| JP07098262 | TWIST VIBRATION TESTER |
| JP11304635 | WAVEFORM COMPENSATING APPARATUS |
| JP2005091080 | VIBRATION TESTING SYSTEM |
Ming XU. (346 Caren Avenue Worthington, OH, 43085, US)
| 1. | l. |
| 2. | comprising an analog to digital converter convening said vibration signal to a digitized vibration signal at a sampling rate that is sufficient to reproduce a highest frequency ot interest in said vibration signal, and a digital signal processing circuit that receives said digitized vibration signal and performs processing steps to isolate lugez frequency pulses in the digitized vibration signal, and process said high frequency pulses to quantify energy content of the isolated pulses. |
| 3. | The data collector of claim 1 wherein as part of quantifying energy content of isolated pulses said digital signal processing circuit detects a maximum peaktopeak amplitude of said digitized vibration signal. |
| 4. | The data collector of claim 2 wherein said digital signal processing circuit detects a peaktopeak amplitude of said digitized vibration signal by comparine a current sample ot said digitized vibration signal to a peak value and wlien the excursion ot said current sample from zero exceeds said peak value then said digital signal processing circuit updates said peak value to be equal to said current sample. |
| 5. | The data collector of claim 3 wherein said peak va) ue is a peak negative value and said digital signal processing circuit detects a peaktopeak amplitude of said digitized vibration signal by comparing a maximum peakto peak value to a difference between said current sample and said peak négative value, and if the difference is largeur, updating said maxunum peaktopeak value. |
| 6. | The data collector of claim 3 wherein said digital signal processing circuit periodically reduces said peak value of said digitized vibration signal by a factor such that said peak value decays over time. |
| 7. | The data collector of claim 5 wherein said digital signal processing circuit is responsive to user input to alter said factor, whereby quantification of high frequency energy content by said digital signal processing circuit can be adapted to frequency cliaracteristics of said vibration signal. |
| 8. | The data collector of claim 6 wherein said digitai signal processus circuit comprises a software programmable processor performing a sot'twaredetuzed signa) processing procedure, said softwaredefined procedure adapted to obtain said factor from a storage location, and to penult a user to define said factor and store said factor in said storage location. |
| 9. | The data collector of claim 2 wherein said digital signal processing circuit periodically reduces said maximum peaktopeak amplitude of said digitized vibration sisnai by a factor such that said maximum peakto peak amplitude decays over time. |
| 10. | The data collector of claim 8 wherein said digital signal processing circuit is responsive to user input to alter said factor, whereby quantificationn of high frequency energy content by said digital signal processor can be adapted to frequency characteristics of said vibration signal. |
| 11. | The data collector of claim 9 wherein said digital signa) processing circuit comprises a software programmable processor performing a softwaredetuieel signal processing procedure, said softwaredefined procédure adapted to obtain said factor from a store location. and to permit a user to define said factor and store said factor in said storage location. |
| 12. | The data collector of claim 2 further comprising a storage device, said digital signal processing circuit storing said maximum peakto peak value in said storage device. |
| 13. | The data collector of claim 2 further comprising a display, said digital signal processinU circuit displavuy said maxunum peaktopeak value on said display. |
| 14. | The data collector of claim 2 wherein said digital signal processing circuit further performs a frequency transformation upon the maximum peaktopeak value to derive a spectrum for the maximum peakto peak value. |
| 15. | The data collector ot claim 13 where in said frequency transformation is a Fast Fourier Transform. |
| 16. | The data collector of claim 1 where in said digital signal processing circuit comprises a software programmable processor performing a softwaredefined signal processing procedure. |
| 17. | A method of detectuy and quantityuig luili frequency energy in a received vibration signal. comprising converti said vibration signal to a digitized vibration signal at a sampling rate that is sufficient to reproduce a highest frequency of interest in said vibration signal, and receivuig said digitized vibration signal. isolating high frequency pulses in the digitized vibration signal, and processing said high frequency pulses to quantify energy content of the isolated pulses. |
| 18. | The method ot claim 16 wherein quantifying energy content of isolated pulses comprises detecting a maximum peaktopeak amplitude of said digitized vibration signal. |
