On April 02, 2013, Tibi, Jonathan and Daniel had a "Mosquito Sprint" to assemble and test a new Mosquito using an LED at 850nm and a photodiode with a TIA (transimpedance amplifier) integrated. The experiment was done on a breadboard, no new Mosquito prototype was made.
The optical design used depicted in the diagram below and described HERE.
We measured the signal through a Y coupler/splitter coming from the force transducer [sensing application]. The signal is still small, because of poor optical coupling between the LED and the MM glass fiber. We tried different LED's, but the one that worked best is a high power and has 4 wires on the semiconductor (normal LEDs have one wire on in the middle of the semiconductor and emit a beam with a dark spot in the middle, which makes it more difficult to couple into a small optical fiber).
The signal from the PD-TIA was measured with the LabJack.
The coupling can be greatly improved if proper optics are used. The optical assembly for launching into the fiber was made with improper optical parts, just for prof of concept. We (Tibi and Jonathan) are actively looking for the best coupling solution (OZ-Optics might help). For now, we are just blasting the fiber with a lot of power.
It turns out that the signal from the force transducer is very stable!!! We also measured some displacement signal using a Joint-Type transducer. Tibi believes that the gap needs to be smaller for this type of transducer if it is operated with less coherent light (the the LED), as opposed to the laser light. We tried a constriction transducer but we were unsuccessful. Is there a difference between the last and the LED light for the constriction transducer?
1) With LED it becomes harder to go for smaller diameter fiber - to decrease the stiffness of our transducers.
2) If we want to go with the Fabry Perot transducer we need coherent light, and the LED is not the solution.
3) The high power LED we used was getting a bit hot - not a big problem...
1) The noise we saw in the past with our laser-based Mosquito was most probably related to light going back into the laser. Therefore, the Chinese isolators attached to their for their laser doesn't work with the MM fiber assembly that we have.
2) No Fabry Perot interference was observed, due to the small coherence length of the LED light. .
If we still want to use an LD, instead of the LED, for different reasons, including also the Fabry Perot, we need to manage the feedback, OZ-Optics seem to have a solution (angled connectors and off-axis coupling of laser and pigtail fiber).
On April 03, 2013 Tibi tested this new Mosquito for displacement sensing, simulating the joint-type transducer. The results are documented here. The figure below shows the experimental setup used.
At first, we explored intensity variation with distance from a "perfect" mirror positioned perpendicular to it.
After, we explored intensity variation with the angle of the mirror, simulating the joint-type transducer.
The sensitivity is not that bad, but not enough for Phil's project: 13% intensity variation for a 3 degree angle of the mirror with the perpendicular. For a 5 mm long lever 3 degree angle means 260 microns tip displacement, assuming that the lever itself doesn't bend a lot. The graph shows that we can detect 80 microns tip displacement of a joint-type with 5mm long lever. This is even more apparent if we do some digital signal conditioning. The two pictures below show signal before and after rotating the mirror of 1 degree. There is some large signal variation when I change the angle. This was done at 150kHz acquisition rate and applied averaging to the raw data.
A 0.5 degree variation get's us to 43 micron resolution for a 5 mm long lever.
The sensitivity can be improved with analog filtering applied to the signal and some averaging done in hardware, if we use fast acquisition and over-sample to average. We can probably get down to 10-20 microns resolution.
But these are only the first tests. Next time I'll explore the constriction transducer.