new web: http://bdml.stanford.edu/pmwiki
TWiki >
Rise Web>ClimbingRobot > ClimbingConcepts > SpinedSurfaces>SurfaceProfileAcquisition (09 Dec 2004, MarkCutkosky)
Rise Web>ClimbingRobot > ClimbingConcepts > SpinedSurfaces>SurfaceProfileAcquisition (09 Dec 2004, MarkCutkosky) Link to Alan's surface profile calculations: SurfaceProfileSimulation
Constant relative displacement of the surface with respect to the sensor is necessary.
This has been achieved using the smaller white belt with motor with adjustable speed provided by AMcClung. At low speed, the motor vibrates too much and surface acquisition as an error unacceptable for our purposes. The final configuration has been improved using a long screw as indicated by prof. MarkCutkosky driven by a 24 V motor with gear 600 RPM, operating @ 6 V. Here the stroke is 5 cm. Again delimited by tape. This last configuration requires a lot of care (not touch the table, verify that motor-screw connections made with tape is ok), and it allows an accuracy in the order of the Optical profilometer instrument. Transfering the motion from motor to slide is a very critical aspect, for vibrations, irregular motion.
The Omron sensor is installed on the small sliding guide from Micro Slides model A-2030 with the magnetic support. The sensor is moving, because the big weight of concrete bricks does not allow to put them on the small slide (the most accurate available in the lab). For this reason the acquisition distance changes because the direction of motion of the slide is never parallel to the concrete surface, so there is always a slope in profiles. This does not affect the parameter calculation because only micro-variations are considered.
The magnet of the sensor support is also used to transfer the motion from the motor to a nut rotating on the screw. The screw has a ratio of about 1.5 mm per revolution. The actul speed is automatically calculated by the acquisition program, because the acquisition time is know and all profiles are at constant distance.
The connection between motor and screw is done by clamping a rubber tube on the motor shaft and on the screw. It allows some small incorrect alignment between the two axes and enough friction to transfer moment at that low speed and load. Acquisitions are performed in both directions by changing the polarity of connections on the power supply. The absolute value of the relative speed should be upper limited to allow complete acquisition. Lower speed implies multiple acquisition. The acquisition length is important. The longer the better. Repeating acquisition orthongonally is useful to assess anysotropy. Althogh climbing is only in one direction, it is easier to estimate the average asperity size and distribution on anysotropical surface. I am putting two stripes of tape at 100 mm. I manually start and stop grabbing the profile when the laser light passes on the tape and then manually select the clean profile between the two, so I can convert the number of samples in real mm. The speed is selectable. Lower speed ==> more samples, but at lower speed the motor is vibrating. Data Acquisition and Processing Software Direct acquisition within Matlab is possible, it is just one string of code, but with more than 10 parameters and references to pointers and so on, as indicated in the Forum link above, so after a few attempt I stopped that way. I am using the acquisition demo program coming with the card @ 300 samples/second (I wanted to use Matlab to sample @ 1000 Hz, as allowed by the card). It is automatically installed with the card. The program saves an Excel file to disk. By plotting the sampled profile I manually find the flat regions (the tape) between the 5 cm to get a clean profile of the surface. Then I copy and paste the column with values (Volts) in an M-file to import a vector in Matlab. Another M-file processes the profile to remove spykes, coming out from out of range (deep holes) or vertical edges (steep asperities) and calculates all the stats. This is still in progress so check the M-file to see the actual algorithm.