| 19. | The method of claim 17 wherein detecting a peaktopeak amplitude of said digitized vibration signal comprises comparing a current sample of said digitized vibration signal to a peak value, and when the excursion ut said current sample from zero exceeds said peak value, updating said peak value to be equal to said current sample. |
| 20. | The method of claim 18 wherein said peak value is a peak negative value and a peaktopeak amplitude of said digitized vibration signal is detected by comparing a maximum peaktopeak value to a difference between said current sample and said peak negative value, and if the différence is larger, updating said maximum peaktopeak value. |
| 21. | The method of claim 18 further comprising periodically reducing said peak value of said digitized vibration signal by a factor such that said peak value decays over time. |
| 22. | The method of claim 20 further comprising responding to user input to alter said factor, whereby quantification of high frequency energy content can be adapted to tiequency characteristics ot said vibration signal. |
| 23. | The method of claim 21 furthe rcomprising obtaining said factor from a storage location, and permitting a user to define said factor and store said factor in said storaxc location. |
| 24. | The method of claim 17 further comprising periodically reducing said maximum peaktopeak amplitude of said digitized vibration signal by a factor such that said maximum peaktopeak amplitude decays over time. |
| 25. | The method of claim 23 further comprismg responding to user input to alter said factor, whereby quantification of high frequency energy content can he adapted to frequency characteristics of said vibration signal. |
| 26. | The method of ciaim 24 further comprising obtaining said factor from a storaze location, and permitting a user to define said factor and store said factor in said storage location. |
| 27. | The method of claim 17 further comprising storing said maximum aktopeak value in a stopalpe devise. |
| 28. | The method of claim 17 further comprising displaying said maximumpeaktopeak value on said display. |
| 29. | The method of claim 17 further comprising performing a frequency transformation upon the maximum peaktopeak value to derivea spectrum for the maximum peaktopeak value. |
| 30. | The method of claim 28 wherein said frequency transformation is a Fast Fourier Transform. |
Background of the Invention There are many well known methods for collecte vibration data for performing predictive maintenance. Typically, in these metliods a vibration sensor such as a piezoelectric accelerometer is mechauclicallv coupled to the machine to be monitored. The vibration sensor collects vibrations from the machine and converts these vibration to an electrical signal. The electrical signal is processed by suitable signal processing and converted from analog to digital form. The resulting digital signal is stored for later analysis.
Analysis of a vibration signal from a machine typically involves one or both of (1) comparing that signal to previously collected signals to identify au variations that would be indicative of machine wear and possible impending failure, and (2) forcing a frequency spectrum of the vibration
signal and evaluating this spectrum for patterns indicative of potential failures.
Typically these operations are performed through the use of a data collector.
A data collector includes analog signal processing electronics for receiving a vibration signal and conditioning that signal, and an analog tu digital converter for digital sampling the analog signal so that it may be stored and analyzed.
Signal analysis may be performed using the data collector itself or by uploadinK collected and digitized signals from the data collector to a host computer where signal analysis is performed.
Typically the analog front end of a data collector includes a number of selectable analog signal conditionin circuits, each selectable by controlling an anal£ switch for connectif the element into the signal path througli the data collector. As seen in Fig. 1, a typical data collector might include a first signal condition section in which analog switches 12 and 22 may be used to selectively pass the incoming analog signal through a high pass filter 14 or other anale signal processing blocks 14'and 14" (these blocks may iiclude, e. g., low pass or band pass filters). Furthermore, a second signal conditioning section in the data collector might include additional analog switches 24 and 31 for selectively connecting the incoming vibration signal through a peak detecting circuit 30 or another analog signal processing circuit 30'. After these processing blocks, the signal may be further conditioned, and
low pass filtered, and then converted from analog to digital form by au analog to digital converter, and stored in a digital memory.
