Matheus improves the design
Feb 24, 2014
Rodrigo's R&D report on the low-cost tape transducer
Aug 28, 2013
New low-cost tape sensor 3D printed
Aug 9, 2013
First attempt
Today Daniel printed the first thing using our new B9Creator. The low-cost tape sensor had the honor to go in first. We are now tuning the printer for better resolution.
3D model of this tape sensor on 3D printed device open model
One prototype after cleaning.
Video of the process
Further notes
This first design had some problems. See Rodrigo's below (open doc)
Comparison between the Low-Cost Tape sensor and the Flex sensor
Jul 16, 2013
We are doing this comparison directly on a hockey stick, in order to make this data directly available to the Hockey Stick project.
Low cost Tape Sensor measures our piezo actuator
Jun 15, 2013
low cost tape sensor
piezo driver and stack
piezo stack
The low cost Tape Transducer was fixed along the piezo stack, which was activated using our new driver. The graph below shows the signal from the tape transducer driven by the piezo. We started with a sin wave and ended with a triangular wave. The entire measurement took 20 seconds.
We tried higher frequencies, but above 3 Hz the kapton tape used to fabricate the tape transducer, which takes time to relax from its stretched state, doesn't follow anymore. We didn't measure the total displacement of the piezo, but it was driven at max voltage, which, if everything was OK, would give 50um motion amplitude. The only thing we can say with certainty is that the amplitude of the motion recorded with the help of the tape transducer was max 50 microns. Why is this so exciting? Because we could test something we built with another thing we also built!!!! In terms of our business strategy, you can now see how these different devices we build combine together into more complex systems. This is in fact the first expression of the touch-sensitive and self-"aware" robots for micromanipulation, combining an actuator and a displacement sensor.
Low cost tape sensor demonstrated and presented
Jun 11, 2013
Beam deflection sensor built and tested
Apr 28, 2013
On 26th of April we finished the assembly of the first beam deflection optical sensor. Tibi worked on the photonics and the optomechanical assembly. Jonathan supplied the electronics. Ivan and Daniel also helped with the understanding of how it actually works. The night before the demo, Tibi, Ivan and Daniel tested the prototype.
This prototype uses the Optical Design - Tape sensor one in (glass) 3 out (PMMA).
Open main document for optomechanical assembly and electronics.
As you can see in the video above, we used an aluminum tube as the main structure of the beam. The fibers were placed into a grove made on a wooden rod, which was about the same diameter as the inner diameter of the aluminum tube. Glue was pored into the grove to keep the fibers in place. For lever we used an 2mm OD, 10cm long acrylic rod, which we coated with silver on a tip. The tip was polished with 1 micron paper before coating. For joint we used a transparent polymer tube. The gap was set at around 150 microns. The tip of the smaller glass fiber was rounded using a modified fiber splicer, in order to reduce the exit cone of the light from the fiber. A high power red LED was used to power the device. The wood rod with the fiber assembly was introduced within the aluminum tube.
The fiber assembly was first tested using a gimbal, see video below.
We assumed that once assembled into a aluminum beam, the sensor will behave very similarly to the bench test using the gimbal, in which the mirror rotates around a point on its surface, facing the laser emitting fiber. In reality, the prototype was assembled with the fibers off axis, i.e. not in the center of the beam. They were placed into a grove made on the surface of the wooden rod. When the beam is bent, the fibers are stretched or compressed, if they find themselves within the bending plane. When stretched, the mirror get's away from the fibers. When compressed, the mirror get's closer to the fibers. This induces an intensity variation that is greater than what is produced by the small tilt of the mirror. The signal is dominated by this compression/stretching mode. We were not able to extract 2D spatial information from the device, but we can easily detect bending in the plane containing the optical fibers.
Characterization and optimization tests
Apr 11, 2013
Enrique used the setup made by TiTiberius Brastaviceanubi to characterize and optimize the tape transducer and recorded the first comprehensive data.
The main goal of this experiment was to find the optimal size of the gap (distance between the mirror and the fibers).
The one in 3 out prototype with round glass fiber tip was used. The setup consisted of a green laser coupled to the 125/62.5 MM glass fiber (the in fiber) using a lens. The fiber was hold in place using the reusable ST connector. The one in 3 out prototype (fiber assembly) was prepared by Tibi before the experiment: PMMA fibers were polished and the glass fiber tip was cleaned with alcohol. The tip of the fiber on the side of laser was freshly cleaved.
We are looking for a linear relation between angle (mirror tilt) and gap size. The preliminary data shows that being too far from, or too close to the mirror is not good. There is a sweet spot somewhere in the middle. Open the spreadsheet used for characterization and optimization of the tape transducer. Next week we'll continue exploring this same prototype and evaluate its sensitivity. Tibi needs to modify the LabView program in order to allow faster acquisition and better statistics.
