(TRF) Module of the KLIPPEL R&D SYSTEM S7 FEATURES Combines linear and nonlinear measurements Provides impulse response and energy-time curve (ETC) Measures linear transfer function and harmonic distortions simultaneously Quasi-anechoic measurement due to windowing of impulse response Fast two channel data acquisition (< 43,6 khz) with noise floor monitoring Automatic SPL calibration Highly adjustable stimulus (bandwidth, spectrum, crest factor) Calculates time delay, minimal phase and group delay Provides cumulative spectral decay (CSD) and sonograph Overlay of up to 20 result curves The TRF module measures two signals simultaneously, and determines the magnitude and phase of the linear transfer function and the harmonic distortion at the same time. The stimulus is a logarithmic sweep with adjustable spectrum, bandwidth and crest factor. Window techniques (gating) applied to the impulse response allows a separation of the direct sound from early reflections, diffuse field and nonlinear artifacts. Finally, post-processing provides the time delay (zero delay plane), minimal phase, group delay and several frequency-time transformations (e.g. cumulative spectral decay and sonograph). Output 1 Stimulus Generator + Shaping Fundamental Windowing Calibration 1 Channel 1 FFT Transfer Function - FFT -1 Postprocessing FFT Channel 2 Data Acquisition Calibration 2 Reference Windowing Harmonic Distortion Article Number: 1000-900 CONTENTS: Excitation Signal (Stimulus)... 2 Signal Acquisition... 2 Post-processing... 2 Graphical representation... 3 Result Windows... 4 Limit Values... 5 Applications... 5 Patents... 8 Document Revision 1.0 updated March 30, 2017 Klippel GmbH Mendelssohnallee 30 01309 Dresden, Germany www.klippel.de info@klippel.de TEL: +49-351-251 35 35 FAX: +49-351-251 34 31
S7 Excitation Signal (Stimulus) Stimulus Frequency band Resolution Voltage Shaping of the stimulus spectrum Signal length The module uses a logarithmic sine sweep excitation signal. The logarithmic sine sweep is an optimal excitation signal as it gives the highest signal to noise ratio and produces no modulation distortion. It allows to separate the linear transfer response and harmonic distortion components. The linear transfer response can so be purified from nonlinear artifacts. Furthermore frequency plots of the individual harmonic distortion components can be extracted from the measurement data. The start frequency fmin and the end frequency fmax of sine sweep can be selected in the range from 0.05 Hz to 43,6 khz. The user may specify the analysis resolution or the corresponding duration of the sweep. The required sample frequency and signal length are selected automatically according to the desired frequency band and resolution. The user may specify either the voltage at the output connector OUT 1 (OUT 2) or the voltage at the terminals of the speaker connected to output SPEAKER 1 (SPEAKER 2). In the later case the amplifier gain is determined at 750 Hz without load prior to the main measurement and the excitation level is adjusted accordingly. The magnitude of the stimulus spectrum can be arbitrarily shaped by the user (e.g. to attenuate low frequencies to protect tweeters). For this a shaping curve may imported from the clipboard. For security reasons the shaping curve is automatically scaled before applying it to the stimulus. The scaling limits the maximal shaping factor to 0 db, i.e. the stimulus amplitude will never be increased. The shape of the spectrum determines the crest factor of the waveform. The default shape of the stimulus spectrum yields to the optimal crest factor of 3dB. Maximal 128k samples synchronous to FFT length Signal Acquisition Signal acquisition Channel 1 Channel 2 Calibration curve Calibration factor Noise floor monitoring Averaging Automatic SPL calibration Continuous loop measurements Two-channel acquisition synchronous to stimulus and FFT length (max. 128k samples) Signal at connector IN1 (microphone) or voltage at speaker terminals (of Speaker 1 or Speaker 2) Signal at connector IN2 (microphone) or current at speaker terminals (of Speaker 1 or Speaker 2) or signal of laser displacement sensor A calibration curve can be imported from the clipboard for each