As an example of the kinds of analysis performed with a data collector, consider a rotating machine that generates vibration signals at a fundamental frequency that corresponds to the frequencv of rotation of the machine. During normal operaiton, additional vibration signals will be generated at higher ti-equencies as well. These higher frequency vibrations correspond to interactions of mechanical parts while the machine rotates, such as movements of moving parts in bearings, sliding pistons and cams, resonances of machine components, and other normal mechanical activity attendant to rotation of the machine. If, however, a bearing or other mechanical system begins to fail, the part will begin to generate difterent frequency patterns. For example, a bearing may develop a crack, causiy a "click"to occur each time weight is applied to the crack in the bearing, which will be reflected as increased liigher frequency vibration in a vibration signal from the machine. Attemativeiy. a sliding mechanical part may begui to fail and scrape undesirably as it moves, aurai causing increased higher frequency vibration.
It will be noted that collection and analysis of high frequency energy in a vibration signal is otten critical in predictive maintenance analysis.
For this reason, various techniques have been developed for isolating lliUh
frequency pulses or other high frequency information in a vibration signal.
Typically, these methods further iivolve generatiy a lower frequency signal that quantifies the lugh frequency energy in the original signal, so that the lower frequency signal may be digitized and analyzed. Several of these techniques will be reviewed below.
So-called"shock pulse"analysis, developed by SPM Instrument AB of Sweden in the 1970's, uses a special transducer having a tuned resonant frequency at 32 l ; Hz. Thus, tllis transducer is most sensitive to "shock pulse"siYnals in this frequency band, which are often indicative of bearing defects and poor lubrication. The output of the resonant transducer is reflective of energy ui the frequency band of the transducer, and is used to develop a measure on a scale of 1 to 100 of the high frequency energy in the signal. with a va) ue near 100 indicating a failure mode.
The Kurtosis method, developed bv British Steel and the University of Southampton in the 1970's. applies a statistical method to isolated frequency bauds, to develop a statistical parameter indicating the distribution of energy of the vibration signal in these various frequency balids.
This analysis is typically performed on bands from 2. 5-5. 0 kHz, 5-10 kHz, 10- 20 kHz, 20-40 kHz, 40-80 kHz, and a sum is generated of the Kurtosis parameters for each of the tive bands to produce an overall measure of the high frequency content in the signal.
Ail enveloping process has been used by various predictive maintenance companies including Computational Systems Incorporated, SKF, and Diagnostics Instruments. In this process, the vibration signal is rectifie and low-pass filtered, which has the effect of demodulating high frequency energy in the signal to base band : the amplitude of the resulting signal is reflective of high frequency energy in the signa) prior to demodulation.
A filial method, known as'spike energy detection"or alternatively "PeakVue", has been used by the assignee of the present application as well as others to generate a measure of the high frequency energy in a vibration signal. Fig. 1 illustrates the typical anatog circuit components that would be used for performing spike energy detection. In Fig.
1, the data collector has been configure for spike energy detection ; thus, the electrical signal received from the vibration sensor is routed through analog switch 12 to high pass filter 14. As can be seen in Fig. 1. in a typical situation where the spike energy method would be useful, the transducer signal would mclude a low frequency vibration signal (seen in Fig. 1 as a sinusoidal waveform 16), superimposed with brief spikes ofliigh frequency vibration 18.
High pass filter 14 removes the low frequency silusoidal wavefonn from the incoming vibration signal, and passes the higher frequencv spikes 18, resulting in a signal as seen in FiL,,. 1 where the spikes 18 are superimposcd upon a flat low frequency baseline signal 20.
The output of high pass filter 14, after passim through a second analog switch 22, is delivered through analog switch 24 to a decayed peak-to- peak detector 30. Detector 30 outputs a signal reflecting the peak-to-peak amplitude of the signal received at its input. This function is aclueved by a combination of resistors and capacitors with op-amp circuits forcing near- idealdiodes.