See the Tape Sensor optimization and characterization doc.
Live demos of the Tape sensor
Feb 6, 2013
Test of the 4 fibers all PMMA - 0.1 degrees resolution
First, we built and tested the all 1mm PMMA transducer: one fiber in and 3 fibers out. The resolution test was done at around 800 microns gap. The device was feed by a red LED. We measured signal steps induced by 0.1 angle degree variations. The acquisition rate was slow in this experiment, only 1kHz. Moreover, 100 points are digitally averaged to reduce noise. The video below shows the results. The signal plotted was intensity of one fiber minus intensity of another fiber. The test was performed on a single axis. The operation of the transducer was simulated using a flat mirror on a gimbal, which is positioned on a XY manual stage.
a stereo microscope was used for visualization and optical alignment
fiber assembly in front of the mirror
Test of the 4 fibers, one glass 3 PMMA - 0.02 degrees resolution
Two versions were made, one with the glass fiber having a flat tip, like in the first picture below, and the other one using a glass fiber with round tip, made by melting, see second picture below.
In both cases a green laser was used. The picture below shows, from left to right, the laser, a lens, and the fiber.
In both cases, a flat mirror on a gimbal was used to simulate the operation of the device. The gimbal was placed on a XY stage.
First prototype
Second prototype
The the case of the flat tip version the results obtained were only sightly better than in the case of the 4 fibers all PMMA prototype. We also noticed that the light coming out of the glass fiber was not very gaussian, almost resembles a flat top, with a lot of speckles.
We moves right a way to the second version, with the round tip, since the results were not encouraging.
The advantage of the round tip is that it makes the exit cone of the glass fiber smaller, and increases the sensitivity of the device. This was confirmed by observing the light out of the round tip of the fiber using a white piece of paper. We used a fiber splicer modified by Jonathan to process the tip of the fiber.
on the computer screen, the 125 micron diameter glass fiber
between the two electrodes, waiting to get zapped
fiber getting zapped on the same screen, you can see the plasma
between the two electrodes. this plasma melts the glass
the kind of shapes we can get, a much smaller sphere was created
in the case of the fiber used in the tested device, see picture above
The video below shows the results obtained. Angle steps of 0.02 degrees were made while the signal was recorded. The signal plotted was intensity of one fiber minus intensity of another fiber. The test was performed on a single axis. We can clearly see that we have a resolution of at least 0.02 degrees. This can be greatly improved by doing some signal conditioning on the electronic board.
Making and testing the 1 in 3 out all PMMA design
Jan 25, 2013
Jonathan improved the electronic circuit for the tape sensor, adding variable gains (third version, see more on this evolving circuit prototype). This is a huge improvement!
Tibi assembled the 1 in, 3 out all PMMA design, connected it to the electronics and tested it with a mirror to see if there is sign of signal crossing, which is an indication that we can map small deflections in 2D. The screenshot below shows signal crossings for small angle variations of the mirror, placed 1mm away from the tip of the fibers.
A video was also made to demonstrate this prototype. Next step is to assemble everything into a single wire, including the mirror on a lever, and test it on a 1m long beam. In order to optimize the sensitivity we'll keep the gap variable.
The making of this prototype
7 fibers were put together like in this picture
Since we only use 3 outer fibers to collect light, the other 3 were pushed back a little, in order to glue the remaining 4.
The remaining 4 fibers were polished
And tested shining light through them.
In order to ease the connections with the LED and PS a plastic tube was put over the tip of all fibers. The end results looks like this. The red one is the hot fiber, i.e. the one that brings light into the sensing area. The other 3 black ones are collecting fibers.
Everything was connected to the electronics and tested.
Building prototypes and testing the electronics and the software
Jan 22, 2013
Tibi tested a different fiber arrangement for the tape sensor for its mechanical integrity. Six 1mm diameter PMMA fibers are placed around a center fiber of the same type. The structure is stable and the 6 outer fibers are distributed almost uniformly around the center one. The center fiber brings light into the sensing region. A mirror reflects the light back. 3 out of these 6 outer fibers collect the light and carry it back to 3 detectors. This is one design among others, described in this document.
At the same time, Tibi finished and tested the LabView program that can accept and process multiple inputs. More about the electronics made by Jonathan here. For the proof-of-concept, all calculations on input signals will be performed digitally. The application will be slower, but this technique is a lot faster. Once we'll learn more about the relation between all input signals, to map input signal variation into 2D spatial information, and if this application takes off, we'll think about a way to do that on hardware, or even use a DSP. Bill and Jonathan are working on the mathematical model for this type of sensor.
We successfully tested a 3 fiber device made by Jonathan - 1 fiber input and 2 output, which is a uni-dimensional but directional sensor. You can see the demonstration recorded during our meeting today.
Francois is also making progress with optical fiber coating and we'll use these new techniques to make a custom-designed mirror for this device.