signal type that can be measured on channel 1 and channel 2. The user can specify a calibration factor for converting physical units to db for each signal type that can be measured on cannel 1 and channel 2. Prior to the main measurement a noise floor measurement can be performed to monitor system noise and to ensure sufficient excitation. The measurement may periodically be repeated 2-256 times. The resulting waveforms are averaged to reduce noise. An automatic procedure is provided to calibrate the TRF for SPL measurements with a pistonphone or calibrator. All the user has to do is to enter the sound pressure level produced by the pistonphone. The user can choose between actually calibrating the TRF and validating the current calibration. A special mode of operation allows repetitive measurements without starting each measurement individually. The time between the individual measurements can be specified by the user. Post-processing Spectral Analysis Windowing The spectrum of the signals acquired on channel 1 and channel 2 is calculated and plotted together with the noise floor. Rectangular, Cosine, Hanning, Hamming, Blackman and Kaiser windows are provided as full (symmetric) and half (asymmetric) windows respectively. The window is appied to the selected range of the impulse response. Furthermore it is used to calculate the timefrequency transformations. KLIPPEL R&D SYSTEM page 2
S7 Denominator and nominator of transfer function Impulse response (linear + nonlinear) Energy-time curve (ETC) Linear transfer function Fundamental + Harmonics Harmonic distortion Reference Time delay (zero delay plane) Minimal phase Excess phase Group delay Cumulative spectral decay (CSD) Wigner distribution Sonograph Several transfer functions can calculated from the measured data (without measuring again), i.e. signal channel 1 / stimulus signal channel 2 / stimulus signal channel 1 / signal channel 2 reciprocal of the above transfer functions The module calculates the impulse response for the selected transfer function. Using two markers the user can select a section of the impulse response for analysis in order to separate direct sound from early reflections and diffuse field. Due to the logarithmic sine sweep excitation the linear and the nonlinear response are separated (if stimulus is selected as transfer function denominator). Using the markers the nonlinear response can be excluded and the transfer function purified from nonlinear artifacts. Magnitude of the envelope of the impulse response given in db. In this representation the direct sound can usually be determined more clearly than in the impulse response. The transfer function is calculated for the selected section of the impulse response. The following representations are provided: Magnitude (Bode plot) Phase (Bode plot) Nyquist plot ( imaginary part vs. real part) Due to the logarithmic sine sweep excitation the transfer function can easily be purified from nonlinear artifacts. Magnitude of fundamental, harmonic distortion components (2 nd to 24 th order). Ratio of harmonic distortion components (2 nd to 24 th order) and the fundamental The measurement can be referred to some reference measurement. For this the transfer function can be divided by a reference curve imported from the clipboard. Furthermore the measurement can be referred to a scalar reference value given in db. Time delay present in the transfer function. The time delay can be determined automatically or specified by the user. The time delay can be converted to the corresponding distance = time delay * sonic speed. Minimal phase of transfer function calculated by the Hilbert transformation. Excess phase of transfer function (phase - minimal phase - effect of time delay) Total group delay (negative derivative of transfer function phase) Excess group delay (negative derivative of excess phase) CSD plot of transfer function. Illustrates the decay of the individual frequency components after exciting the system with a sine that is suddenly switched off. Wigner distribution of the transfer function Sonograph of the transfer function. A short symmetric data window is used to separate the impulse response into joined sections. The plot shows the spectra of the sections