Specifically, a first portion of the detector 30 comprises capacitor Cl, resistor Rl and an op-amp circuit that behaves as a near-ideal diode and is therefore illustrated as a diode DI. Through the action of diode D1, upon each négative voltage swing of the signal delivered to detector 30, capacitor C ! will accumulate a sufficient charge to have a peak voltage equal to the negative peak amplitude of the input signal. This charge will discharge from capacitor Cl through resistor RI whenever the input signal is above its peak negative amplitude. The rate of discharge is determined by the time constant formed by multiplying R1 times C1. Typically, this time constant is chosen to approximate the reciprocal of the cutoff frequencv of the liigh pass filter 14. As a consequence, only those fiequencies of interest above the cutoff frequency will pass through capacitor Cl and be delivered to the second portion of detector 30.
The second portion of detector 30 comprises a second op-amp circuit that behaves as a buffer and near-ideat diode, and accordingly is
illustrated as a bufter B 1 anl diodc D2. Thc output of tlus op-amp circuit is delivered to a parallel connection of a capacitor C2 and resistor R2. Due to the présence @e of diode D2, capacitor C2 will be charged to the voltage across resistor Rl and diode Dl whenever that voltage is ureater than the voltage currently across capacitor C2. Thus. capacitor C2 charges to a value representative of the peak-to-peak value of the input signal, comprised of the sum of the capacitor Cl voltage produced by the first portion of detector 30 and the positive peak amplitude of the input signal. Capacitor C2 discharges charge accumulated in this manner through resistor R2, at a rate determined by the time constant formed by multiplying R2 times C2. This time constant is normally chosen to be proportional to the period of repetition of the spikes in the input signal.
As a consequence of the long time constant R2C2, the wavefontt output from detector 30 lias a sawtooth-like wavefom 32 with a substantial low frequency component. This waveform can be readily digitized, and compared with previously recordeecl vibration silnals and/or frequency transformed for analysis, as described above.
A difficulty inherent in the various high frequency energy detection methods described above is their lack of tlexibilitv. In each of the above-described metliods, specified frequency bands of the incoming vibration signal are isolated using special purpose analog circuitry. If different
applications or different machines require the use of ditferent frequency bands, redundant analog circuitry would need to be included in the data collector : i. e., the data collector would need to have multiple high pass filters and multiple peak-to-peak detectors, one for each frequency band of interest. Altemativeiy, the filters and detectors in the data collector may be made adjustable, but this would also require complexity. namely, in the filters and detectors, analog switches would need to be included to select between circuit components of different values, in order to change the frequency bands and time constants of the circuit. Tlus approach may also encounter problems with noise due to the number of switches that are included in the analog signal path.
Accorduigly, there is a need for a data collector and data collection method that is suitable for detection ut high frequency energy in a signal, facilitating use in a wide range ot applications.
Summary of the Invention In accordance with principles of the present invention, this need is met by a data collector having a digital signal processing circuit that receives a digitized vibration signal after the signal has been converted by an ånaloz to digital converter to a digital fonn, and performs processing steps to isolate high frequency pulses in the digitized vibration signal and quantify energy content of those pulses. In accordance with principles of the present invention, the vibration signal is digitized by an analog to digital converter at a
sampling rate that is sufficient to reproduce highest frequency of interest in the input signal. Then, the digital signal processing circuit performs high frequency pulse isolation and energy quantification upon the resulting digital signal.
In the described embodiment. the digital signal processing circuit performs amplitude detection by detecting a peak-to-peak amplitude of the received vibration signal. Specifically, the digital signal processing circuit compares the digital signal to a peak negative value and when the signal exceeds the peak négative value the peak negative value is updated to be equal to the digital signal. Then, the digital signal processing circuit compares a maximum peak-to-peak value to the difference between the digital signal and the peak negative value, and if the difference is larger, the maximum peak-to- peak value is updated. Periodically, the maxi num peak-ro-peak valuc and peak native value are reduced in magnitude by a decay, factor, such that the peak negative value and maximum peak-to-peak value decay over time.
