of each ch room. Al Also, in both cases es, the e same e method to obtain the e impulse e res esponse e was used ed.
Jorge Otero and Ivan Felis.
The signal emitted depends on the frequency cy response of the transduce cer evaluated whereby an exponential sine sw sweep is generated in a width band suitable ! " = $%& 2()
*
ln )
*
. /
* 0 12 3
4
3
5 − 1
Where 89 and 8: are the initial and end frequencies of the signal respectively and ; represent the time.
Once ce the signal has been emitted by the device ce under test (DUT) and record in digital system acq cquisition (DAQ), the spect ctrums of the record and the inverse signal emitted are multiplied to get the impulse response.
Jorge Otero and Ivan Felis.
General scheme of the process Th The pre-ri riging ng effect ct produce ced at low frequency cy due to to sound car can be correct cted by:
! " = $$% ℎ(") → *(")
+,-. /(0) +,-. /(0) 1/ 0 23(0) → 4 5 = 6$$% *(")
The filter is gets in the frequency cy domain as a co conjugated operation where 7(8) is is a a small, all, frequency cy-de depe pende ndent r regul ulari rization pa n parameter. . Finally, an IFFT brings back ck the inverse filter to ti time me domain.
A sweep sinus measurement with a single micr crophone was performed. A loudspeaker Gen enel elec ec 8010A w 8010A was s set u up
ce separated from floor 1.6 meters. A rece ceiver is a Behringer ECM 8000 micr crophone separated 1 meter of the source ce
Jorge Otero and Ivan Felis.
Sample rate: 48kS/ S/s.
Signal sweep from 20hz to 20kHz.
Audio sound card.
Signal calibrated to 91dB with white noise Th The in instan antan aneous sweep power should ld be controllab llable le acc ccording to the frequency cy just being swept through. Wi With this, s, it avoids s the sa saturation on the so sound card.
The proce cess of impulse measurement was assessed in a couple of RESON TC 4034 transduce
en environmen ent is a water er tank with a water er volume e of 0.64 m&. B . Both s h similar s r sens nsors a are s sepa parated d 30cm each ch
cquisition system was used with a sample rate of 10 ⁄ +, ,. .
Jorge Otero and Ivan Felis.
Th This is sweep was as recorded for -../0 an and in in order to suppress all all reflect ctions a time window of 12./0 was used to extricate the direct ct si signal from the file before the first sc scatter appear and delete the elect ctromagnetic c signal from the elect ctric c components. Th The exponential ial sweep sig ignal al em emitted ed has a duration of 200us.
As a result of the experiments, the impulse response from device ces has been calcu culated acc ccording to the method descr cribed.
Jorge Otero and Ivan Felis.
IR of the loudspeaker in an anech choic c ch chamber. IR IR of under erwater er sen ensor.
Impulse e mea easurem emen ent of a mech echanical and ultrasound transducer cer can be e got with the e method des escr cribed ed in this paper er. Th This m made p possible t to d derive a a g general p procedure f for c calculating t g these r responses, w which c can b be u used f for d different tr transducer.
With h the he expo pone nential ial sweep p technique hnique, it it is is po possible ible to simult imultane aneous usly ly de deconvolv lve the he line linear ar impuls impulse respo pons nse of the he sy system and to selectively separa rate each impulse response corresponding to the harmonic distortion orders rs considered. Wit With h this his in informa matio ion, n, it it is is po possible ible to de determine mine, for example ample, the he kind ind of sig ignals nals sho hould uld be be sent to eac ach h trans ansduc ducer whe hen n th the signal expected is known.
The i impulse r response d deconvolution p process i is r realized b by l y lineal c convolution o
the m measured o
with t the a analytical in inverse filt ilter pr processed d form the he excit itatio ion n sig ignal.
ing line lineal al convolut lutio ion n allo allows tim ime-alias aliasing ing pr proble blems to be be avoide ided.
n fa fact, even if the time analysis window has the same length as the emitted signal, the tail of the system response may be lo lost, but but this his will ill no not in introduc duce tim ime alias aliasing ing.
Experimental with th piezoelectr tric tr transducer shows th that t it t is possible to suppress reflecti tion from given surface of th the ex experiment and obtain good results spite of the increase of the noise. In order to perform an objective comparison of the im impuls pulse respo pons nse qualit qualitie ies, the he optim imal al sig ignal nal-to to-no nois ise ratio ios ac achie hievable able for eac ach h trans ansduc ducer ha have be been n compar mpared. d.
Jorge Otero and Ivan Felis.