vs. frequency and time. Graphical representation Spectral Analysis Transfer function Curve post-processing: Spectra can be plotted integrated (RMS sum) over IEC standard 1/3 octave and octave bins respectively or without post-processing. Noise floor: The result of the prior noise floor measurement is plotted together with the spectral lines of the signal Effective resolution: The effective frequency resolution depends on the window length and type applied to the impulse response. In all transfer function related plots results are shown only for frequencies above the relative resolution. Curve post-processing: Transfer function related curves can be plotted KLIPPEL R&D SYSTEM page 3
S7 averaged over IEC standard 1/3 octave and octave bins respectively, smoothed (moving average) over 1/n th octave (n=1,2,,99) or without post-processing. Harmonic distortion Phases Units: Transfer function related curves can be plotted in real physical units (linear or logarithmic), in db or in level meter style (fixed x-axis from 10 Hz to 40 khz, y-axis in db with dynamic range fixed to 50 db). in percent or in db (100 % corresponds to 0 db ) wrapped unwrapped Result Windows Impulse Response Energy-time curve H(f) Magnitude Fundamental + Harmonics Harmonic distortion Reference for H(f) H(f) Phase H(f) Nyquist H(f) Minimum Phase H(f) Excess Phase Shows the measured and the windowed impulse response. Using two markers, a section of the impulse response can be selected for analysis, e.g. to exclude room reflections. The response comprises the fundamental and 2 nd, 3 rd and higher order harmonics separated in the time domain. Magnitude of the envelope of the impulse response given in db. In this representation the direct sound can usually be determined more clearly than in the impulse response. Magnitude of the transfer function Shows the magnitudes of the fundamental, the harmonic distortion components (from 2 nd to maximal 24 th order). Ratio of harmonic distortion components (from 2 nd to maximal 24 th order) and fundamental of transfer function in percent or db respectively Reference curve for the transfer function that can be imported from clipboard Phase of transfer function Nyquist plot of transfer function (imaginary part vs. the real part). Minimum phase of transfer function Excess phase of transfer function. The excess phase is Φexcess = phase - minimum phase - effect of time delay. H(f) Excess Delay H(f) Total Delay Cumulative Decay Wigner Distribution Sonograph Excess delay of transfer function. The excess delay is the negative derivative of the excess phase. Total delay of transfer function. The total delay is the negative derivative of the transfer function phase. Cumulative spectral decay for transfer function H(f). The CSD plot illustrates the decay of the individual frequency components after exciting the system with a sine that is suddenly switched off. Wigner distribution plot for the transfer function Sonograph of the transfer function. A short symmetric data window is used to separate the impulse response into joined sections. The plot shows the spectra of the sections vs. frequency and time. KLIPPEL R&D SYSTEM page 4
S7 Stimulus(f) Spectrum Spectrum of the stimulus Y1(f) Spectrum Spectrum of the signal acquired on channel 1 Y2(f) Spectrum Spectrum of the signal acquired on channel 2 Calibration Curves Stimulus(t) Shows the imported calibration curves. Calibration curves can be imported from the clipboard. Waveform of stimulus signal Y1(t) Waveform of signal acquired on channel 1 Y2(t) Waveform of signal acquired on channel 2 Table Results + Settings Collection of result and setup parameters Limit Values Parameter Min Max Unit Minimal frequency fmin > 0 < fmax Hz Maximal frequency fmax > fmin 43.6 khz Resolution 0,05 187,5 Hz Sweep time 0.005 21.8 s Averaging 2 256 Output Voltage at OUT 1 and OUT2 ±9 V peak Input Voltage at IN1and IN2 ±10 V peak Voltage at Terminal SPEAKER 1 300 V RMS Current at Terminal SPEAKER 1 20 A RMS Applications Stimulus Stimulus (t) Stimulus (t) vs time KLIPPEL 1,0 0,5 0,0 [V] -0,5-1,0 0 50 100 150 200 250 300 Time [ms] The module uses a logarithmic sine sweep excitation. This excitation signal gives the highest signal to noise ratio and produces no modulation distortion. If no spectral shaping is applied to the stimulus the crest factor of the time signal is automatically adjusted to the optimal 3 db. KLIPPEL R&D SYSTEM page 5