In this described embodiment, the maximum pcak-to-peak value comprises a peak-detected digital output signal that may be stored audlor displayed to a user as a reflection of the high frequency content of the vibration signal. Furthermore, the digital signal processing circuit also perfores a frequency transformation (e.g., a Fast Fourier Transform) upon the
maximum peak-to-peak value to derive a spectrum for the maximum peak-to- peak value for storale and/or display to a user.
This digital methodology for high frequency analysis produces a resulting peak-detected output signal that is easily analogized to those produced using analog circuitry in known data collectors. Accordingly, users that are knowledgeable in the analysis of peak-detected waveforms produced by prior data collectors, may readily reapply this knowledge to the peak- detected digital output signal produced in accordance witli principles of the present invention.
It is a significant aspect of the preseut uivention that the decay factors used in computing the maximum peak-to-peak value and peak negative value are determined by the digital signal processing circuit and may he readily changez. As a consequence, Iiigh frequency processing performed by the digital signal processing circuit may be readily adapted to file particular frequency characteristics of the machine being analyzed for predictive maintenance. Importantly, this adaptation may be made without any modification to the analog circuitry of the data collector.
In the described embodiment, the digital signal processing circuit is in the foriii of a software programmable processor perfonniy a software-detuied signal processing procedure to implement the methods described above. In this embodiment, adjustment of parameters is further
simplified by adapting the software-defined procedure to optai these parameters from a storage location, and to permit a user to define these parameters as desired.
The above and other objects and avantages of the present invention shall be made apparent from the accompanying drawings and the descriptionthereof.
Brief Description of the Drawing The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with a general description of the invention given above, and the detailed description of the embodiments given below, serve to explain the principes of the invention.
Fieu. 1 is a block diagram of a conventional data collector including spike energy analog signal processing circuitry: Fiv. 2 ils a block diagram of a data coliector in accordance with principles ut the present invention for performing digital signal processing of liigh frequency energy in a vibration signal : Fie. 3 is a process diagram illustrating the processing steps carried forward by the digital signal processor illustrated in Fig. 2 ; and
Fig. 4 is a tlow chart of operations performed by the digital signal processor of claim 2 in processing high frequency energy in a vibration signal in accordance with principes of the present invention.
Detailed Description of Specific Embodiments Referring now to Fils. 2. a data collector 40 in accordance with principles of the present invention can be described. Functional blocks of the data collector are illustrated in Fig. 2 to facilitate the description of the operations contemplated by the present invention. It will be appreciated. however, that a data collector suitable for carrying out the pruiciples of the present invention may take many forms. In one embodiment, a suitable data collector comprises a data collector available from the assignee of the present invention under the model name dataPAC 1500. Another suitable data collector is disclosed in U. S. Patent 5. 633, 811 issued to Canada et al., and assigned to Computational Systems Incorporated, which describes a data collector sold by Computational Systems Incorporated, the disclosure of which is incorporated herein by reference. Alternatively, another suitable data collector is a palmtop computer havi lu a sound interface card therein for receivuig digital vibration data such as is disclosed in U. S. Patent 5,808,903 issued to Schiltz et al. and assigne to the assignee of the present application. the disclosure of which is also incorporated herei by reference.
Regardless of the embodiment of data collector that is utilized, the data collector will comprise andalou signal processing circuits 42 for receiving a signal from a vibration sensor such as a piezoelectric accelerometer and conditioning this signal for sampling and conversion to a digital fonmat. The operations perfonned by circuits 42 may include gain and offset correction and various filtering operations. as well as low pass filtering for anti-aliasing.