Acquisition of surface profiles
-- MicheleLanzetta? - 17 Jul 2004 Trying to achieve a simple and objective assessment/characterization of surfaces to compare the performance of different SpinedSurfaces. Also applies to foot and toe testing, etc. Instrumentation The Omron Z4M? -N30V measurement sensor available @ the lab can be used after consulting the online documentation with any analog data acquisition card connected to terminals 2-8 and powering terminals 1-7. VirgilioMattoli's card can be used inserting a 220 Ohm resistor (red red brown gold) to convert the 4 to 20 mA output to 0.88 to 4.4 V. Differential input is acquired for higher accuracy. The three cables going to the card are twisted for lower interferences (signal noise). I have not reset the Omron unit and hope the company setup is still ok (setup mode is available, but we do not have instruments to reset it in the lab). I did not have quick access to a silicon wafer so I did not try it on the very flat surface. I just measured some tape and metal thickness to verify that the accuracy of the instrument is as claimed in the manual. For actual tests I am using my own 12 bit LabJack cheap (130$) USB card, Win-plug&play. To order and download free acquisition SW, including LabView, Matlab, Excel, Linux and much more. Forum and main: Also a novice like me has been very happy about it. A (1D) surface profile is obtained and can be imported in Excel or Matlab for further processing, like parameter estimation. Basic instructions for use with Data Acquisition Toolbox Setup An overview of a previous version of the profile acquisition system can also be seen in VisionExperiments.Constant relative displacement of the surface with respect to the sensor is necessary.
This has been achieved using the smaller white belt with motor with adjustable speed provided by AMcClung. At low speed, the motor vibrates too much and surface acquisition as an error unacceptable for our purposes. The final configuration has been improved using a long screw as indicated by prof. MarkCutkosky driven by a 24 V motor with gear 600 RPM, operating @ 6 V. Here the stroke is 5 cm. Again delimited by tape. This last configuration requires a lot of care (not touch the table, verify that motor-screw connections made with tape is ok), and it allows an accuracy in the order of the Optical profilometer instrument. Transfering the motion from motor to slide is a very critical aspect, for vibrations, irregular motion.
The Omron sensor is installed on the small sliding guide from Micro Slides model A-2030 with the magnetic support. The sensor is moving, because the big weight of concrete bricks does not allow to put them on the small slide (the most accurate available in the lab). For this reason the acquisition distance changes because the direction of motion of the slide is never parallel to the concrete surface, so there is always a slope in profiles. This does not affect the parameter calculation because only micro-variations are considered.
The magnet of the sensor support is also used to transfer the motion from the motor to a nut rotating on the screw. The screw has a ratio of about 1.5 mm per revolution. The actul speed is automatically calculated by the acquisition program, because the acquisition time is know and all profiles are at constant distance.
The connection between motor and screw is done by clamping a rubber tube on the motor shaft and on the screw. It allows some small incorrect alignment between the two axes and enough friction to transfer moment at that low speed and load. Acquisitions are performed in both directions by changing the polarity of connections on the power supply. The absolute value of the relative speed should be upper limited to allow complete acquisition. Lower speed implies multiple acquisition. The acquisition length is important. The longer the better. Repeating acquisition orthongonally is useful to assess anysotropy. Althogh climbing is only in one direction, it is easier to estimate the average asperity size and distribution on anysotropical surface. I am putting two stripes of tape at 100 mm. I manually start and stop grabbing the profile when the laser light passes on the tape and then manually select the clean profile between the two, so I can convert the number of samples in real mm. The speed is selectable. Lower speed ==> more samples, but at lower speed the motor is vibrating. Data Acquisition and Processing Software Direct acquisition within Matlab is possible, it is just one string of code, but with more than 10 parameters and references to pointers and so on, as indicated in the Forum link above, so after a few attempt I stopped that way. I am using the acquisition demo program coming with the card @ 300 samples/second (I wanted to use Matlab to sample @ 1000 Hz, as allowed by the card). It is automatically installed with the card. The program saves an Excel file to disk. By plotting the sampled profile I manually find the flat regions (the tape) between the 5 cm to get a clean profile of the surface. Then I copy and paste the column with values (Volts) in an M-file to import a vector in Matlab. Another M-file processes the profile to remove spykes, coming out from out of range (deep holes) or vertical edges (steep asperities) and calculates all the stats. This is still in progress so check the M-file to see the actual algorithm.
- LabJackStream.zip: Old Backup Matlab files and surface profiles
Copyright &© by the contributing authors. All material on this collaboration platform is the property of the contributing authors. Ideas, requests, problems regarding TWiki? Send feedback