S7 Shaping of the Stimulus -45 Signal lines Spectrum of stimulus KLIPPEL -50-55 -60 [db] -65-70 -75-80 5 10 20 50 100 200 500 1k 2k 5k Frequency [Hz] The stimulus spectrum can be arbitrarily shaped by importing a shaping curve from the clipboard. This is for instance useful to adapt the stimulus to the noise floor (by increasing the excitation level for noisy frequency bands). Another application is to attenuate low frequencies to protect tweeters. Spectral Analysis S p e c tr u m o f s i g n a l Y 1 ( C h a n n e l 1 ) S ig n a l a t I N 1 [ d B ] 0 d B = 1. 0 0 V - 6 0-7 0-8 0-9 0-1 0 0-1 1 0 S ig n a l lin e s N o is e f lo o r SNR K L I P P E L - 1 2 0-1 3 0 5 10 20 50 100 200 500 1k 2k 5k F r e q u e n c y [ H z ] A noise floor measurement can be performed prior to the main measurement. Both noise floor and main measurement are subjected to a FFT and plotted in the same graph. This way sufficient excitation and the signal to noise ratio can be checked for the whole frequency band and disturbances (humming etc.) are identified easily. Impulse Response I m p u l s e r e s p o n s e h ( t ) H ( f ) = S i g n a l a t I N 1 / S t i m u l u s 5 0 0 0 M e a s u r e d W i n d o w e d K L I P P E L 2 5 0 0 [ V / V ] 0-2 5 0 0 fundamental 5th 4th 3rd harmonics 2nd - 5 0 0 0-7 5 0 0 0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 T i m e [ m s ] Using two markers the user can separate the direct sound response from early reflections and diffuse field. This way quasi-anechoic measurements can be performed. Due to the logarithmic sweep excitation the linear and the nonlinear response are separated (note the nonlinear harmonic response between 200 ms and 300 ms). KLIPPEL R&D SYSTEM page 6
S7 F u n d a m e n t a l + H a r m o n i c d i s t o r t i o n c o m p o n e n t s H ( f ) = S i g n a l a t I N 1 / S t i m u l u s F u n d a m e n t a l T H D N 2 n d H a r m o n i c 3 r d H a r m o n i c 4 t h 5 t h 6 t h 7 t h 8 t h 9 t h 1 0 t h 1 1 t h 1 2 t h K L I P P E L - 2 5-5 0 [ d B ] - 7 5-1 0 0 5 0 1 0 0 2 0 0 5 0 0 1 k 2 k F r e q u e n c y [ H z ] Due to the separation of linear and nonlinear response the transfer function can easily be purified from nonlinear artifacts. Furthermore the individual harmonic distortion components can be extracted from the measured data. Impedance Electrical Impedance H(f)= Voltage Speaker 1 / Current Speaker 1 17,5 15,0 12,5 10,0 Magnitude KLIPPEL [Ohm] 7,5 5,0 2,5 0,0 50 100 200 500 1k 2k 5k 10k Frequency [Hz] The two channels can be used to measure voltage and current at transducer terminals. Choosing voltage and current as transfer function nominator and denominator the impedance is calculated. Time-Frequency Transformations 313.0 Cumulative spectral decay H(f)= Signal at IN2 / Stimulus 100 db - [V / V] 75 50 0 Time [ms] 250 500 50 100 200 1k 500 2k 20k 10k 5k Frequency [Hz] 20 750 The CSD plot (cumulative spectral decay) illustrates the decay of the individual frequency components after exciting the system with a sine that is suddenly switched off. Box resonance and other unwanted modes are so detected easily. KLIPPEL R&D SYSTEM page 7
S7 S o n o g ra p h H ( f) = S ig n a l a t IN 1 / S tim u lu s 38 30 22 14 6-2 - 1 0-1 8-2 6-3 4-4 2-5 0-5 8-6 6-7 4-8 2-9 0-9 8-1 0 6-1 1 4-1 2 2-1 3 0-1 3 8-1 4 6-1 5 4-1 6 2-1 7 0-1 7 8-1 8 6-1 9 4-2 0 2 [ d B ] 0 d B = 1. 0 0 V / V K L IP P E L 2k 1k 500 200 100 Linear response Nonlinear response 0 100 200 300 400 500 600 T im e [m s] The figure left shows the Sonograph of the transfer function. It provides the spectral contents of the impulse response vs. time. It is for instance useful to identify the direct sound and the linear and the nonlinear response. Patents Germany 102009033614, P10214407 USA 12/819,455, 7,221,167 China 201010228820.8, 03108708.6 Find explanations for symbols at http://www.klippel.de/know-how/literature.html Klippel GmbH Mendelssohnallee 30 01309 Dresden, Germany www.klippel.de info@klippel.de updated March 30, 2017 TEL: +49-351-251 35 35 FAX: +49-351-251 34 31 KLIPPEL R&D SYSTEM page 8