An analog to digital (A/D) converter 44 receives the output of signal processing circuits 42 and converts this analo signal to a digitized form A/D converter 44 is a high sampling rate Iiigh resolution data converter programmez to sample its analog input signal at a sampling rate at least twice the maximum frequency of interest in the vibration signal. A suitable A/D convener for this purpose is sold by Burr-Brown as part no. DSP 101 or DSP 102.
The digitized vibration signal generated by A/D converter 44 is delivered to a digital signa) processor 46. Digital signal processor 46 may comprise a special purpose digital signa) processor operating under the control of suitable software, a general purpose microprocessor operatin under the control of suitable software, or a discrete 1couic circuit for processing digital signals in the manner described below. Each ot these embodiments is within the scope of the present invention, however, the following description will
focus upon embodiments of the invention in which digital signal processor 46 is a programmable circuit operating under the control of software.
Processor 46 is coupled to a keypad and display 48 for interacting with a user of the data collector 40. Keypad and display may comprise the keypad and display incorporated into a laptop computer, or the touch screen uicorporated into a palmtop computer. or a custom display and surrounding function keys on a special purpose data collector.
Processor 46 is further coupled to a storage device 50 for store digital data. including (where processor 46 is a programmable circuit) software for operating the processor to carry out the methods of the present invention, as well as stored digitized vibration signals and/or processed versions thereof.
Storax, device 50 and processor 46. or either of them. may optionally interface tlirough a liost interface circuit 5 to deliver stored data from storage device 50 to a host computer 54. Host computer 54 may include software for evaluating or archiving digital signals stored by the data collector 40. In such an embodunent, some or all of the analysis steps described in connection with Fig. 3 may be performed in a liost computer upon digital data stored by data collector 40. Alternatively, data collector 40 may include sufficient processing power and digital storage to fully process and archive digital signals collected by the data collector. in which case liost interface 52
and host computer 54 may not be needed. Wliile both embodiments are within the scope of the present invention. the following discussion will focus upon embodimentsof the invention in which digital signal processor 46 is a progranunable circuit operatuy under the control of software.
Host interface 52 may take any of a variety of forms, including a serial interface such as RS-232 or USB coupled to a PC. an Ethemet or other network connection connected through network cabling, a parallel port connection, or other suitable fonts. Furthennore, host computer 54 may comprise a palmtop, laptop, or desktop PC, or a mono-or multi-processor server.
Refcn-uiu now to Fig. 3. an explanation can he uiven of steps performed by the data collector and/or host computer of Fig. 2. in accordance with principles of the present invention. A digitized vibration signal 59 delivered by A/D convener 44 to digital signal processor 46 mat is a good candidate for high frequency processing is illustrated in FiL'. 3. This signal 59 comprises a low frequency vibration signal (seen in Fig. 3 as a sinusoidal wavefomm 16), superimposed with brief spikes of high frequency vibration 18.
Digital signal processor 46 performs an initial routine for filtering and optionally decimating the digitized signal. Specifically, processor 46 perfores a digital signal processing routine upon the received digital signal that implements a low-pass or hand-pass filter. As illustrated at 60 in Fig. 3,
this filtering routine may be responsive to stored parameters available to software ui processor 46 to select a desired frequency band of interest : e. g., the filter may be a baud-pass filter having a lower cutoff frequency selectable between 100 Hz, 200 Hz, 500 Hz. 1 kHz, 2 kHz and 5 kHz, and an upper cutoff freqquency of 65 kHz. Other frequency bands may also be utilized for different applications. Essentially, the lower cutoff frequency must he above the fundamental machine vibration frequency that is reflected by the si lusoidal low frequency signal 16 received by digital signal processor 46. The upper cutoff freequency must be at a sufficiently high frequency to allow the brief spikes 18 of high frequency signal to be passed with minimal attenuation or distortion. The upper cutoff frequency is also limited by the analog ami-alias filtering that is performed on the signal prior to A/D conversion. In the illustrated embodiment, the upper cutoff frequency is 65 kHz.
Band pass filtering performed by processor 46 removes the low frequencv sinusoidal waveforn 16 from the incoming vibration signal, and passes the higher frequency spikes IS. resulting in a signal 61 as seen in Fig. 3 where the spikes lE are superunposed upon a flat low frequency haseline signal 20.
Having thus isolated the lush frequency spikes through band pass tilteruy, ui a second processing step, digital signal processor 46 perfores a digital signal processing routine, described in greater detail below with
reference to Fig. 4. to generate a peak-to-peak detected signal. The peak-to- peak@ detection signal processing routine includes decay time constants causing the resulting waveform to have a sawtooth-like waveform 32 with a substantial low frequency component. This sawtootli-Like digital signa) waveform is analogons to the analog signal produced using analog circuitry in known data collectors. Accordiigly, it can be processed in an analogous manner by displaying its aérage amplitude or time waveform as depicted at 62 or 64 in Fig. 3. or by performing a Fast Fourier Transform or other frequency transform and displaying the energy spectrum for the signal as depicted at 66 in Fig. 3.
Users that are knowledgeable in the analysis of peak-detected waveforms produced by prior data collectors, and in their amplitude, time and frequency domain appearance. may readily reapply this knowledge to the peak-detected digital output signal produced in accordance with principles of the present invention to perform predictive maintenance analysis.
Referring now to Fig. 4, the process performed by the digital signal processor 46 in processing a vibration signal in response to user input can be explained in greater detail. In a first step 70, a user of the system identifies band pass fi-equencies for the digital band pass filtering illustrated at 60 in Fig. 3. Specifically, the user specifies a lower cutoff frequency, e.g., by selecting one of 100 Hz, 200 Hz, 500 Hz. 1 kHz, 2 kHz or 5 kHz, and optionally selects an upper cutoff frequency. In step 70, the user also
identifies the expected frequency tD at which a detect signal is Likely to be seen, and the number of harmonics, n, of this expected frequency that are of interest for display and analysis. The upper cutoff frequency, if it is specified hy the user, must be large enough to pass all of the brief spikes 18, and a confirmation of this may be included in step 70.
In a subsequent step 72, this user-provided information is used to compute parameters for the digital processing that is perfonned in the subsequent processes. Specifically, these parameters include a time constant Tl and a second time constant T2. The cune constant rl is computed as 0. 000324 seconds when the user specifies in step 70 a lower bandpass cutoff frequency of 2 kHz or 5 kHz, and #1 is computed as 0.0069 seconds when the user specifies ui step 70 a lower bandpass cutoff frequency of less than or equal to 1 kHz. The time constant T2 is computed as 2.07/2*#*@D) or 2.07*n/(2*#*fmax).
As will be appreciated, the tirst formula #2 time constant establishes an appropriate decay rate to display the desired defect frequency tD at near maximum amplitude, whereas the second formula for the #2 time constant ensures that n harmonics of the defect frequency tD will be displayed, where fmax is the maximum frequency that will be captured by the A/D converter.
After computiy these parameters in step 72. digital processing commences. First, in step 74, values for the peak négative sample and the
maximum peak to peak value are initialized at zero. Then, in step 76, the digitized vibration signal is received from the A/D converter 44. The signal is then band-passed filtered in step 78 by applying the bandpass filtering algorithm with the frequencies specified in step 70. The resultmg filtered stream of samples are received bv a peak detection algorithm, In step 80, a sample is received, and in a loop comprisuig steps 82, 84, 86.88.90 and 92, this sample is processel. Tliereafter, ii step 96 a subsequent sample is received, and processing returns to step 82 to process the new sample.
In the turst step 82 of the loop identified above, the current sample is evaluated to determine whether it is more négative than tlle stored peak negative sample. If so, then in step 84 the peak negative sample is replaced with the current sample value. As a consequence, the peak negative sample will always have the most negative sample that has been previously encounterecl, or a decayed version thereof.
If in step 82 the current sample is not more negative than the peak negative sample, processing continues to step 86 in which the difference between the current sample and the peak negative sample is computed, and compared to the stored decayed maximum peak to peak value. If the current difference is larger than the stored decayed maximum peak to peak value, the in step 88 the stored decayed maximum peak to peak value is replaced with the cuitent difference. As a consequence. the stored decayed maximum peak
to peak value will always reflect trie largest decayed peak-to-peak value of the incoming samples that has been previously encountered.
After step 84 or 88. or immediately after step 86 if the current difference is less than the stored maximum peak to peak valuc, processing continues to steps 90 and 92, in which steps are taken to apply the decay time constants #1 and #2 to the stored peak negative value and maximum peak to peak vatue. Specifically, in step 90, the peak negative sample is multiplied by a factor exp(-T/#1), where exp () represents the natural exponential operation, and T is the time period between digital samples. Sunilarly, in step 92. the maximum peak to peak value is multiplied by a factor exp (-T/r2).
It will be appreciated that these operations cause the peak negative vatue and maximum peak to peak value to decay over time with an exponential decay characteristic having the time constants #1 and #2, respectively.
After step 92. in addition to receiving a new sample value in step 96, other operations are performed upon the computed vatue for the maximum peak to peak value. Specifically, in step 94, the decayed maximum peak to peak value computed in the preceduig steps is delivered to subsequent processes that display a spike energy time waveform, or compute a Fast
Fourier Transform of the decayed maximum peak to peak value to generate a spectral display. The resulting time and spectral displays are analogons to those produced by prior instrumentation utilizing solely analog spike energy detection circuitry, and thus can be utilized readily by those persons experienced in using prior analog spike energy detection circuitry.
The decayed maximum peak to peak value is also used in step 98 to compute an "overall" spike energy value, analogous to that created by some prior data collectors. In prior data collectors sold by the assignee of the present application, the"overall"spike energy value is created by au analog spike energy circuit. In this circuit, the time constant of the RC t'ilter tomled by capacitor C2 and resistor R2 of Fig. 1. is 0.03 seconds. The output of the spike detection circuit is then fed through a low pass filter having a cutoff at 0.1 Hz. The resultinlT essentially DC value uidicates the"overall"energy of spikes in the input signal. To produce an analogous signal using the present invention, a compensation is performed to correct the time constant T2. specifically. r) is correcte such that it is no less tiiati 0. 03 seconds. To do this, it r) is less than 0.03 seconds, in step 98 the maximum peak to peak value is multiplied by a compensatory factor equal to exp(-T/003-#2)@.
The combined effect of this compensatorv factor and the factor used in step 92 is equal to multiplying the maximum peak to peak value by a factor of
exp (-T/. 03). Thus, if the value of #2 is less than 0.03 seconds, the compensatory step 98 adjusts the maximum peak to peak value to sunulate a T2 of 0.03 seconds.
After the compensation of step 98 is performed, as needed, in step 100 the resultuia maximum peak-to-peak value samples are fed to a digital low pass filtering process, having a cutoff frequency at 0.1 Hz. This is analogous to the low pass filtering used to compute"overall"spike energy values in prior data collectors. Then, the resulting low pass filtered samples are displayed, as a DC value, to indicate the"overall"spike energy in the received vibration signal. This"overall"value is readily analogous to those generated by prior data collectors and can therefore be readily used hy those persons experienced with the analogous values generated by data collectors having analog peak detection functions.
While the present invention has been illustrated by a description of various embodiments and while these embodiments have been described in consiclerable detail, it is not the intention of the applicants to restrict or in any way limit the scope of the appended claims to such detail.
Additional avantages and modifications will readily appear to those skilled in the art. The invention in its broader aspects is therefore not Limited to the specific details, representative apparats and method, and illustrative example shown and described. Accordiigly, departures may be made from such details without departing from the spirit or scope of applicant's general inventive concept.
What is clained is: